article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
radio sources with a total projected size of @xmath15 mpc are categorized into giant radio galaxies .
they constitute one of the largest classes of astrophysical objects in the universe . with a typical spectral age
@xmath16 being higher than a few @xmath17 myr @xcite , they are recognized as relatively old systems .
therefore , the giant radio galaxies are utilized to investigate the activity of jets and associated lobes in evolved radio sources , which has been yet to be explored .
lobes of radio galaxies accumulate an enormous amount of relativistic plasma , comprised of electrons and magnetic field ( and possibly heavy particles including protons , although the present paper does not deal with them ) .
the plasma energy in the lobes is sourced from the bulk kinetic energy of the jets through their terminal hot spots .
this makes the lobes one of the most valuable probes for the past jet activity .
it is able to estimate roughly the plasma energy in the lobes by their synchrotron radio radiation .
however , the radio information alone is not useful to disentangle the electron and magnetic - field energies , and thus , the minimum energy condition @xcite was widely assumed , without any concrete physical justification .
another valuable information to study the energetics should be provided by inverse compton ( ic ) x - ray emission from the synchrotron - emitting electrons in the lobes , where the cosmic microwave background ( cmb ) photons are boosted up @xcite .
the ic x - ray spectrum , in comparison with the synchrotron radio one , enables an independent estimate of the electron and magnetic energies . in the last two decades since its discovery from the lobes of fornax a @xcite ,
the ic x - ray radiation has been a standard tool to diagnose the lobe energetics ( e.g. , @xcite ) .
the technique was strengthen by the ic @xmath18-ray detection with the fermi observatory from the outer lobes of centaurus a @xcite .
as a result , the electron energy is found to surpass the magnetic one in the lobes , typically by a factor of @xmath19 @xmath20 . until recently ,
ic x - ray examination on lobes of giant radio galaxies has been hampered by mainly their faintness .
for diffuse x - ray sources with a low surface brightness , the x - ray imaging spectrometer ( xis ; @xcite ) onboard the suzaku observatory @xcite exhibits a great advantage , owing to its moderate field of view ( a @xmath21 square ) and to its rather low and stable instrumental background @xcite .
actually , a systematic study with the xis has been just conducted for the lobes of several giant radio galaxies @xcite .
this research arrived at a significant detection of ic x - ray photons from these lobes , and pioneeringly hinted that the current jet activities are significantly declined in the giant radio galaxies , compared with those with a moderate size ( e.g. , @xmath22 kpc ) .
this motivates us to enlarge urgently the sample of ic - detected giant radio galaxies . the giant radio galaxy , 3c 236 , is located at the redshift of @xmath23 @xcite .
assuming the cosmology with @xmath24 km s@xmath25 mpc@xmath25 , @xmath26 , and @xmath27 , the redshift gives an angle - to - size conversion ratio of @xmath28 .
it is optically classified as a low excitation radio galaxy , with weak emission lines @xcite . with a classical fanaroff - riley ii radio morphology @xcite
, the radio source has an angular extent of @xmath21 @xcite , which corresponds to a projected linear size of @xmath29 mpc .
therefore , this had been known as the largest radio source all over the sky , until the giant radio galaxy j1420 - 0545 with a size of @xmath30 mpc was discovered @xcite .
the age of the source is estimated as @xmath31 myr by the synchrotron aging technique @xcite , indicating that it is really an elderly radio galaxy .
these make 3c 236 as an ideal target for the ic x - ray study with suzaku .
actually , a faint x - ray emission associated with its west lobe has been concretely detected in the suzaku observation , as is reported in the present paper .
the log of the suzaku observations for the giant radio galaxy , 3c 236 , is given in table [ tab : log ] .
we focused on the west lobe , since radio images of 3c 236 ( e.g. , @xcite ) indicate that diffuse radio emission is more pronounced in the west lobe while a significant dominance of double bright hot spots is recognized in the east one .
the central region of the west lobe was placed at the nominal position of the x - ray telescope ( xrt ; @xcite ) for the xis .
we fixed the satellite roll angle at @xmath32 deg , in order to avoid contamination onto the west lobe from calibration sources at the specific corners of the xis field of view , and distortion by the anomalous columns of the xis 0 .
unfortunately , in this configuration , the 3c 236 nucleus is located slightly outside the xis field of view .
we observed another sky field on the southwest of the west lobe field , which is known to be relatively free from contaminating bright x - ray point sources ( e.g. , @xcite ) , and evaluated the x - ray background ( xrb ) spectrum .
this xrb field has not yet been observed with asca , chandra , and xmm - newton . the present paper deals with the xis results only , because no significant x - ray signals were detected with the hard x - ray detector @xcite from both sky field .
the standard software package , heasoft 6.16 , was adopted for data reduction and analysis .
we referred the calibration database ( caldb ) as of 2014 july 1 and 2011 june 30 for the xis and xrt , respectively .
all the xis events were reprocessed with the suzaku tool , aepipeline .
we filtered the data by the following standard criteria ; the satellite is outside the south atlantic anomaly ( saa ) , the time after an saa passage is larger than @xmath33 s , the geomagnetic cut - off rigidity is higher than 6 gv , the source elevation is higher than and above the rim of day and night earth , respectively , and the xis data are unaffected by telemetry saturation .
the filtering procedures yielded 78.0 ks and 42.4 ks of good exposures for the west - lobe and xrb fields , respectively . in the scientific analysis
below , we picked up those events with an xis grade of 0 , 2 , 3 , 4 , or 6 . in order to mitigate the artificial increase in the non - x - ray background ( nxb ) level related to the charge injection , those pixels flagged as sci_2nd_trailing_row were masked out for the backside - illuminated ( bi ) ccd chip ( xis 1 ) .
( 80mm,80mm)figure1a.ps ( 80mm,80mm)figure1b.ps thanks to its sub - arcsecond angular resolution , chandra is very useful to evaluate the x - ray flux of possible contaminating faint point - like sources located within the west lobe , which are unresolved with the suzaku xis .
3c 236 has ever been observed by the chandra acis - i array with three exposures . among them , we selected the observation aiming at its west lobe ( obsid = 10246 ; see table [ tab : log ] ) .
the acis data were reduced with the software package ciao 4.6 .
we reprocessed the data and created a new level 2 event file with the tool chandra_repro , by referring to caldb 4.6.2 . because the nxb count rate integrated over the whole acis - i array
was found to be fairly stable throughout the observation , we performed no additional data screening to the new level 2 event file .
correspondingly , we obtained 29.4 ks of good exposure on the west lobe . when the science products were derived , the grade selection criterion similar to that for the suzaku xis ( i.e. , 0 , 2 , 3 , 4 , or 6 ) was applied .
llr ( r.a .
, decl . ) & @xmath34 & @xmath35 + ( , ) & @xmath36 & @xmath37 + ( , ) & @xmath38 & @xmath39 + ( , ) & @xmath40 & @xmath41 + ( , ) & @xmath42 & @xmath43 + ( , ) & @xmath44 & @xmath45 + ( , ) & @xmath46 & @xmath47 + ( , ) & @xmath46 & @xmath48 + ( , ) & @xmath46 & @xmath49 + ( , ) & @xmath50 & @xmath51 + ( , ) & @xmath50 & @xmath52 + ( , ) & @xmath46 & @xmath53 + ( , ) & @xmath46 & @xmath54 + ( , ) & @xmath40 & @xmath55 + ( , ) & @xmath46 & @xmath56 + ( , ) & @xmath57 & @xmath58 +
the left panel of figure [ fig : img_wl ] displays the xis image of the 3c 236 west lobe . here , we extracted the xis events in the 0.5 5.5 kev range , where the observed data are free from photons from the calibration sources at the field - of - view corners . the data from all the ccd chips ( xis 0 , 1 and 3 ) were summed up .
neither the xrb nor nxb was subtracted .
the image was smoothed with a two - dimensional gaussian kernel of radius .
the radio image of 3c 236 at 608.5 mhz @xcite was superposed with contours .
the figure [ fig : img_wl ] left panel suggests faint x - ray emission associated with the west lobe , although it is certainly contaminated by several unresolved x - ray sources behind the lobe . the chandra acis - i image of the same sky field in @xmath59 @xmath60 kev is presented in the right panel of figure [ fig : img_wl ] .
the image clearly reveals numbers of point - like x - ray sources .
we performed a source detection procedure by the ciao tool wave_detect with a standard parameter set .
the exposure and point - spread - function maps over the acis field of view was correctly taken into account for the source detection . in total , @xmath61 x - ray sources were detected with a significance of @xmath62 within the acis field of view .
these sources are plotted with the circles or boxes in figure [ fig : img_wl ] . utilizing the 608.5 mhz radio map ,
we evaluated the size of the west lobe .
the standard deviation , @xmath63 , of the radio intensity in the source free pixels was regarded as the noise level of the radio map .
the emission region was defined as those pixels with a radio surface brightness of @xmath64 .
we successfully approximated the envelope of the emission region of the west lobe by an ellipse , which is shown in figure [ fig : img_wl ] with the solid line .
after decomvolving the radio beam size , the ellipse has a major and minor radius of and , respectively , corresponding to the projected size of @xmath65 kpc and @xmath66 kpc at the redshift of 3c 236 .
the x - ray spectrum from the west lobe is accumulated in this ellipse ( see [ sec : westlobe ] ) . on the region ,
the @xmath67 chandra sources , plotted with the circles in figure [ fig : img_wl ] , were detected .
these sources are listed in table [ tab : src ] .
( 80mm,80mm)figure2.ps ll parameter & value + @xmath68 ( cm@xmath25 ) & @xmath70 + @xmath71 & @xmath72 + @xmath73 & @xmath74 + @xmath75 & @xmath76 + @xmath77 ( erg s@xmath25 @xmath78 str@xmath25 ) & @xmath79 + @xmath80 ( erg s@xmath25 @xmath78 str@xmath25 ) & @xmath81 + @xmath82 ( erg s@xmath25 @xmath78 str@xmath25 ) & @xmath83 + @xmath84 & @xmath85 +
it is inevitably required to precisely subtract the nxb and xrb levels , for the spectral analysis of extended x - ray emission with a low surface brightness .
it is reported that the nxb level of the xis is predictable with an accuracy of @xmath86% for a typical exposure of @xmath87 ks by utilizing the xisnxbgen tool @xcite .
we make use of the xis data accumulated from the xrb field ( see table [ tab : log ] ) to model the xrb spectrum in the direction to 3c 236 .
the @xmath59 @xmath88 kev xis image of the xrb field is plotted in figure [ fig : img_xrb ] .
we extracted the xrb spectrum within the solid rectangle in the image .
the region is selected so as not to be affected by the anomalous columns of the xis 0 .
in addition to the previously known x - ray sources ( the crosses in figure [ fig : img_xrb ] ) , listed in the second rosat source catalog of pointed observations with the position sensitive proportional counter , a few point sources and faint extended x - ray emission were recognized within the regions .
then , the dashed rectangle and circles in the image were discarded .
figure [ fig : spec_xrb ] shows the xrb spectrum derived from the region , after the nxb spectrum generated by xisnxbgen was subtracted .
we co - added the data from the two front - illuminated ( fi ) ccd chips ( xis 0 and 3 ) .
the xis response matrix functions ( rmfs ) of this region were created by xisrmfgen . assuming a diffuse source with a radius of , which has a flat spatial distribution
, we simulated the auxiliary response files ( arfs ) by xissimarfgen @xcite . the @xmath89 @xmath17 kev xrb spectrum is widely known to be decomposed into a power - law ( pl ) component and a two - temperature thermal plasma emission @xcite .
the pl component is reported to exhibit an average surface brightness of @xmath90 ergs @xmath78 s@xmath25 str@xmath25 in @xmath91
@xmath17 kev , with a rather small spatial fluctuation of @xmath92% .
its photon index is accurately determined as @xmath93 @xcite .
this component is believed to be dominated by unresolved faint sources , including distant active galactic nuclei . by adopting the mekal code @xcite ,
the temperatures of the thermal plasma emission were determined as @xmath94 kev and @xmath95 kev @xcite .
the thermal components are thought to be of galactic and local origin , and its surface brightness is reported to be highly variable from field to field .
we fitted the xis spectrum from the xrb region with a composite model consisting of the pl component ( @xmath93 ) and the two mekal ones ( @xmath94 kev and @xmath95 kev ) .
all the components were subjected to the galactic absorption toward 3c 236 with a hydrogen column density of @xmath96 @xmath78 @xcite .
the model was found to be reasonable ( @xmath97 ) , yielding the best - fit parameters tabulated in table [ tab : spec_xrb ] .
the measured @xmath91
@xmath17 kev surface brightness of the pl component , @xmath98 ergs @xmath78 s@xmath25 str@xmath25 , falls near the lowest end of the xrb surface brightness distribution @xcite .
therefore , we have judged that the best - fit xrb model is safely applied to the spectral analysis of the west lobe .
( 80mm,80mm)figure3.eps the chandra acis signals from each x - ray source , detected on the integration region for the xis spectrum from the west lobe ( the ellipse in figure [ fig : img_wl ] ) , were integrated within a circle of a @xmath17 pixel ( ) radius centered on the source position ( see table [ tab : src ] ) .
the nxb plus xrb events were accumulated from a concentric annuls with a radius of @xmath67
@xmath99 pixel ( ) .
the spectral files , together with the rmf and arf ones , were created by the ciao script specextract .
the energy - dependent aperture correction was applied to the arfs .
( 80mm,80mm)figure4.eps ll parameter & value + @xmath68 ( cm@xmath25 ) & @xmath100 + @xmath71 & @xmath101 + @xmath102 ( erg s@xmath25 @xmath78 ) & @xmath103 + @xmath84 & @xmath104 + due to low signal statistics for the individual sources , we presents the acis spectrum summed over the 15 sources in figure [ fig : spec_src ] .
the rmf was simply averaged over these sources , after it was multiplied by the corresponding arf .
the spectrum was successfully approximated by a pl model modified by a free absorption ( @xmath105 ) .
the best - fit parameters are summarized in table [ tab : spec_src ] .
the summed absorption - inclusive x - ray flux from the sources was measured as @xmath106 erg s@xmath25 @xmath78 in @xmath59 @xmath107 kev .
the photon index , @xmath108 , was found to agree with a typical value of nearby active galaxies in the similar energy range @xcite .
lllllll & & + & & + + + + + + data & @xmath109 & @xmath110 & & @xmath111 & @xmath112 & + nxb & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath117 & @xmath118 + signal & @xmath119 & @xmath110 & @xmath115 & @xmath120 & @xmath112 & @xmath118 + xrb & @xmath121 & & @xmath118 & @xmath122 & & @xmath110 + sources & @xmath123 & & @xmath118 & @xmath124 & & @xmath110 + excess & @xmath125 & @xmath110 & @xmath126 & @xmath113 & @xmath112 & @xmath112 + figure [ fig : spec_wl ] shows the xis spectra of the west lobe integrated within the ellipse in figure [ fig : img_wl ] , after the nxb events simulated by the xisnxbgen tool were subtracted .
the data below 0.6 kev were discarded , since we noticed considerable inconsistency between the xis fi and bi data in the spectral fitting over this energy range , probably due to the calibration uncertainty .
the signal statistics from the region in the @xmath127
@xmath128 kev range are summarized in table [ table : stat : wl ] .
we adopted the typical nxb systematic uncertainty of @xmath129% @xcite .
significant x - ray signals exceeding the nxb were detected , with an fi and bi count rate per one ccd chip of @xmath130 cts s@xmath25 and @xmath131 cts s@xmath25 , respectively . here , the first and second error represents the statistical and systematic one , respectively .
the nxb - subtracted xis spectra of the west lobe plotted in figure [ fig : spec_wl ] inevitably include the xrb signals and contamination from the point sources , in addition to the signals from the west lobe itself . in the similar manner to
that adopted in [ sec : xrb ] , the rmf and arf of the west lobe region to the xrb were generated .
the dash - dotted lines in the left panel of figure [ fig : spec_wl ] indicate the xrb model spectrum ( table [ tab : spec_xrb ] ) , which is convolved with the xis rmf and arf .
the fi and bi count rates of the xrb within the west lobe region in @xmath127
@xmath128 kev were estimated as @xmath132 cts s@xmath25 and @xmath133 cts s@xmath25 , respectively .
the rmfs and arfs of the west lobe region to the @xmath67 chandra x - ray sources listed in table [ tab : src ] were individually calculated , and they were simply averaged over all the sources . the best - fit pl model to the sum of the chandra spectra from these sources ( see figure [ fig : spec_src ] and table [ tab : spec_src ] ) is drawn with the dotted lines in the figure [ fig : spec_wl ] left panel , after it is convolved with the averaged rmf and arf . the contamination from these sources onto the @xmath127
@xmath128 kev fi and bi count rate of the west lobe were estimated as @xmath134 cts s@xmath25 and @xmath135 cts s@xmath25 , respectively .
we neglected the chandra sources detected outside the signal integration region , since their contribution is inferred as at most @xmath136% of the sum signals from the @xmath67 sources .
the solid lines in the left panel of figure [ fig : spec_wl ] present the sum spectrum of the xrb and source contamination .
the residuals plotted in the bottom clearly visualizes excess signals over these two components ( @xmath137 ) .
the @xmath127 @xmath128 kev excess count rate is estimated as @xmath138 cts s@xmath25 and @xmath139 cts s@xmath25 with the xis fi and bi , respectively . here and hereafter ,
the errors of the best - fit spectral models to the xrb and to the contaminating sources were propagated to the systematic error ( i.e. , the second one ) .
thus , the statistical significance of the excess with the fi and bi was evaluated as @xmath140 and @xmath141 , respectively . even if the systematic errors are taken into account
, the excess has remained meaningful .
( 75mm,75mm)figure5a.eps ( 75mm,75mm)figure5b.eps in order to reproduce the excess spectrum from the west lobe region , we employed an additional absorbed pl component .
the arf to this pl component is calculated for a diffuse source filling the ellipse in figure [ fig : img_wl ] with a uniform surface brightness .
first , we fitted the xis spectrum with the photon index @xmath71 and absorption column density @xmath68 both left free .
the confidence map in the figure [ fig : confidence ] visualizes that the galactic absorption toward 3c 236 with the column density of @xmath96 @xmath78 @xcite is fairly preferable , although the acceptable range is found to be slightly wide .
@xcite reported a massive nuclear outflow of neutral gas with a column density of @xmath142 @xmath78 , which could affect the nuclear jet emission .
such a high column density , which requires a very soft spectrum ( @xmath143 ) , is outside the @xmath144% statistical confidence range .
we think that the nuclear outflow should not impinge upon the x - ray emission from the mpc - scale lobe , since it is reported to exhibit a spatial extent of only @xmath145 kpc .
based on the argument above , we decided to adopt the galactic absorption for the west lobe .
the excess spectrum is successfully described with the pl model modified by galactic absorption as is indicated by the dashed lines in the right panel of figure [ fig : spec_wl ] ( @xmath146 ) .
the best - fit spectral parameters for the excess are tabulated in table [ tab : spec_wl ] ( case 1 ) .
the photon index and flux density at 1 kev is measured as @xmath0 and @xmath147 njy , respectively . in figure
[ fig : sed ] , the spectral energy distribution of the x - ray excess from the west lobe is compared with that of the radio synchrotron emission .
the radio data of the west lobe are taken from @xcite . since the x - ray spectral integration region adopted here ( the ellipse in figure [ fig : img_wl ] ) corresponds to the combination of the _ western lobe _ and _ western hot spot _
regions defined in @xcite , we evaluated the radio flux of the west lobe by adding those from these two regions .
the low - frequency synchrotron radio photon index of the west lobe was estimated as @xmath1 in the @xmath2
@xmath3 mhz range , where the radio image reveals a dominance of diffuse emission .
the best - fit index of the x - ray excess from the west lobe agrees with this value , within the errors .
therefore , we re - examined the pl model with the photon index fixed at the radio value ( @xmath148 ) . as a result , a reasonable fit ( @xmath149 ; case 2 in table [ tab : spec_wl ] ) was obtained .
the flux density of the pl component was derived as @xmath5 njy at 1 kev .
lll parameter & case 1 & case 2 + @xmath68 ( @xmath78 ) & + @xmath150 & @xmath151 & @xmath152 + @xmath153 ( njy ) & @xmath154 & @xmath155 + @xmath84 & @xmath156 & @xmath157 + ( 75mm,75mm)figure6.ps
after carefully subtracting the nxb , xrb and source contamination , we have detected the excess x - ray emission , associated with the west lobe of the giant radio galaxy 3c 236 , with a high significance .
its x - ray spectrum is approximated by the pl model , and the photon index was determined as @xmath0 .
this is found to be consistent with the @xmath2
@xmath3 mhz synchrotron radio index @xmath1 , when the statistical and systematic errors are considered .
thus , a reasonable fit was derived , by fixing the photon index at @xmath4 .
as shown in figure [ fig : sed ] , the x - ray spectrum of the west lobe does not smoothly connect to the synchrotron radio one .
therefore , the x - ray emission is not attributable to the highest frequency end of the pl - like synchrotron spectrum from the same electron population as for the radio emission .
referring to the previous x - ray studies on lobes of giant radio galaxies ( e.g. , @xcite ) , these properties are thought to support an ic origin for this x - ray emission .
the synchrotron x - ray hypothesis is further rejected by an argument on the electron radiative time scale .
we evaluate the synchrotron life time of electrons which could produce the synchrotron x - ray photons , by simply assuming the minimum energy condition @xcite . because the magnetic field strength under this assumption is found to be not so far from the more realistic value to be discussed in
[ sec : energetics ] , the following discussion basically holds . corresponding to the emission region
defined in [ sec : image ] , the three - dimensional shape of the west lobe is assumed to be an ellipsoid with a major and minor radius of @xmath65 kpc and @xmath66 kpc , respectively . from the synchrotron radio flux density , @xmath158 jy at 608.5 mhz ( calculated from @xcite ) , and the spectral slope of @xmath159 , the minimum energy magnetic field in the west lobe
was estimated as @xmath160 @xmath13 g , neglecting the proton contribution .
a similar magnetic field strength was derived by @xcite under the equipartition condition as @xmath161 @xmath13 g . in this magnetic field ,
the synchrotron photons at 1 kev are expected to be radiated from electrons with a lorentz factor of @xmath162 .
such electron is estimated to loose half of their energy in a time scale of @xmath163 myr through the synchrotron radiation , where @xmath164 is the electron mass , @xmath165 is the speed of light , @xmath166 is the thomson cross section , and @xmath167 is the minimum - energy magnetic energy density .
it is important to note that this @xmath168 value overestimates the actual radiative cooling time scale in the lobe of 3c 236 , since the compton loss is neglected .
nevertheless , the time scale is by three orders of magnitude shorter than the spectral age of 3c 236 , @xmath31 myr @xcite .
we , thus , regard the synchrotron x - ray explanation as encountering a serious problem due to the significant radiative loss .
based on the above consideration , we safely ascribe the detected x - ray emission to the ic radiation from the synchrotron - emitting electrons within the west lobe of 3c 236 .
this result has made 3c 236 the largest radio galaxy of which the lobe is ever studied through the ic photons . in the case of radio lobes , three candidate seed photon sources for the ic scattering
were widely proposed ; the synchrotron radio photons in the lobes themselves , the infra - red radiation field from the nucleus @xcite , and the cmb radiation @xcite .
the energy density of the synchrotron radiation spatially averaged over the west lobe is estimated to be very low as @xmath169 erg @xmath9 from the radio flux density @xmath158 jy at 608.5 mhz .
the infra - red flux density from the nucleus of 3c 236 , @xmath170 mjy at the k - band @xcite , is converted to the nuclear monochromatic k - band luminosity as @xmath171 ergs s@xmath25 .
this nuclear photon field yields an infrared energy density of @xmath172 ergs s@xmath25 at the nearest edge of the west lobe , which is located at the projected distance of @xmath173 kpc from the nucleus . at the redshift of 3c 236 ( @xmath174 ; @xcite ) , the cmb energy density is precisely predicted as @xmath175 ergs s@xmath25 .
therefore , we have concluded that the seed photons are dominantly provided by the cmb radiation and the other candidate sources have only a negligible contribution . by comparing the ic x - ray and synchrotron radio spectral energy distributions presented in figure [ fig : sed ] , we can measure the energy densities of the electrons and magnetic field , @xmath14 and @xmath176 respectively , in the west lobe of 3c 236 .
the input observables , required to calculate @xmath14 and @xmath176 , are listed in the upper columns of table [ tab : spec_ueum ] .
the synchrotron radio flux density and photon index ( @xmath177 jy at 608.5 mhz and @xmath1 in the @xmath2 @xmath3 mhz range , respectively ) were derived from @xcite .
we adopted the ic x - ray flux density at 1 kev of @xmath5 njy derived in case 2 where the x - ray photon index was fixed at the radio value .
corresponding to the radio and x - ray spectral slope , the electron number density spectrum was assumed to be a simple pl form described as @xmath178 .
the volume of the west lobe was estimated as @xmath179 @xmath180 , since we approximated the shape of the west lobe to be the ellipsoid with the major and minor radius of @xmath65 kpc and @xmath66 kpc , respectively ( see [ sec : image ] and [ sec : origin ] ) .
we analytically evaluated the energetics in the west lobe of 3c 236 by referring to @xcite .
the result is summarized in the lower columns of table [ tab : spec_ueum ] . here and hereafter , the statistical error of @xmath153 is propagated to the first one , while all the other errors ( including those in @xmath181 , @xmath4 and so forth ) are taken into account in the second one . the specially averaged magnetic field strength is derived as @xmath12 @xmath13 g from the equation ( 11 ) in @xcite described as @xmath182 , where @xmath183 is the synchrotron radio ( and hence the ic x - ray ) energy index , @xmath184 and @xmath185 are the frequencies at @xmath181 and @xmath153 are measured respectively , @xmath186 is approximately constant at @xmath187 over the range of @xmath188 @xmath189 , and @xmath190 is evaluated as @xmath191 at @xmath192 ( @xmath159 ) .
this magnetic field strength is equivalent to the magnetic energy density of @xmath10 ergs @xmath9 .
for the electron spectral slope of @xmath193 ( i.e. , @xmath194 ) , the @xmath14 estimate is known to be rather sensitive to the assumption on the minimum lorentz factor of the electrons . however , it has not yet been observationally explored . following the previous studies ( e.g. , @xcite ) , we first evaluated the energy density of electrons with a lorentz factor of @xmath195 @xmath196 , because they are directly observable through the ic and synchrotron radiations .
the energy density of these electrons was derived as @xmath8 ergs @xmath9 .
this indicates a slight electron dominant condition in the west lobe , as parameterized by the electron to magnetic field energy density radio of @xmath11 .
second , we lowered the minimum electron lorentz factor down to @xmath197 , since the existence of such low - energy electrons was suggested in a number of strong and compact radio galaxies @xcite .
in addition , a similar value of the minimum lorentz factor was adopted to interpret low - frequency radio data of some giant radio galaxies @xcite . as a result ,
the electron energy density increased to @xmath198 erg @xmath9 .
correspondingly , the electron dominance in the lobe was found to become more prominent , as @xmath199 .
figure [ fig : ue - um ] compiles the relation between @xmath14 ( for @xmath195 @xmath196 ) and @xmath176 in the lobes of radio galaxies , from which the ic x - ray emission was detected ( @xcite , and reference therein ) .
the figure clearly demonstrates that a typical electron dominance of @xmath200
@xmath20 is realized in the radio lobes , regardless of their physical size @xcite .
our suzaku result on the west lobe of the giant radio galaxy 3c 236 is found to be fully compatible with this picture .
due to their low energy densities , the ic - detected giant radio galaxies are distributed in the bottom - left end of the @xmath14@xmath176 plot ( i.e. , figure [ fig : ue - um ] ) . in the discussion , we neglect the lobes of the giant radio galaxy , 3c 457 , which is located around the center in figure [ fig : ue - um ] , since it is reported to be one of the youngest sources ( @xmath201 myr ; @xcite ) among the giant radio galaxies . in the lobes of giant radio galaxies , a dominance of the ic radiative loss over the synchrotron one
is theoretically pointed out , because their magnetic energy density is predicted to be less than the cmb energy density @xcite .
the recent suzaku results on the giant radio lobes @xcite , including our present result on 3c 236 , have observationally confirmed this idea , by demonstrating a condition of @xmath202 .
especially , the ic loss is extremely dominant in 3c 236 as @xmath203 . in figure
[ fig : ue - length ] , the electron energy density in the lobes are plotted against the total projected length of the radio galaxies , @xmath204 .
@xcite proposed that a lobe evolves along a track of @xmath205 on the plot , while it is actively energized by its jet .
the lobe is also predicted to deviate from the @xmath14@xmath204 track toward a lower @xmath14 value , due to significant radiative and adiabatic losses , after the energy import by its jet has ceased .
the relation of @xmath206 observed for the radio galaxies with a moderate size of @xmath207 kpc ( the dashed line in figure [ fig : ue - length ] ; @xcite ) is thought to qualitatively back up the former idea .
the latter scenario was observationally demonstrated by the radio galaxy fornax a which is reported to host a very dormant nucleus and jet . as is shown in figure [ fig : ue - length ] , the electron energy density in the lobes of fornax a is by more than a factor of 10 smaller than the value expected from the @xmath14@xmath204 regression line for the radio galaxies with @xmath208 kpc .
the recent systematic studies with suzaku ( e.g. , @xcite ) revealed that the lobes of the giant radio galaxies tend to exhibit a lower @xmath14 value by nearly an order of magnitude , in comparison to the simple extension of the regression line for the @xmath209 kpc radio galaxies .
this is clearly visualized in figure [ fig : ue - length ] . by analogy with fornax
a , this property indicates that the activity of the jets in the giant radio galaxies have been already declined , and their energy input to the lobes is currently defeated by the radiative and adiabatic losses of the electrons @xcite .
in contrast , @xmath14 in the west lobe of 3c 236 coincides with the value estimated from the @xmath14@xmath204 relation for the smaller sources .
this result suggests that the lobe is still supplied with a sufficient energy from the jet .
even though the current lobe is energetically decoupled from the jet , it has not yet entered a cooling - dominant regime .
the ongoing energy transport ( or transport until recently ) to the west lobe of 3c 236 appears to be reinforced by the observational fact that the radio spectrum around the west hot spot ( e.g. , @xmath210 in @xmath2 mhz
ghz for the _ western hot spot _ region in @xcite ) is harder than those in the surrounding regions . @xcite
closely investigated similarities between the small - scale ( @xmath212 kpc ) and large - scale ( @xmath21 mpc ) radio structures .
they pointed out that the interaction of the jet with the interstellar medium in the central region of 3c 236 , creating the small - scale structure , continued for more than an order of magnitude longer than @xmath213 myr .
the interaction is inferred to have consequently formed the the mpc - scale lobes .
the energy supply to the lobe is possible to be maintained by such a long - lasting inteaction .
in contrast , the lack or weekness of the radio jet connecting the radio structures in small and large scales especially toward the west direction @xcite , which is reminiscent of giant double - double radio galaxies @xcite , implies that the jet activity in 3c 236 is highly variable or episodic , rather than stable over its life time ( e.g. , @xcite ) . by comparing the dynamical age of the small - scale structure ( @xmath214 myr ) and the radiative age of young electrons contained in the hot spots within the large - scale lobes @xcite
, @xcite proposed that the jet had been possibly switched off for @xmath213 myr .
a high - resolution vlbi image @xcite supports a recent ignition of a one - sided jet to the west in a 10 mas scale ( corresponding to @xmath215 pc at the 3c 236 rest frame ) .
the jet reactivation was suggested to be triggered @xmath216 myr ago by a minor merger event that took place @xmath213 myr ago @xcite .
we briefly examine the impact of adiavatic and radiative losses during the suggested jet hibernation .
when the distance from the nucleus to the lobe edge ( @xmath217 mpc ) is devided by the source age ( @xmath218 myr ; @xcite ) , the time - averaged advance speed of the 3c 236 west lobe is roughly evaluated as @xmath219 .
this agrees with the typical lobe speed measured for fanaroff - riley ii radio galaxies ( @xmath220 @xmath221 ; @xcite ) . at the advance speed
, the lobe head proceeds to @xmath222 kpc in @xmath17 myr .
this results in an expansion of the west lobe volume by less than @xmath223 % .
therefore , the adiabatic loss during the jet dormancy is considered to be relatively insignificant in the 3c 236 west lobe .
the radiative time scale of the electrons in the west lobe is evaluated as @xmath224 myr @xmath225 where both the synchrotron and ic radiation is taken into account . here
, the electron lorents factor of @xmath226 corresponds to the synchrotron photons of @xmath227 ghz under the magnetic field strength of @xmath228 @xmath13 g which is measured in this study .
this means that the radiative cooling due to the jet hibernation is neglidgible to the electrons , for which we integrated the energy density @xmath14 ( i.e. , @xmath229 @xmath230 or @xmath231 @xmath230 ) .
as mentioned above , the flat synchrotron spectrum in the hot spot is compatible with the scenario that the west lobe of 3c 236 has not yet been in the cooling domain .
therefore , we have concluded that our result does not contradict to the reported suspention of the jet ejection .
another energy injection channel is hinted in the 1.4 ghz radio image by @xcite , which reveals a wiggling inner ridge in the central region of the west lobe .
they interpreted that the rigde was created by a recent jet penetrating the west lobe .
this possibly yields a very attractive idea that the new jet re - energizes the west lobe and prevents it from being cooled in the future .
we thank the anonymous reviewer for her / his constructive and supportive suggestions .
we are grateful to all the members of the suzaku team , for the successful operation and calibration .
this research has made use of the archival chandra data and its related software provided by the chandra x - ray center ( cxc ) .
abdo , a. a. et al .
, 2010 , science , 328 , 725 adgie , r. l. , crowther , j. h. , & gent , h. 1972 , , 159 , 233 alexander , p. , & leahy , j. p. , 1987
, , 225 , 1 barthel , p. d. , miley , g. k. , jagers , w. j. , schilizzi , r. t. , & strom , r. g. , 1985 , , 148 , 243 becker , r. h. , white , r. l. , & edwards , a. l. , 1991 , , 75 , 1 becker , r. h. ; white , r. l. , & helfand , d. j. , 1995 , , 450 , 559 brunetti , g. , setti , g. , & comastri , a. , 1997 , , 325 , 898 cohen , a. s. , et al .
2007 , , 134 , 1245 colla , g. , et al . , 1973 , , 11 , 291 condon , j. j. , et al . , 1998 , , 115 , 1693 croston , j. h. , hardcastle , m. j. , harris , d. e. , belsole , e. , birkinshaw , m. , & worrall , d. m. , 2005 , , 626 , 733 douglas , j. n. , bash , f. n. , bozyan , f. a. , torrence , g. w. , & wolfe , c. , 1996 , , 111 , 1945 feigelson , e. d. , laurent - muehleisen , s. a. , kollgaard , r. i. , & fomalont , e. b. , 1995 , , 449 , l149 genzel , r. , pauliny - toth , i. i. k. , preuss , e. , & witzel , a. , 1976 , , 81 , 1084 gregory , p. c. , & condon , j. j. , 1991 , , 75 , 1011 hales , s. e. g. , baldwin , j. e. , & warner , p. j. , 1988 , , 234 , 919 harris , d. e. , & grindlay , j. e. , 1979 , , 188 , 25 hill , g. j. , goodrich , r. w. , & depoy , d. l. , 1996 , , 462 , 163 ishisaki , y. , et al . 2007 , , 59 , 113 ishwara - chandra , c. h. , & saikia , d. j. , 1999 , , 309 , 100 isobe , n. , et al . , 2002 , , 580 , l111 isobe , n. , et al
. , 2009 , , 706 , 454 isobe , n. , makishima , k. , tashiro , m. , & hong , s. , 2005 , , 632 , 781 isobe , n. , makishima , k. , tashiro , m. , itoh , k. , iyomoto , n. , takahashi , i. , & kaneda , h. , 2006 , , 645 , 256 isobe , n. , seta , h. , gandhi , p. , & tashiro , m.s . , 2011a , , 727 , 82 isobe , n. , seta , h. , & tashiro , m.s . , 2011b , 63 , s947 kalberla , p. m. w. , burton , w. b. , hartmann , dap , arnal , e. m. , bajaja , e. , morras , r. , & poppel , w. g. l. , 2005 , , 440 , 775 kaneda , h. , et al . , 1995 , , 453 , l13 kellermann , k. i. , pauliny - toth , i. i. k. & williams , p. j. s. , 1969 , , 157 , 1 kellermann , k. i. & pauliny - toth , i. i. k. , 1973 , , 78 , 828k konar , c. , hardcastle , m. j. , croston , j. h. , & saikia , d. j. 2009 , , 400 , 480 koyama , k. , et al . , 2007 , , 59 , s23 khr , h. , witzel , a. , pauliny - toth , i. i. k. , & nauber , u. , 1981 , , 45 , 367 kushino , a. , ishisaki , y. , morita , u. , yamasaki , n. y. , ishida , m. , ohashi , t. , & ueda , y. , 2002 , , 54 , 327 laing , r. a. , riley , j. m. , & longair , m. s. , 1983 , , 204 , 151 labiano , a. , et al . , 2013 , , 549 , a58 lumb , d. h. , warwick , r. s. , page , m. , & de luca , a. , 2002 , , 389 , 93 machalski , j. , koziel - wierzbowska , d. , jamrozy , m. , & saikia , d. j. , 2008 , , 679 , 149 mack , k .- h . , klein , u. , odea , c. p. , & willis , a. g. , 1997 , , 123 , 423 , mack et al . , 1998 , a&a , 329 , 431 mewe , r. , gronenschild , e. h. b. m. , & van den oord , g. h. j. 1985 , , 62 , 197 migliori , g. , grandi , p. , palumbo , g. g. c. , brunetti , g. & stanghellini , c. 2007 , , 668 , 203 miley , g. , 1980 , , 18 , 165 mitsuda , k. , et al . , 2007 , , 59 , s1 miyazawa , t. , haba , y. , & kunieda , h. 2009 , , 61 , 1331 morganti , r. , tadhunter , c. n. , & oosterloo , t. a. 2005 , , 444 , l9 odea , c. p. , et al . , 2001 , , 121 , 1915 orr , e. , et al . 2010 , , 515 , 50 pauliny - toth , i. i. k. , wade , c. m. , & heeschen , d. s. , 1966 , , 13 , 65 pilkington , j. d. h. & scott , j. f. , 1965 , , 69 , 183p roger , r. s. , costain , c. h. , & stewart , d. i. , 1986 , , 65 , 485 schilizzi , r. t. , et al . , 2001 , , 368 , 398 schoenmakers , a. p. , de bruyn , a. g. , rttgering , h. j. a. , & van der laan , h. , 1999 , , 341 , 44 schoenmakers , a. p. , et al . , 2000 , , 146 , 293 serlemitsos , p. j. , et al . , 2007 , , 59 , s9 skrutskie , m. f. , et al . , 2006 , , 131 , 1163 stawarz , . , et al . , 2013 . , 766 , 48 strom , r. g. , & willis , a. g. , 1980 , , 85 , 36 takahashi , t. , et al . , 2007 , , 59 , s35 takeuchi , y. et al . , 2012 , , 749 , 66 tashiro , m. , et al . , 1998 , , 499 , 713 tashiro m. , isobe , n. , seta h. , yaji , y. , & matsuta k. , 2009 , , 61 , s327 tawa , n. , et al . , 2008 , , 60 , s11 viner , m. r. , & erickson , w. c. , 1975 , , 80 , 931 voges , w. , et al . , 1999 , , 349 , 389 waldram , e. m. ; yates , j. a. ; riley , j. m. ; warner , p. j. , 1996 , , 282 , 779 willis , a. g. , strom , r. g. , & wilson , a. s. , 1974 , , 250 , 625 witzel , a. , pauliny - toth , i. i. k. , geldzahler , b. j. , & kellermann , k. i. 1978 , , 83 , 475 | x - ray emission associated with the west lobe of the giant radio galaxy , 3c 236 , was investigated with the suzaku observatory , to evaluate the energetics in the lobe . after removing contamination from point - like
x - ray sources detected with chandra and subtracting the x - ray and non - x - ray backgrounds , the suzaku spectrum from the lobe was reproduced by a power - low model with a photon index of @xmath0 where the first and second errors represent the statistical and systematic ones , respectively . within the errors ,
the x - ray index was consistent with the radio synchrotron one , @xmath1 , estimated in the @xmath2
@xmath3 mhz range .
this agreement supports that the x - ray emission is attributed to the inverse - compton radiation from the synchrotron electrons filling the lobe , where the cosmic microwave background photons are up - scattered .
this result made 3c 236 the largest radio galaxy , of which the lobe has ever been probed through the inverse - compton x - ray photons .
when the photon index was fixed at @xmath4 , the x - ray flux density at 1 kev was measured as @xmath5 njy .
a comparison of the x - ray flux to the radio one ( @xmath6 jy at @xmath7 mhz ) yields the energy densities of the electrons and magnetic field in the west lobe as @xmath8 ergs @xmath9 and @xmath10 ergs @xmath9 , respectively , indicating a mild electron dominance of @xmath11 .
the latter corresponds to the magnetic field strength of @xmath12 @xmath13 g .
these are typical among the lobes of giant radio galaxies .
a compilation of the @xmath14-size relation for the inverse - compton - detected radio galaxies implies that the west lobe of 3c 236 is still actively energized by its jet .
llllll observatory & obsid & date & ( r.a .
, decl . ) & gti ( ks ) & target + observatory & obsid & date & ( r.a .
, decl . ) & gti ( ks ) & target + + suzaku & 707005010 & 2012 may 6 8 & ( , ) & 78.0 & west lobe + suzaku & 707006010 & 2012 may 8 9 & ( , ) & 42.4 & xrb + chandra & 10246 & 2012 march 10 & & 29.4 & west lobe + |
several important materials , such as iron - based high critical temperature superconductors ( fesc),@xcite nickelates,@xcite cobaltites,@xcite manganites,@xcite and many others , have several active orbitals . if electronic correlation effects are important , then multiorbital hubbard models must be employed for their analysis . however , these models are difficult to study in two ( 2d ) or three ( 3d ) dimensions due to their complexity . while at particular electronic densities , such as integer fillings , zero temperature mean field ( mf ) approximations are often reliable , the phase diagrams of multiorbital hubbard models varying _ temperature _ @xmath3 are basically unknown .
this is because thermal mean - field approximations provide qualitatively incorrect results at robust hubbard repulsion @xmath2 and intermediate temperature since they can not generate local moments with short - range magnetic order because of the absence of spatial fluctuations .
moreover , `` sign problems '' complicate the application of quantum monte carlo ( mc ) when several orbitals are active.@xcite since not only @xmath2 but also the hund coupling @xmath4 are important , it is imperative to apply alternative computational tools , even if crude , to study multiorbital models at intermediate temperatures because the potential for new states in these systems is considerable .
recently,@xcite the `` monte carlo - mean field '' ( mc - mf ) technique that mixes the mc and mf approximations was tested using the half - filled one - orbital hubbard model .
the method captured all qualitative features of this model known from quantum monte carlo , including the non - monotonic behavior of the nel temperature @xmath5 with increasing @xmath2 , and it was even quantitatively accurate with errors of only @xmath620% in @xmath5.@xcite this method was applied before to the bec - bcs crossover in the cold atom context and to other problems.@xcite early studies within the spin - fermion model context showed that this type of techniques are also reliable in studies of superconductors.@xcite the new method relies on the hubbard - stratonovich decomposition of the interacting problem via auxiliary fields ( auxf ) @xcite ( in the hartree channel in our study , but could be in other channels as well ) . neglecting the auxf s imaginary - time dependence but retaining their spatial fluctuations leads to a hamiltonian with quantum fermions coupled to classical degrees of freedom , similarly as in double - exchange models for manganites.@xcite classical monte carlo is used for the auxf at any temperature , while the fermionic sector is treated via the traveling cluster approximation ( tca ) that allows access to large lattices.@xcite in this publication , the mc - mf technique is applied for the first time to a 2d two - orbital hubbard model , varying temperature @xmath3 and repulsion @xmath2 .
the unveiled phase diagram is rich , including a narrow nematic phase above @xmath5.@xcite even more importantly , here we report an unexpected novel regime , dubbed orbital selective directional conductor ( osdc ) , where a remarkable anisotropy in transport is observed , with one direction insulating and the other conducting , leading to a dimensional reduction from 2d to 1d .
this dimensional reduction is different from that in tl@xmath7ru@xmath7o@xmath8 and bacusi@xmath7o@xmath9 because they are insulating in all directions , @xcite and different from layered sr@xmath10ru@xmath7o@xmath8 because it requires a high magnetic field .
@xcite our results are also different from layered oxides that are metallic in - plane but insulating out - of - plane,@xcite because their crystal structure already establishes an asymmetry . on the contrary ,
our two - dimensional model is fully symmetric between the @xmath11 and @xmath12 directions but _ spontaneously _ becomes insulating in one direction and metallic in the other , without the help of the lattice , magnetic fields , or impurities .
although five orbitals are needed for a faithful electronic description of iron superconductors , below for simplicity we will focus on the two most important orbitals @xmath0 and @xmath1.@xcite the two - orbital hubbard model studied here is defined as : @xmath13 where @xmath14 creates an electron at site @xmath15 , orbital @xmath16 ( either @xmath17 or @xmath18 ) , and with spin projection @xmath19 .
the number operator is @xmath20 , @xmath21=@xmath22 , and @xmath23 .
@xmath2 , @xmath24 , @xmath4 , and @xmath25 are the kanamori parameters .
the usual constraints @xmath26 and @xmath27 are assumed . in the hund term
only the ising portion is used because the expected magnetic order is collinear , most materials have an easy - axis , and it is technically simpler for this first study .
the hopping parameters reproduce the fermi surface of the undoped fesc,@xcite but our conclusions could be realized as well in other materials with local tetragonal symmetry .
the crystal location of the se , as , or p atoms , used by electrons to tunnel from fe to fe , justify that both nearest- ( nn ) and next - nearest - neighbor ( nnn ) hoppings are needed .
the explicit hopping amplitudes are in previous publications .
@xcite the nn sites hoppings @xmath28 and @xmath29 are only intraorbital . along the plaquette diagonals , the intraorbital ( interorbital ) hopping is @xmath30 ( @xmath31 ) .
their values are @xmath28=@xmath32 , @xmath29=@xmath33 , and @xmath30=@xmath31=@xmath34 and the bandwidth of this tight binding model is @xmath35=@xmath36.@xcite hereafter , @xmath28 will be denoted by @xmath37 , and it will be the energy unit . to convert to ev , the _
ab - initio _ derived bandwidth for the @xmath0-@xmath1 bands is @xmath38 ev.@xcite the density is fixed to two electrons per site ( @xmath39 ) .
mean - field approximations have already been applied to two - orbital hubbard models at @xmath3=@xmath40,@xcite showing several phases with increasing @xmath41 : a paramagnetic metal , a metal with ( @xmath42 ) spin order , and an insulator also with ( @xmath42 ) spin order .
these previous @xmath3=@xmath40 studies and others @xcite showed that @xmath43 is relevant for fesc , and we will fix @xmath44 in all the results below . since fesc materials vary substantially in their degree of electronic correlation , results will be presented varying the ratio @xmath41 .
our main focus are the temperature effects since their influence on model eq .
( 1 ) are unknown .
our study is performed in two dimensions , mainly on @xmath45 lattices .
the tca traveling cluster @xcite is @xmath46 . via the parallelized version of tca,@xcite lattices as large as @xmath47 were reached .
the mean - field approach was chosen to be the hartree approximation that works well for the half - filled one - orbital model.@xcite
our main results are in the phase diagram of model eq.(1 ) at @xmath39 ( fig .
[ rf-1 ] ) .
it was constructed based on data for the two spin structure factors of relevance , @xmath48 and @xmath49 , the spin nematic order parameter @xmath50,@xcite their temperature derivatives , and the resistivity and density of states ( dos ) .
we also monitored local moment formation at intermediate temperature and large @xmath41.@xcite this phase diagram is surprisingly rich because it contains _ three _ regimes with @xmath51 long - range magnetic order [ degenerate with @xmath52 .
the insulator at large @xmath41 is induced by the robust @xmath4 that produces @xmath53=1 local spins interacting via a frustrated heisenberg model known to have @xmath51-@xmath54 magnetic order .
the other two states at intermediate couplings are more subtle .
intuitively , their magnetic order can be considered as arising from fermi surface nesting effects .
however , the presence of @xmath55 , instead of one , regions is unexpected .
the first one , dubbed af - m ( antiferromagnetic metallic ) , is metallic in both directions , albeit anisotropic , as observed in fesc experiments.@xcite but the second region , the osdc , is metallic along the spin staggered direction but insulating along the spin uniform direction , leading to a surprising dimensional reduction at intermediate couplings .
a typical magnetic order parameter ( @xmath56 ) displays a negative curvature increasing temperature(fig .
[ rf-2]a ) , with a diverging slope at the critical temperature @xmath57 in the bulk limit .
thus , for a finite system the temperature where the first derivative d@xmath56/d@xmath3 is maximized , or where the associated susceptibility maximizes , provides an estimation of @xmath57 ( here @xmath5 and @xmath58)(fig .
[ rf-2]b ) .
in addition , upon cooling @xmath58 can also be estimated from the temperature @xmath59 where @xmath48 and @xmath49 split . while for a finite system @xmath58 and @xmath59 may be different , they should merge in the bulk limit .
typical results are in fig .
[ rf-2 ] for a @xmath45 lattice , at a fixed ratio @xmath41 .
our mc statistics and lattice sizes are sufficient to observe a robust order parameter behavior for @xmath48 and @xmath50 : nonzero at @xmath60 and decreasing with increasing temperature with a negative curvature , with the exception of a small temperature window where the curvature is positive due to size effects .
although very close in temperature , systematically for all the couplings @xmath41 with long range spin order at @xmath60 and for all lattices , we find @xmath58 slightly larger than @xmath5 suggesting a small region of nematicity .
in general , we also found that @xmath59 tends to be larger than @xmath58 .
the observed narrow window of nematicity in fact appears to survive a finite - size scaling analysis ( fig .
[ rf-4](a ) ) , at least for the investigated coupling @xmath61 where information for lattices as large as @xmath47 were gathered ( since these simulations are very time consuming the scaling analysis could be done only for one coupling ) .
remarkably , within the error bars the bulk limit extrapolated @xmath58 and @xmath59 seem to lead to the same nematic critical temperature .
more specifically , our results unveil a small nematic window of @xmath62 at @xmath61 .
the narrowness of this nematic regime is likely exaggerated by the z(2 ) nature of the ising approximation used here for the hund term .
such a fragile nematic phase , reported here for the first time in a hubbard model , is compatible with previous studies using spin models,@xcite spin - fermion models,@xcite and with experiments,@xcite that have also reported very narrow nematic temperature windows .
as explained before it is surprising that there are three distinct regions below @xmath5 : two metals ( af - m and osdc ) and one insulator ( af - i ) . the distinction between the two metals and the insulator can be understood via the density - of - states in figs .
[ rf-4](b , c ) .
panel ( c ) displays a canonical insulating behavior : in the temperature range shown , a pseudogap ( pg ) is observed in the local moments regime,@xcite transforming into a full gap when the magnetic order develops at @xmath5 .
this is the af - i state ( fig .
[ rf-1 ] ) . in panel
( b ) , upon cooling toward @xmath5 a pseudogap opens probably because of fermi surface nesting effects .
but even at low temperatures , and independently analyzing the zero - temperature hartree equations , in both metallic regions the total dos has a finite weight at the fermi energy @xmath63 .
what is then the difference between af - m and osdc ?
their physical distinction is illustrated in fig .
[ rf-5 ] ( a ) where the resistivity @xmath64 vs. @xmath3 is presented at three values of @xmath41 , corresponding to the three low - temperature regions of fig .
@xmath64 is calculated from the optical conductivity @xmath65 , integrating in a narrow range near @xmath66 and then inverting.@xcite at @xmath67 , @xmath64 in the @xmath12 spin uniform direction is larger than in the @xmath11 spin staggered direction , as in previous calculations @xcite and experiments.@xcite this is understood from the orbital resolved dos of fig .
[ rf-4](d ) : in the magnetic @xmath51 state that breaks rotational invariance , near @xmath63 the orbital @xmath1 , related to conduction in the spin uniform direction , is more suppressed than @xmath0 , related to conduction in the spin staggered direction .
but since both orbitals have a sizable dos weight at the @xmath63 , both directions are metallic . at large @xmath68 , the dos fig .
[ rf-4](c ) displays a sharp gap at low temperature , and both directions must be insulating .
note that experiments affected primarily by the vicinity of @xmath63 may suggest strong orbital order , but the @xmath69-integral of the orbital - resolved dos , i.e. the orbital population @xmath70 and @xmath71 , is only different by 0.5% ( fig .
[ rf-4 ] ( f)).@xcite for completeness , we also monitored @xmath72 vs. temperature .
the results ( not shown ) indicate that upon cooling from high @xmath3 , @xmath73 remains smaller than 0.001 at @xmath41=@xmath74 until @xmath58 is reached , eventually converging , as @xmath3 is further reduced , to the results derived from fig .
[ rf-4 ] ( f ) . within our mc accuracy
we conclude that in our model there is no additional orbital - order critical temperature above @xmath58 .
the interpolation between small and large @xmath41 unveils a surprise : at intermediate couplings such as @xmath75 the spin staggered direction remains metallic , but the spin uniform direction becomes @xmath76 ( fig .
[ rf-5 ] ( a ) ) . intuitively , this is because the interpolation between the density - of - states of fig .
[ rf-4](d ) , where both orbitals have nonzero weight at @xmath63 , and ( c ) , where both orbitals have negligible weight at @xmath63 , is not smooth . instead
, there is an intermediate coupling range where the @xmath1 weight at @xmath63 is almost negligible while that of @xmath0 is still finite ( fig .
[ rf-4](e ) ) .
the @xmath1 weight is very small but not zero , even at @xmath60 , because of the broadening used for density of states .
therefore , strictly speaking it can not be used as a sharp order parameter : af - m and osdc are likely analytically connected because they break the same symmetries .
however , our study shows that the orbital population difference at @xmath63 in the osdc is sufficiently large to induce one - dimensional transport .
results for the full @xmath65 ( see fig .
[ rf-5 ] ( b ) and appendix ) show that at @xmath77 only one orbital dominates . moreover , from @xmath70 and @xmath71 ( fig .
[ rf-4 ] ( f ) ) note that at large @xmath41 both orbital populations converge to one since @xmath4 is large , while in the pm - m regime they are also one by symmetry .
however , in the osdc region _ both _ @xmath70 and @xmath71 are different from one ( and different among themselves because @xmath78 is spontaneously broken ) : the osdc regime is not the same as an orbital selective mott phase @xcite where one orbital has population exactly one .
another interesting observation is that for both af - m and osdc there is an insulating region @xmath79 in both directions immediately above @xmath5 due to the opening of the pseudogap in the local moments region and concomitant coexisting patches of ( @xmath80,0 ) and ( 0,@xmath80 ) order,@xcite in agreement with spin fermion studies @xcite and experiments.@xcite the analysis presented in fig .
[ rf-2 ] and fig .
[ rf-4](a ) are just examples of the vast computational study carried out at various values of @xmath41 and used to construct the phase diagram of fig .
[ rf-1 ] . for completeness , in fig . [ srf-1 ] this substantial effort
is illustrated further by providing data corresponding to other lattice sizes at @xmath61 .
the information gathered from these efforts for @xmath5 , @xmath58 , and @xmath59 were used for the finite - size scaling analysis of fig .
[ rf-4 ] ( a ) .
the phase diagram of a layered two - orbitals hubbard model was studied with emphasis on temperature effects .
we report a novel intermediate coupling @xmath2 region , called the osdc , that is conducting in one direction via the @xmath0 orbitals , but insulating in the other because the associated @xmath1 orbitals have nearly vanishing weight at @xmath63 .
although our low temperature calculations do not include quantum fluctuations , the osdc starts at relatively high temperatures @xmath81 and for this reason our approach emphasizing thermal effects should be sufficient .
moreover , we tested that using other hoppings,@xcite the osdc is also found .
experimentally , materials of the family of iron superconductors are the most likely to realize the osdc ( in fact , indications of osdc are already present in ref . for the case of doped materials ) , and a mixing in the chemical formula of as , associated with weak coupling , and se , associated with strong coupling , may be needed .
but the osdc could be realized in other layered materials where a transition metal atom m is coordinated with four ligand atoms x , establishing mx@xmath82 tetrahedral cages with near degenerate @xmath0 and @xmath1 orbitals , and where a magnetic state that breaks lattice rotational invariance is stabilized .
e. dagotto thanks guangkun liu for useful conversations . a. mukherjee and n. patel were supported by the national science foundation grant no .
e. dagotto and a. moreo were supported by the u.s .
department of energy , office of basic energy science , materials science and engineering division .
the lattices used all have periodic boundary conditions . in our calculations ,
a total of 4000 mc system sweeps were typically performed : 2000 to thermalize the system , and the rest for calculating observables .
a mc system sweep consists of sequentially visiting every lattice site and updating the local auxiliary fields followed by the fermionic diagonalization or tca procedure to accept / reject via the metropolis algorithm .
we start the simulation at high temperature with a random configuration of auxiliary fields and then slowly cool down to lower temperatures to avoid being trapped in metastable states .
the mc runs start at @xmath83 , which corresponds to about 1000 k and cool down to @xmath84 , where temperature steps as small as @xmath85 were used for the temperatures relevant to the magnetic and nematic transitions .
this slow process allows us to obtain reliable results independent of the initial conditions of the calculation . in our finite systems , there is no energy difference between the @xmath51 and @xmath54 magnetic states . as a consequence , mc simulations that start at high temperature in a random state for the auxiliary fields may end up in @xmath51 or @xmath54 with equal chance upon cooling . in practice , we simply discarded all cooling down mc processes that led to a @xmath54 state at low temperatures .
the antiferromagnetic order is studied via the spin structure factors , @xmath86 the two wavevectors of interest in fesc are @xmath87 and @xmath54 .
the expectation value is generated by using the eigenvectors of the mc equilibrated configurations .
for the study of the nematic regime above @xmath5 , we compute the nematic order parameter @xmath88 where @xmath89 and @xmath90 are unit vectors connecting site @xmath91 with its nearest neighbors .
the @xmath92 summation is over all nearest neighbors , and @xmath93 in the @xmath51 magnetic phase . to better locate critical temperatures
, we also calculate numerically the magnetic ( @xmath94 ) and nematic ( @xmath95 ) susceptibilities using the standard variance calculation .
in this work , we have also evaluated the orbital resolved density of states ( dos ) , @xmath96 , where @xmath97 are the eigenvalues of the fermionic sector and the summation runs up to @xmath98 , i.e. the total number of eigenvalues of a @xmath99 system with spin .
@xmath100 is the weight of the @xmath101 eigenstate for orbital @xmath16 in the state @xmath102 .
@xmath103 is calculated by implementing the usual lorentzian representation of the @xmath104 function .
the broadening needed to obtain @xmath105 from the lorentzian is @xmath106 , where @xmath35 is the fermionic bandwidth at @xmath107 .
finally , @xmath108 , the total dos , is the sum of the different orbital densities of states .
numerically , e.g. , for the 8@xmath109 system the broadening is about @xmath110 .
two hundred @xmath111 samples are obtained from the 2000 measurement system sweeps at every temperature . we discard 10 mc steps between measurements to reduce self - correlations in the data .
the 200 @xmath111 samples are used to obtain the thermally averaged @xmath112 at a fixed temperature .
these are further averaged over data obtained from 10 independent runs with different random number seeds .
a very similar procedure is followed to calculate the optical conductivity , that involves matrix elements of the current operator . to determine the crossover temperature between the weak coupling paramagnetic state and the regime with preformed local moments above the magnetic order
, we compute the specific heat , @xmath113 , by numerically differentiating the average energy with respect to temperature , as well as the orbital resolved double occupation @xmath114 .
the details of the procedure we followed is presented in our earlier work on the one - orbital hubbard model,@xcite but , briefly , @xmath115 has a high - temperature peak that corresponds to local moment formation that can be tracked varying @xmath2 . in addition
, the double occupation has to be below a cutoff for the system to have local moments.@xcite this is of particular importance at small @xmath2 values because @xmath115 can have a considerable contribution from the electronic delocalization and in this regime it can not be used to track local moment formation . the many - body technique used in this publication
was already introduced and tested in previous efforts.@xcite however , since an easy axis is present in most materials and considering that empirically the monte carlo convergence is improved under such circumstances ( colloquially , ising is easier than heisenberg ) , in this project it was decided to use the ising approximation in the hund term . in the language of the one - orbital hubbard model that corresponds to using the hartree approximation instead of the hartree - fock approximation employed before .
this requires , then , a test of the hartree assumption for the three - dimensional one - orbital standard hubbard model at half - filling .
there is no need to provide explicitly the hamiltonian for such well known model , thus we move immediately to discuss the results , which are provided in fig .
[ srf-4 ] .
the results shown in fig . [ srf-4 ] are encouraging .
the critical temperatures found with the mc - mf technique capture the `` up and down '' behavior of @xmath5 with increasing @xmath116 and they converge close to the expected scaling at large @xmath116 .
moreover , in the entire range of @xmath116 investigated the mc - mf results are close to those of quantum monte carlo ( with the largest discrepancy being about 20% ) .
the successful test presented in fig . [ srf-4 ] suggests that the mc - mf technique captures the essence of the one - orbital problem , not only qualitatively but also quantitatively .
this gives us confidence that the results for two orbitals in the main text , that have not been studied before in the literature , are reliable . to illustrate the physics of the three different states reported here
, the optical conductivity was calculated .
the results are in fig .
[ srf-3 ] .
panel ( a ) is in the af - m regime : while the @xmath51 magnetic order breaks the symmetry between the @xmath11 and @xmath12 directions , the difference is not dramatic and leads to both directions being metallic .
panel ( c ) corresponds to the mott insulating regime : here a gap is present both when the electric field points along the @xmath11 and @xmath12 directions .
the first excitations occur at the scale of the hund coupling , along the @xmath11 spin staggered direction .
the most novel result is shown in panel ( b ) , already presented in the main text and reproduced here for the benefit of the readers , corresponding to the new osdc region : here at @xmath117 there is a finite weight in the @xmath11 direction but negligible weight in the @xmath12 direction , compatible with the calculation of the resistivity shown in the main text .
some subtle technical details are worth discussing , particularly with regards to the window in @xmath69 , around zero , used to define the resistivity .
consider a @xmath99 lattice .
its associated mean level spacing is roughly estimated as @xmath118 where @xmath35 is the bandwidth . in practice
the individual @xmath104-functions broadening used in the conductivity calculation is @xmath119 times a factor , which it has been chosen to be 4 in our calculations .
as example , for a @xmath120 system @xmath121 , and the broadening used is then 0.03 .
the integration range is decided as follows : the smallest frequency ( the starting @xmath69 ) is chosen to be at least one order of magnitude smaller than @xmath119 .
the frequency increment is chosen to be @xmath122 .
the data of these 10 frequency points is used for the integration and the outcome is ascribed as the average conductivity for the frequency value at the lower end of the interval . in the example
given above , then the frequency step is 0.003 .
the integration is performed over 10 frequency steps .
the calculation is started at an initial frequency of @xmath123 . to derive the mf hamiltonian we follow the treatment in _ quantum theory of finite systems _ , by blaizot , j .-
& ripka , g. , the mit press ( 1985 ) .
the advantage of this approach is that one can derive a single general expression for the mean field parameters , for any number of orbitals .
this expression can be easily coded in , thus avoiding the need to derive all possible mean field decouplings by hand . for this purpose
we introduce the following notation : @xmath131 where @xmath132 are elements of the single particle density matrix .
we now make the hartree - fock approximation assumption : the state of the system can be represented by a single slater determinant , @xmath133 . by using the wick s theorem ,
we can then write down the expectation value of @xmath134 in @xmath133 , denoted by @xmath135 $ ] , as : @xmath136= \sum_{\langle i , j \rangle,\alpha,\beta,\sigma } t^{i , j}_{\alpha,\beta}\langle\beta,\sigma|\rho_{i , j}|\alpha,\sigma\rangle + \\ \sum_{i,\sigma,\sigma^\prime}\sum_{\alpha,\alpha^\prime,\beta,\beta^\prime}u_{\sigma,\sigma^{\prime}}(\alpha,\alpha^{\prime},\beta,\beta^{\prime})\times\\ [ \langle\beta,\sigma|\rho_{i , i}|\alpha,\sigma\rangle\langle\beta^{\prime},\sigma^{\prime}|\rho_{i , i}|\alpha^{\prime},\sigma^{\prime}\rangle\\ - \langle\beta^{\prime},\sigma^{\prime}|\rho_{i , i}|\alpha,\sigma\rangle\langle\beta,\sigma|\rho_{i , i}|\alpha^{\prime},\sigma^{\prime}\rangle ] .
\end{aligned}\ ] ] we now take the derivative of @xmath135 $ ] with respect to a generic density matrix element , @xmath137 , to get an explicit formula for hartree - fock mean field parameters , @xmath138 .
running over all values of orbitals and spin in @xmath137 , for taking the derivatives , generates all possible mean field decouplings in the hartree - fock channel .
thus , the general formula is as follows : @xmath139\times \\ \langle\beta^{\prime},\sigma^{\prime}|\rho_{i , i}|\alpha^{\prime},\sigma^{\prime}\rangle\\ -[u_{\bar{\sigma}'\bar{\sigma}}(\bar{\beta},\alpha',\beta',\bar{\alpha})+u_{\bar{\sigma}\bar{\sigma}'}(\alpha',\bar{\beta},\bar{\alpha,}\beta')]\times \\ \langle\beta^{\prime},\bar{\sigma}^{\prime}|\rho_{i , i}|\alpha^{\prime},\bar{\sigma}\rangle ) .
\end{aligned}\ ] ] r. tiwari and p. majumdar , arxiv:1301.5026 ; r. tiwari and p. majumdar , arxiv:1302.2922 ; s. tarat and p. majumdar , europhys . lett . * 105 * , 67002 ( 2014 ) ; s. tarat and p. majumdar , arxiv:1402.0817 ; s. tarat and p. majumdar , arxiv:1406.5423 .
chu , j. g. analytis , k. de greve , p. l. mcmahon , z. islam , y. yamamoto , and i. r. fisher , science * 329 * , 824 ( 2010 ) ; i. r. fisher , l. degiorgi , and z. x. shen , rep .
* 74 * , 124506 ( 2011 ) .
r. m. fernandes , a. v. chubukov , and j. schmalian , nature phys . *
10 * , 97 ( 2014 ) ; r. m. fernandes , a. v. chubukov , j. knolle , i. eremin , and j. schmalian , phys .
b * 85 * , 024534 ( 2012 ) ; and references therein .
h. m. ronnow , ch .
renner , g. aeppli , t. kimura , and y. tokura , nature * 440 * , 1025 ( 2006 ) ; t. valla , p. d. johnson , z. yusof , b. wells , q. li , s. m. loureiro , r. j. cava , m. mikami , y. mori , m. yoshimura , and t. sasaki , nature * 417 * , 627 ( 2002 ) .
a. nicholson , w. ge , x. zhang , j. riera , m. daghofer , a. m. ole , g. b. martins , a. moreo , and e. dagotto , phys .
* 106 * , 217002 ( 2011 ) ; a. nicholson , w. ge , j. riera , m. daghofer , a. moreo , and e. dagotto , phys . rev .
* b 85 * , 024532 ( 2012 ) .
y. kamiya , n. kawashima , and c. d. batista , phys .
b * 84 * , 214429 ( 2011 ) ; a. l. wysocki , k. d. belashchenko , and v. p. antropov , nat
* 7 * , 485 ( 2011 ) ; m. capati , m. grilli , and j. lorenzana , phys . rev .
b * 84 * , 214520 ( 2011 ) .
general arguments @xcite suggest that either @xmath141 in a first - order transition , or @xmath142 with both transitions being second order .
since we do not observe indications of metastabilities in the mc time evolution , or in histograms of observables , our results are compatible with the existence of a nematic regime .
v. i. anisimov , i. a. nekrasov , d. e. kondakov , t. m. rice , and m. sigrist , eur .
j. b * 25 * , 191 ( 2002 ) ; a. georges , l. d medici , and j. mravlje , annu .
rev . condens .
matter phys .
* 4 * , 137 ( 2013 ) . | employing a recently developed many - body technique that allows for the incorporation of thermal effects , the rich phase diagram of a two dimensional two orbital ( degenerate @xmath0 and @xmath1 ) hubbard model is presented varying temperature and the repulsion @xmath2 .
our main result is the finding at intermediate @xmath2 of a novel antiferromagnetic orbital selective state where an effective dimensional reduction renders one direction insulating and the other metallic .
possible realizations of this state are discussed .
in addition we also study nematicity above the nel temperature . after a careful
finite - size scaling analysis the nematicity temperature window appears to survive in the bulk limit although it is very narrow . |
in this note we present a new class of gauss processes , generalizing the wiener bridge , for which karhunen love ( kl ) expansion can be given explicitly .
we point out that there are only few gauss processes for which the kl expansion is explicitly known . to give some examples , we mention the wiener process ( see , e.g. , ash and gardner ( * ? ? ?
* example 1.4.4 ) ) , the ornstein uhlenbeck process ( see , e.g. , papoulis ( * ? ? ?
* problem 12.7 ) or corlay and pags ( * ? ? ?
* section 5.4 b ) ) , the wiener bridge ( see , e.g. , deheuvels ( * ? ? ?
* remark 1.1 ) ) , kac kiefer
wolfowitz process ( see , kac , kiefer and wolfowitz @xcite and nazarov and petrova @xcite ) , weighted wiener processes and bridges ( deheuvels and martynov @xcite ) , jandhyala macneill process ( jandhyala and macneill ( * ? ? ?
* section 4 ) ) , a generalization of wiener bridge ( pycke @xcite ) , generalized anderson darling process ( pycke @xcite ) , rodrguez viollaz process ( pycke @xcite ) , scaled wiener bridges or also called @xmath13-wiener bridges ( barczy and igli @xcite ) , detrended wiener processes ( ai , li and liu @xcite ) , additive wiener processes and bridges ( liu @xcite ) , additive slepian processes ( liu , huang and mao @xcite ) and spartan spatial random fields ( tsantili and hristopulos @xcite ) .
we also mention that kl expansions of gauss processes have found several applications in small deviation theory , for a complete bibliography , see lifshits @xcite . here
we only mention two papers of nazarov and nikitin @xcite , @xcite .
let @xmath14 , @xmath15 and @xmath16 denote the set of non - negative integers , positive integers and real numbers , respectively . for @xmath17 , we will use the notation @xmath18 .
let @xmath2}$ ] be a standard wiener process , and let @xmath3\to { \mathbb{r}}$ ] be a twice continuously differentiable function such that @xmath4 and @xmath5 .
let us introduce the process @xmath19 .
\end{aligned}\ ] ] one can consider @xmath20}$ ] as a generalization of the wiener bridge corresponding to the function @xmath21 . in the special case @xmath7 ,
@xmath1 $ ] , we have @xmath22 , @xmath1 $ ] , i.e. , it is a wiener bridge over @xmath8 $ ] from @xmath9 to @xmath9 .
however , in general , @xmath23 is not a bridge .
note that @xmath23 is a bridge in the sense that @xmath24 with some @xmath25 ( i.e. , @xmath23 takes some constant value at time 1 with probability one ) if and only if @xmath26 , and in this case @xmath27 .
indeed , @xmath24 with some @xmath25 if and only if @xmath28 .
since @xmath29 ( see proposition [ pro1 ] ) , we have @xmath24 with some @xmath25 if and only if @xmath26 , as desired .
further , since @xmath30 , in this case we have @xmath27 . in the present paper
, we do not intend to study whether the process @xmath31}$ ] given by can be considered as a bridge in the sense that it can be derived from some appropriate stochastic process ( for more information on this procedure , see barczy and kern @xcite ) . to give a possible
motivation for the definition of the process @xmath23 , let us write in the form @xmath32,\ ] ] where @xmath33 can be _ formally _ interpreted as the orthogonal projection of the derivative of @xmath34 ( in notation @xmath35 ) onto @xmath36 in @xmath37 , since @xmath5 ( it is only a _ formal _
one because the derivative of @xmath38 does not exist ) .
so , from this point of view , the derivative of @xmath39 ( in notation @xmath40 ) is _ formally _ the orthogonal component of @xmath35 with respect to @xmath36 in @xmath37 , and one can call @xmath40 as the @xmath36-detrendization of @xmath35 . further , we point out that if @xmath21 additionally satisfies @xmath41 , then the gauss process @xmath31}$ ] given in coincides in law with one of the gauss processes introduced in nazarov ( * ? ? ?
* formula ( 1.3 ) ) , for more details , see appendix [ appendix_nazarov ] . in the spirit of nazarov @xcite
, one can say that @xmath31}$ ] is a perturbation of the wiener process @xmath2}$ ] by the function @xmath21 .
[ pro1 ] the process @xmath31}$ ] is a zero - mean gauss process with continuous sample paths almost surely and with covariance function @xmath42 , @xmath43 $ ] .
the proof of proposition [ pro1 ] can be found in section [ section_proofs ] .
the continuity of the covariance function @xmath44 yields that @xmath31}$ ] is @xmath37-continuous , see , e.g. , theorem 1.3.4 in ash and gardner @xcite .
we also have @xmath45 ^ 2)$ ] .
so , the integral operator associated to the kernel function @xmath44 , i.e. , the operator @xmath46)\rightarrow l^2([0,1])$ ] , @xmath47 , \qquad \phi\in l^2[0,1 ] , \end{aligned}\ ] ] is of the hilbert
schmidt type , thus @xmath31}$ ] has a karhunen love ( kl ) expansion based on @xmath8 $ ] : @xmath48,\ ] ] where @xmath49 , are independent standard normally distributed random variables , @xmath50 , are the non - zero ( and hence positive ) eigenvalues of the integral operator @xmath51 and @xmath52,\ k\in{\mathbb{n}}$ ] , are the corresponding normed eigenfunctions , which are pairwise orthogonal in @xmath53)$ ] , see , e.g. , ash and gardner ( * ? ? ?
* theorem 1.4.1 ) . for completeness , we recall that the integral operator @xmath51 has at most countably many eigenvalues , all non - negative ( due to positive semi - definiteness ) with @xmath9 as the only possible limit point , and the eigenspaces corresponding to positive eigenvalues are finite dimensional .
observe that has infinitely many terms . indeed ,
if it had a finite number of terms , i.e. , if there were only a finite number of eigenfunctions , say @xmath54 , then , by the help of , we would obtain that the wiener process @xmath2}$ ] is concentrated in an at most @xmath55-dimensional subspace of @xmath53)$ ] , and so even of @xmath56)$ ] , with probability one .
this results in a contradiction , since the integral operator associated to the covariance function ( as a kernel function ) of a standard wiener process has infinitely many eigenvalues and eigenfunctions .
we also note that the normed eigenfunctions are unique only up to sign . the series in converges in @xmath57 to @xmath39 , uniformly on @xmath8 $ ] , i.e. , @xmath58}{\operatorname{\mathbb{e}}}\left ( \left\vert y_t - \sum_{k=1}^n\sqrt{\lambda_k}\,\xi_k e_k(t)\right\vert^2\right)\to 0 \qquad \text{as $ n\to\infty$.}\ ] ] moreover , since @xmath44 is continuous on @xmath8 ^ 2,$ ] the eigenfunctions corresponding to non - zero eigenvalues are also continuous on @xmath8,$ ] see , e.g. ash and gardner @xcite ( this will be important in the proof of proposition [ klgenthm ] , too ) .
since the terms on the right - hand side of are independent normally distributed random variables and @xmath31}$ ] has continuous sample paths with probability one , the series converges even uniformly on @xmath8 $ ] with probability one ( see , e.g. , adler ( * ? ? ? * theorem 3.8 ) ) .
[ klgenthm ] if @xmath59 is a non - zero ( and hence positive ) eigenvalue of the integral operator @xmath51 and @xmath60 is an eigenfunction corresponding to it , then @xmath61 , \end{aligned}\ ] ] with boundary conditions @xmath62 conversely , if @xmath59 and @xmath63 , @xmath1 $ ] , satisfy and , then @xmath59 is an eigenvalue of @xmath51 and @xmath60 is an eigenfunction corresponding to it .
note that for the converse statement in proposition [ klgenthm ] we do not need to know in advance that @xmath59 is non - zero .
the proof of proposition [ klgenthm ] can be found in section [ section_proofs ] . to describe the solutions of and , for a fixed @xmath64 we introduce the notations @xmath65 [ thm_kl_gen ] in the kl expansion of the process @xmath31}$ ] given in , the non - zero ( and hence positive ) eigenvalues are the solutions of the equation @xmath66 and the corresponding normed eigenfunctions take the form @xmath67 , \qquad t\in[0,1 ] , \end{split } \end{aligned}\ ] ] where @xmath68 is chosen such that @xmath69 .
( note that @xmath70 may depend on @xmath59 , but we do not denote this dependence . ) the proof of theorem [ thm_kl_gen ] can be found in section [ section_proofs ] .
we emphasize that in theorem [ thm_kl_gen ] we give kl expansion for a new class of gauss processes with the advantage of an explicit form of the eigenfunctions appearing in , while in the recent papers on kl expansions such as for detrended wiener processes ( ai , li and liu @xcite ) , additive wiener processes and bridges ( liu @xcite ) and additive slepian processes ( liu , huang and mao @xcite ) , the form of the eigenfunctions remains somewhat hidden . as we have already mentioned , in the case of @xmath41 , the gauss process @xmath31}$ ]
coincides in law with one of the gauss processes ( 1.3 ) in nazarov @xcite , where he presented a procedure for finding the kl expansion for his more general gauss processes . in theorem [ thm_kl_gen ]
we make the kl expansion of @xmath31}$ ] as explicit as possible by solving the underlying eigenvalue problem directly .
we note that theorem [ thm_kl_gen ] is applicable in the case of @xmath71 as well .
[ rem_lambda0 ] note that @xmath9 may be an eigenvalue of the integral operator @xmath51 defined in , which is in accordance with corollary 2 in nazarov @xcite . for an example , see section [ section_proofs ] . in the next remark we recall an application of the kl expansion .
[ rem_laplace ] the laplace transform of the @xmath53)$]-norm square of @xmath31}$ ] takes the form @xmath72 indeed , by , we have @xmath73 , \end{aligned}\ ] ] and hence using the fact that @xmath74 is an orthonormal system in @xmath53)$ ] , we get @xmath75 which is nothing else but the parseval identity in @xmath53)$ ] . since @xmath76 , @xmath77 , are independent standard normally distributed random variables , we get @xmath78 next we study the special case @xmath79 , @xmath1 $ ] , yielding @xmath80 . [ klcor1 ] if @xmath79 , @xmath1 $ ] , then in the kl expansion of @xmath81 , @xmath1 $ ] , the non - zero ( and hence positive ) eigenvalues are the solutions of the equation @xmath82 and the corresponding normed eigenfunctions take the form @xmath83 = c \left[\sin\left(\frac{t}{\sqrt{\lambda}}\right ) + \sqrt{\lambda}\pi \cos\left(\frac{1}{\sqrt{\lambda}}\right ) \sin(\pi t)\right ] \end{aligned}\ ] ] for @xmath1 $ ] , where @xmath68 is chosen such that @xmath69 , i.e. , @xmath84 the proof of corollary [ klcor1 ] can be found in section [ section_proofs ] .
in fact , we will provide two proofs .
the first one is an application of theorem [ thm_kl_gen ] , which is based on the method of variation of parameters , while the second proof is based on the method of undetermined coefficients .
the equation has a root in every interval @xmath85 , @xmath77 .
indeed , the left hand side of at @xmath86 is equal to @xmath87 , and the right hand side of at @xmath88 is equal to @xmath89 , and consequently , the continuous function of the left hand side of as a function of @xmath64 changes sign and hence has a root on every interval @xmath85 , @xmath77 .
the equation has no root greater than @xmath90 , since @xmath91 since @xmath92 , @xmath77 , are the eigenvalues of the integral operator corresponding to the covariance function @xmath93 , @xmath43 $ ] , of a standard wiener process , we can say that there is a kind of interlacement between the eigenvalues of the integral operators corresponding to the underlying standard wiener process @xmath38 and to the perturbed process @xmath23
. for more details on this phenomenon in a general setup , see , e.g. , nazarov @xcite . using the rootsolve package in r
, we determined the first five roots of listed in decreasing order : @xmath94 in figure [ fig ] , we plotted the left hand side of as a function of @xmath95 . , @xmath64.,width=453,height=264 ] using the above five roots , by , for small values of @xmath96 , the laplace transform @xmath97 of @xmath98 , can be approximated by @xmath99 finally , we study the special case @xmath100 , @xmath1 $ ] , which is nothing else but the case of a usual wiener bridge over @xmath8 $ ] from @xmath9 to @xmath9 .
note that the kl expansion of a wiener bridge has been known for a long time , see , e.g. , deheuvels ( * ? ? ?
* remark 1.1 ) .
[ klcor2 ] if @xmath100 , @xmath1 $ ] , then in the kl expansion of the wiener bridge @xmath101 , @xmath1 $ ] , the non - zero ( and hence positive ) eigenvalues are the solutions of the equation @xmath102 and the corresponding normed eigenfunctions take the form @xmath103 , \end{aligned}\ ] ] satisfying @xmath69 .
the proof of corollary [ klcor2 ] can be found in section [ section_proofs ] .
* proof of proposition [ pro1 ] .
* the fact that @xmath23 is a zero - mean gauss process with continuous sample paths almost surely follows from its definition . indeed , since for all @xmath104 , @xmath105 , @xmath106 to check that @xmath23 is a gauss process it is enough to show that @xmath107 is normally distributed for all @xmath104 , @xmath105 .
this follows from the fact that @xmath38 is a gauss process and from the definition of @xmath108 taking into account that an @xmath37-limit of normally distributed random variables is normally distributed ( for a more detailed discussion on a similar procedure , see , e.g. , the proof of lemma 48.2 in bauer @xcite ) .
further , since @xmath109 , @xmath1 $ ] , the process @xmath110}$ ] is a martingale , and consequently @xmath111 , yielding that @xmath112 , @xmath1 $ ] .
moreover , for @xmath43 $ ] , @xmath113 where for the last but one equality we used @xmath114 and @xmath5 .
* proof of proposition [ klgenthm ] .
* let @xmath59 be a non - zero ( and hence positive ) eigenvalue of the integral operator @xmath51 .
then we have @xmath115,\ ] ] and hence @xmath116 .
\end{aligned}\ ] ] then @xmath117 .
\end{split } \end{aligned}\ ] ] the right - hand ( and hence the left - hand ) side of is differentiable with respect to @xmath118 , since @xmath60 is continuous ( see the introduction ) , and , by differentiating with respect to @xmath118 , we have @xmath119 , \end{aligned}\ ] ] yielding that @xmath120.\end{aligned}\ ] ] differentiating with respect to @xmath118 yields ( the differentiation is allowed , since @xmath21 is twice continuously differentiable ) . with the special choice @xmath121 in , using that @xmath4 and @xmath64 , we have the boundary condition @xmath122 , yielding the first part of .
further , by with @xmath123 , we have @xmath124 yielding the second part of . conversely , let us suppose that @xmath59 and @xmath63 , @xmath1 $ ] , satisfy and .
then integration of from @xmath118 to @xmath125 gives @xmath126,\ ] ] i.e. , @xmath127.\ ] ] by , we have @xmath128,\ ] ] i.e. , @xmath129,\ ] ] which is nothing else but .
integration of from @xmath9 to @xmath118 gives @xmath130,\ ] ] i.e. , by integration by parts , @xmath131.\ ] ] by and using also @xmath114 , we have @xmath132 , \end{aligned}\ ] ] i.e. , holds , as desired .
* proof of theorem [ thm_kl_gen ] .
* let @xmath64 and @xmath60 be solutions of ( 1.4 ) and ( 1.5 ) , and introduce the notation @xmath133 then ( 1.4 ) and ( 1.5 ) take the form @xmath134,\label{de_e2withk}\\
e(0)=0\qquad\mbox{and}\qquad\lambda e'(1)=-kg'(1),\label{eq : boundary}\end{gathered}\ ] ] respectively .
these are , strictly speaking , not equations for the unknown function @xmath60 and scalar @xmath59 , since @xmath60 is hidden also in the coefficient @xmath135 .
however , it will prove convenient to consider temporarily as a second - order linear differential equation ( de ) for @xmath60 .
the general solution of the homogeneous part of is @xmath136,\ ] ] where @xmath137 , @xmath138 . to find the solution of the inhomogeneous equation , we use the method of variation of parameters , i.e. , we are looking for @xmath60 in the form @xmath139,\label{eq : varparam}\ ] ] with some twice continuously differentiable functions @xmath140\to{\mathbb{r}}$ ] . from this
, we obtain the system of equations @xmath141 for @xmath142 and @xmath143 . solving this and substituting the solutions into ,
we obtain @xmath144 where @xmath137 , @xmath138 . if we take into account the initial condition @xmath145 , we can write this in the form @xmath146 where @xmath138 . applying integration by parts twice in both integrals and taking into account
the condition @xmath114 , from this we obtain @xmath147 with the notation @xmath148 the first two terms can be contracted into one : @xmath149 now we substitute this into the definition of @xmath135 : @xmath150 where @xmath151 putting all these together , we obtain the equation @xmath152 now we want to substitute @xmath60 into the second equation of , therefore we calculate the derivative of : @xmath153.\end{aligned}\ ] ] substituting this into the second equation of , we get @xmath154 or , using again the notations , @xmath155 this , together with , yields the following homogeneous system of linear equations for the unknowns @xmath156 and @xmath135 : @xmath157 in what follows , we show that excluding two special cases , namely , @xmath7 , @xmath1 $ ] , and @xmath158 , @xmath1 $ ] , the function @xmath60 given in can be identically zero if and only if @xmath159 . to prove this
, it is enough to check that the functions @xmath160 , @xmath1 $ ] , and @xmath161 , \end{split } \end{aligned}\ ] ] are linearly independent . on the contrary ,
let us suppose that they are linearly dependent , i.e. , there exist constants @xmath162 such that @xmath163 and @xmath164 .
\end{split } \end{aligned}\ ] ] by differentiating twice , one can check that @xmath165 .
\end{split } \end{aligned}\ ] ] comparing and , we have @xmath166 , @xmath1 $ ] .
if @xmath167 , then @xmath168 , @xmath1 $ ] , which is a contradiction
. thus @xmath169 , and we have @xmath170 , @xmath1 $ ] . using that @xmath4 and @xmath171 , we get @xmath172 , @xmath1 $ ] or @xmath173 , @xmath1 $ ] , which cases were excluded .
this leads us to a contradiction .
hence , excluding the two special cases @xmath7 , @xmath1 $ ] , and @xmath158 , @xmath1 $ ] , we see that @xmath60 is not identically zero if and only if at least one of the two coefficients @xmath156 and @xmath135 is different from zero , i.e. , the system has a nontrivial solution for @xmath156 and @xmath135 .
this is equivalent to the condition that its determinant is zero , which in turn yields equation . if @xmath172 , @xmath1 $ ] , or @xmath173 , @xmath1 $ ] , then , by integration by parts , we have @xmath174 , \end{aligned}\ ] ] yielding that the function @xmath160 , @xmath1 $ ] , and the function in are linearly dependent
further , by , using @xmath175 , we have @xmath176,\\ & k=\int_0 ^ 1 g(t)e(t)\,{\mathrm{d}}t = \pm c_2 \int_0 ^ 1 t \sin\left(\frac{t}{\sqrt\lambda}\right){\mathrm{d}}t = \pm c_2 \sqrt{\lambda } \left ( -\cos\left(\frac{1}{\sqrt\lambda}\right ) + \sqrt{\lambda }
\sin\left(\frac{1}{\sqrt\lambda}\right ) \right ) .
\end{aligned}\ ] ] by the second equation of , we have @xmath177 ( which is in fact in the special cases @xmath172 , @xmath1 $ ] , and @xmath173 , @xmath1 $ ] ) .
this together with the above form of @xmath135 , yields @xmath178 , i.e. , @xmath179 , @xmath77 .
next we check that the equation @xmath178 is nothing else but the equation in the special cases @xmath172 , @xmath1 $ ] , and @xmath173 , @xmath1 $ ] . using integration by parts
, the constants defined in take the forms @xmath180 hence , using @xmath181 , the equation takes the form @xmath182 by some algebraic transformations , it is equivalent to @xmath183 . since @xmath64 , we have @xmath184 , yielding @xmath185 , @xmath77 , as desired .
all in all , for every possible @xmath21 , the equation holds .
it remains to study the form of the eigenfunctions . if @xmath186 , then from the second equation of we have @xmath187 where @xmath188 finally ,
if we substitute these expressions for @xmath135 and @xmath156 into , then we obtain with some appropriately chosen @xmath68 .
if @xmath189 , then we also show that @xmath190 , \qquad t\in[0,1 ] , \end{split } \end{aligned}\ ] ] is a normed eigenvector corresponding to the eigenvalue @xmath59 , where @xmath68 is chosen such that @xmath69 .
note that is a special case of . by proposition [ klgenthm ] ,
it is enough to verify that @xmath63 , @xmath1 $ ] , given in satisfies and .
first , note that , by integration by parts , one can calculate @xmath191 \cos\left(\frac 1{\sqrt\lambda}\right ) .
\end{aligned}\ ] ] further , @xmath192 , \qquad t\in[0,1 ] , \end{aligned}\ ] ] and @xmath193 , \qquad t\in[0,1 ] , \end{aligned}\ ] ] yielding that @xmath194 .
\end{aligned}\ ] ] hence , taking into account that @xmath195 , to verify it remains to check that @xmath196 \cos\left(\frac 1{\sqrt\lambda}\right)g''(t ) , \qquad t\in[0,1],\ ] ] which is equivalent to @xmath197 \cos\left(\frac 1{\sqrt\lambda}\right ) = 0 , \qquad t\in[0,1 ] .
\end{aligned}\ ] ] taking into account and that @xmath189 , we have @xmath198 \cos\left(\frac 1{\sqrt\lambda}\right ) = 0 , \end{aligned}\ ] ] yielding .
the boundary conditions hold as well .
indeed , the boundary condition @xmath145 is satisfied , since @xmath4 , and the boundary condition @xmath199 is equivalent to @xmath200 = 0 , \end{aligned}\ ] ] which is satisfied due to .
* an example for the assertion in remark [ rem_lambda0 ] .
* let @xmath3\to{\mathbb{r}}$ ] , @xmath201 , @xmath1 $ ] .
then @xmath114 , @xmath41 , @xmath171 , and @xmath9 is an eigenvalue of @xmath51 with @xmath202 as an eigenfunction corresponding to @xmath9 , which is in accordance with nazarov ( * ? ? ?
* corollary 2 ) . indeed , @xmath203
\end{aligned}\ ] ] * first proof of corollary [ klcor1 ] . * we will apply theorem [ thm_kl_gen ] with the function @xmath3\to{\mathbb{r}}$ ] , @xmath79 , @xmath1 $ ] .
first , we check that @xmath204 can not be an eigenvalue . on the contrary ,
let us suppose that @xmath205 is an eigenvalue .
then , using @xmath206 , would imply that @xmath207 which is equivalent to @xmath208 .
calculating the integral , this leads us to @xmath209 , being a contradiction . using @xmath210 and the addition formulas for cosine and sine
, the constants defined in take the forms @xmath211{\mathrm{d}}u \\ & \phantom{a_g(\lambda ) } = \frac{\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \left(1 + \cos\left(\frac{1}{\sqrt{\lambda}}\right)\right ) , \qquad \lambda>0,\\ & b_g(\lambda)=\frac{\sqrt{2}}{\pi } \int_0 ^ 1 \sin(\pi u ) \sin\left(\frac{u}{\sqrt{\lambda}}\right){\mathrm{d}}u = \frac{1}{\sqrt{2}\pi}\int_0 ^ 1\bigg[\cos\left ( \left(\pi-\frac{1}{\sqrt{\lambda}}\right ) u \right ) - \cos\left ( \left(\pi+\frac{1}{\sqrt{\lambda}}\right ) u \right ) \bigg]{\mathrm{d}}u \\ & \phantom{b_g(\lambda ) } = \frac{\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \sin\left(\frac{1}{\sqrt{\lambda}}\right ) , \qquad \lambda>0 , \end{aligned}\ ] ] and @xmath212\\ & \phantom{c_g(\lambda)\ ; } = \frac{1}{2\pi^2 \left(\pi^2-\frac{1}{\lambda}\right ) } \left[\frac{1}{\sqrt{\lambda } } + \frac{\pi^2}{\pi^2-\frac{1}{\lambda } } \sin\left ( \frac{2}{\sqrt{\lambda } } \right ) \right ] , \qquad \lambda>0 .
\end{aligned}\ ] ] hence , using @xmath213 , the equation takes the form @xmath214 which , by some algebraic transformations , yields .
further , the normed eigenfunctions take the form @xmath215\\ & = c\bigg [ \frac{\sqrt{2}}{\pi } \sqrt{\lambda } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin(\pi t ) + \frac{\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \left ( 1 + \cos\left ( \frac{1}{\sqrt{\lambda } } \right)\right ) \sin\left ( \frac{t}{\sqrt{\lambda } } \right ) \\ & \phantom{= c\bigg [ } + \frac{1}{\sqrt{2}\pi}\cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \cos\left ( \frac{t}{\sqrt{\lambda } } \right ) \left ( \frac { \sin\left ( \left ( \pi - \frac{1}{\sqrt{\lambda}}\right)t\right ) } { \pi-\frac{1}{\sqrt{\lambda } } } - \frac { \sin\left ( \left ( \pi + \frac{1}{\sqrt{\lambda}}\right)t\right ) } { \pi+\frac{1}{\sqrt{\lambda } } } \right)\\ & \phantom{= c\bigg [ } - \frac{1}{\sqrt{2}\pi}\cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin\left ( \frac{t}{\sqrt{\lambda } } \right ) \left ( - \frac { \cos\left ( \left ( \pi + \frac{1}{\sqrt{\lambda}}\right)t\right ) } { \pi+\frac{1}{\sqrt{\lambda } } } - \frac { \cos\left ( \left ( \pi - \frac{1}{\sqrt{\lambda}}\right)t\right ) } { \pi-\frac{1}{\sqrt{\lambda } } } + \frac{1}{\pi+\frac{1}{\sqrt{\lambda } } } + \frac{1}{\pi-\frac{1}{\sqrt{\lambda } } } \right ) \bigg]\\ & = c\bigg [ \frac{\sqrt{2}}{\pi } \sqrt{\lambda } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin(\pi t ) + \frac{\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \sin\left ( \frac{t}{\sqrt{\lambda } } \right)\\ & \phantom{= c\bigg [ } + \frac{1}{\sqrt{2}\pi \left(\pi - \frac{1}{\sqrt{\lambda}}\right ) } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \left ( \sin\left ( \left ( \pi - \frac{1}{\sqrt{\lambda}}\right)t\right ) \cos\left ( \frac{t}{\sqrt{\lambda } } \right ) + \cos\left ( \left ( \pi - \frac{1}{\sqrt{\lambda}}\right)t\right ) \sin\left ( \frac{t}{\sqrt{\lambda } } \right ) \right)\\ & \phantom{= c\bigg [ } + \frac{1}{\sqrt{2}\pi \left(\pi + \frac{1}{\sqrt{\lambda}}\right ) } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \left ( \sin\left ( \frac{t}{\sqrt{\lambda } } \right ) \cos\left ( \left ( \pi + \frac{1}{\sqrt{\lambda}}\right)t\right ) - \cos\left ( \frac{t}{\sqrt{\lambda } } \right ) \sin\left ( \left ( \pi + \frac{1}{\sqrt{\lambda}}\right)t\right ) \right)\bigg]\\ & = c\bigg [ \frac{\sqrt{2}}{\pi}\sqrt{\lambda } \cos\left ( \frac{1}{\sqrt{\lambda } }
\right ) \sin(\pi t ) + \frac{\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \sin\left ( \frac{t}{\sqrt{\lambda } } \right)\\ & \phantom{= c\bigg [ } + \frac{1}{\sqrt{2}\pi \left(\pi - \frac{1}{\sqrt{\lambda}}\right ) } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin(\pi t ) + \frac{1}{\sqrt{2}\pi \left(\pi + \frac{1}{\sqrt{\lambda}}\right ) } \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin(-\pi t ) \bigg ] \\ & = \frac{c\sqrt{2}}{\pi^2 - \frac{1}{\lambda } } \left ( \sin\left ( \frac{t}{\sqrt{\lambda } } \right ) + \sqrt{\lambda}\pi \cos\left ( \frac{1}{\sqrt{\lambda } } \right ) \sin(\pi t ) \right ) , \end{aligned}\ ] ] where @xmath68 is such that @xmath69 .
hence @xmath216 merging the factor @xmath217 into @xmath70 , taking into account and that @xmath218 this yields .
* second proof of corollary [ klcor1 ] .
* in the special case @xmath79 , @xmath1 $ ] , the de and the boundary conditions take the form @xmath219 , \end{aligned}\ ] ] and @xmath220 respectively . with the special choice @xmath121 ,
using @xmath122 and @xmath64 , we have @xmath221 .
the de is a second order linear inhomogeneous de of the type @xmath222 , @xmath1 $ ] , where @xmath223 ( for more details , see the proof of theorem [ thm_kl_gen ] ) . by the method of undetermined coefficients , its general solution takes the form @xmath224 , \end{aligned}\ ] ] where @xmath225 .
hence @xmath226,\\ \label{help4 } e''(t ) & = - \frac{a}{\lambda}\sin\left(\frac{t}{\sqrt{\lambda}}\right ) - \frac{b}{\lambda } \cos\left(\frac{t}{\sqrt{\lambda}}\right ) - c\pi^2\sin(\pi t ) - d\pi^2\cos(\pi t ) , \qquad t\in[0,1 ] .
\end{aligned}\ ] ] then the de takes the form @xmath227 . \end{aligned}\ ] ] since @xmath122 , by , we have @xmath228 . since @xmath221 , by , we have @xmath229 , and then , since @xmath228 , we get @xmath230 or @xmath204 .
next , we check that @xmath204 can not be an eigenvalue . on the contrary ,
let us suppose that @xmath204 is an eigenvalue . then , since @xmath228 , by , we have @xmath231,\ ] ] where @xmath232 .
further , by , @xmath233 yielding @xmath234 , @xmath1 $ ] , which leads us to a contradiction . since @xmath235
, then we have @xmath230 . by and
together with @xmath236 , we have @xmath237 , @xmath1 $ ] , and @xmath238 .
hence @xmath239 then @xmath240 by and @xmath241 , we have @xmath242 and hence @xmath243 note that the eigenfunction @xmath60 is identically zero if and only if @xmath244 . indeed ,
the functions @xmath245 , @xmath1 $ ] , and @xmath246 , @xmath1 $ ] , are linearly independent provided that @xmath235 , as we check below . on the contrary let us suppose that they are linearly dependent , i.e. , there exist constants @xmath232 such that @xmath247 and @xmath248 , @xmath1 $ ] . without loss of generality
, one can assume that @xmath249 .
then @xmath250 , @xmath1 $ ] , and , by differentiating twice , we have @xmath251 , @xmath1 $ ]
. hence @xmath252 , @xmath1 $ ] .
since @xmath235 , we have @xmath253 yielding @xmath254 , @xmath1 $ ] , which is a contradiction . the system and
has a nontrivial solution @xmath255 if and only if its determinant is zero , which yields @xmath256 which is equivalent to , since @xmath257 finally , we check . by , using @xmath230 and @xmath235 , we have @xmath258 and consequently @xmath259 which yields @xmath260 , \qquad t\in[0,1 ] . \end{aligned}\ ] ] using @xmath261 , one can finish the proof as in the first proof of corollary [ klcor1 ] .
* proof of corollary [ klcor2 ] .
* one can apply theorem [ thm_kl_gen ] with the function @xmath3\to{\mathbb{r}}$ ] , @xmath100 , @xmath1 $ ] . in the proof of theorem [ thm_kl_gen ]
we have already checked that the equation @xmath178 is nothing else but the equation in the special case @xmath172 , @xmath1 $ ] .
further , taking into account @xmath185 , @xmath77 , by partial integration , we obtain that the normed eigenfunctions take the form @xmath262 \\ & = c(-1)^k\bigg [ \frac{1}{k\pi}t + \frac{(-1)^k - 1}{(k\pi)^2}\sin(k\pi t ) + \cos(k\pi t ) \int_0^t u \sin(k\pi
u)\,{\mathrm{d}}u - \sin(k\pi t ) \int_0^t u \cos(k\pi
u)\,{\mathrm{d}}u \bigg]\\ & = \frac{c}{(k\pi)^2}\sin(k\pi t ) , \qquad t\in[0,1 ] , \end{aligned}\ ] ] where @xmath68 is such that @xmath69 . since @xmath263 , @xmath77 , we have @xmath264 , yielding @xmath265 , @xmath1 $ ] , i.e. , we have .
* appendix *
in this appendix we compare the gauss process given in with the gauss process given in ( 1.3 ) in nazarov @xcite , and then we also compare our theorem [ thm_kl_gen ] with the results in section 3 in nazarov @xcite for the kl expansions of the gauss processes in question .
let @xmath266}$ ] be a zero - mean gauss process , and suppose that its covariance function @xmath267 is continuous .
let @xmath268\to{\mathbb{r}}$ ] be a measurable function such that @xmath269 .
introduce the function @xmath270\to{\mathbb{r}}$ ] , @xmath271 , @xmath1 $ ] .
further , suppose that @xmath272 is not the identically zero function , and let @xmath273 for all @xmath274 , introduce the stochastic process @xmath275 .
\end{aligned}\ ] ] due to nazarov ( * ? ? ?
* formula ( 1.4 ) ) , @xmath276}$ ] is a zero - mean gauss process with covariance function @xmath277 .
\end{aligned}\ ] ] nazarov ( * ? ? ?
* section 3 ) gave a procedure for finding the kl expansion of @xmath276}$ ] supposing that we know the kl expansion of @xmath266}$ ] .
he also provided several examples ( specifying @xmath278 and @xmath279 ) , where he made the kl expansion of @xmath276}$ ] as explicit as possible .
the process @xmath280 can be considered as a one - dimensional linear perturbation of the gauss process @xmath278 .
now we consider two special cases of the above construction .
let @xmath266}$ ] be a standard wiener process , @xmath268\to{\mathbb{r}}$ ] be a twice continuously differentiable function such that @xmath281 $ ] and @xmath282 , and let @xmath283 .
then @xmath284 , @xmath285 $ ] , @xmath270\to{\mathbb{r}}$ ] , @xmath286,\ ] ] and @xmath287 .
further , @xmath288 , @xmath1 $ ] , and @xmath289 , @xmath1 $ ] , satisfying @xmath290 , @xmath291 , and @xmath292 . hence , taking into account proposition [ pro1 ] and , choosing @xmath21 as the function @xmath272 , the gauss process @xmath31}$ ] given in coincides in law with the gauss process @xmath276}$ ] given in .
let @xmath266}$ ] be a standard wiener process , and @xmath3\to{\mathbb{r}}$ ] be a twice continuously differentiable function with @xmath4 , @xmath41 , and @xmath293 .
let us define the gauss process @xmath276}$ ] given in with @xmath294 and @xmath295 . since @xmath296,\ ] ] and @xmath297,\ ] ] we have @xmath298 , @xmath299 and @xmath287 .
note that it is a so - called critical case , since @xmath300 , see nazarov ( * ? ? ?
* corollary 2 ) .
hence the gauss process @xmath276}$ ] given in coincides in law with the gauss process @xmath31}$ ] given in . based on the above discussion ,
if @xmath3\to{\mathbb{r}}$ ] is a twice continuously differentiable function with @xmath4 , @xmath41 , and @xmath293 , then the gauss process @xmath31}$ ] given in coincides in law with one of the gauss processes introduced in nazarov ( * ? ? ?
* formula ( 1.3 ) ) .
we would like to thank endre igli for providing us with the formal motivation of the process defined in presented in the introduction .
we are grateful to yakov nikitin for drawing our attention to the paper @xcite of nazarov , and sending us some recent papers on explicit karhunen
love expansions as well .
special gratitude to alexander nazarov for explaining us in detail how the setup in his paper @xcite can be specialized to our case , and for pointing out a mistake in an earlier version of the paper related to the fact that zero can be an eigenvalue of the integral operator under consideration . | we derive a karhunen love expansion of the gauss process @xmath0 , @xmath1 $ ] , where @xmath2}$ ] is a standard wiener process and @xmath3\to { \mathbb{r}}$ ] is a twice continuously differentiable function with @xmath4 and @xmath5 .
we formulate two special cases with the function @xmath6 , @xmath1 $ ] , and @xmath7 , @xmath1 $ ] , respectively .
the latter one corresponds to the wiener bridge over @xmath8 $ ] from @xmath9 to @xmath9 .
* karhunen love expansion for a generalization of wiener bridge * + mtys @xmath10 and rezs l. @xmath11 0.1 cm faculty of informatics , university of debrecen , pf .
400 , h4002 debrecen , hungary .
* institute of mathematics , university of debrecen , pf .
400 , h4002 debrecen , hungary .
e mails : [email protected] ( m. barczy ) , [email protected] ( r. l. lovas ) . @xmath12 corresponding author . |
entanglement @xcite , a characteristic feature of quantum mechanics , plays a key role in many quantum information tasks @xcite . in order to implement those tasks , it is important to devise schemes for the creation of entangled states and to observe their dynamical characteristics in the presence of enviromental effects .
cavity quantum electrodynamics ( qed ) systems , the cleanest open quantum systems known so far , are interesting from the point of view of transfering quantum information from matter to radiation , from flying to static systems , and viceversa
. moreover , cavity qed setups @xcite have been used in many fundamental experiments in quantum mechanics , such as the observation of decoherence @xcite , the generation of atomic epr pairs @xcite and the observation of quantum jumps @xcite .
this paper studies the entanglement dynamics under dissipation in a tripartite system , simple enough to allow for a quasi - analytical solution but rich enough to display a variety of dynamical characteristics , which can be implemented experimentally with present day technology .
figure [ fig : scheme ] shows the system , which is composed by several cavities crossed by a two - level atom .
similar proposals with multiple cavities , attempting to approach the heisenberg limit for the estimation of an atomic phase shift @xcite or producing field - field entangled states @xcite , can be found in the literature . in our system , the first and the last cavities are high quality cavities and the second one is a ramsey zone .
we calculate the degree of entanglement of the three pairs of subsystems , which are shown to behave as qubits for appropriate initial conditions , and discuss the role of the initial photon number and dissipation rate in the entanglement properties as measured by concurrence .
this paper has four sections : the first , section [ sec : model ] , establishes the mathematical model that describes the proposed setup ; the second ( section [ sec : dynamics ] ) presents the evolution superoperator for the model ; in the third section the state of the system when the atom traverses the first cavity and the atom - field concurrence are given ; finally , section [ sec : cavity2 ] describes the behavior of pairwise concurrence obtained numerically by using typical experimental values .
we consider an apparatus composed of three cavities crossed by a two - level atom , as depicted in fig .
[ fig : scheme ] .
as it is clear from the picture , the atom undergoes several dynamical processes : in the first step the atom interacts with a high quality dissipative cavity containing a coherent state , when the atom leaves this first cavity it will accumulate a phase due to free motion until the ramsey zone is reached , where its state is rotated and , immediately thereafter a second free flight , the atom interacts with the final high - q dissipative cavity prepared in a different coherent state .
although the cavities contain , in principle , infinite modes of the electromagnetic field , we consider only one mode per cavity field , almost resonant with the relevant atomic transition .
it is important to realize that we are assuming that the center - of - mass coordinate of the atom is treated as if it were a classical variable , which is possible when the velocity of the atom is large enough not to be affected by the changes of the potential energy caused by the interaction with the fields .
( color online ) scheme of the proposed experimental setup.,scaledwidth=48.0% ] we assume that the whole setup is cooled to very low temperatures ( of the order of 1k in this case ) which allows us to model cavity field decay as interaction with a zero - temperature bath . we suppose that the atom - field detuning , in each cavity , is large enough to be in the dispersive regime .
we can describe the electromagnetic field of the ramsey zone as a classical one , in its interaction with the atom @xcite .
taking into account these considerations we can write the equation of motion of @xmath0 , the atom - fields state , as @xmath1+{\mathcal{l}}_d({\hat{\rho}}(t)),\ ] ] where @xmath2 , which models the dissipative effects , is written as @xmath3 the rate at which photons in the i - th cavity are lost , @xmath4 , is proportional to the cavity frequency , @xmath5 , and inversely proportional to the quality factor of the cavity , @xmath6 .
since we assume that the dissipative processes of the atom are much slower than those of the field , we ignore these processes in the description of the dynamics of the whole system .
the hamiltonian @xmath7 comprises two parts , which we designate by @xmath8 and @xmath9 .
the first part of the hamiltonian , the free hamiltonian @xmath8 , is given by @xmath10 where @xmath11 is the transition frequency between the relevant states of the atom , @xmath12 ( ground ) and @xmath13 ( excited ) , and @xmath14 is the operator @xmath15 .
the creation @xmath16 and annihilation @xmath17 operators of the cavity fields , satisfy the usual bosonic commutation relations @xmath18=\delta_{ij}$ ] and @xmath19= 0 = [ a_i^\dag , a_j^\dag]$ ] .
the second part of the hamiltonian is the interaction hamiltonian @xmath20 where the function @xmath21 is equal to @xmath22 if @xmath23 belongs to the interval @xmath24 $ ] , and zero otherwise .
hence , the atom interacts with the first ( second ) cavity field in the interval of time @xmath25 $ ] ( @xmath26 $ ] ) and with the ramsey - zone field in the interval @xmath27 $ ] .
the coupling constant between the atom and the i - th cavity field is @xmath28 , where the vacuum rabi frequency @xmath29 is proportional to the product of the atomic dipole with the electric field having the root mean square amplitude of the vacuum in the cavity .
the probability of detecting @xmath30 photons in the i - th cavity , @xmath31 , is usually employed to specify the conditions under which the dispersive approximation is valid .
we define @xmath32 , the relevant number of excitations in the i - th cavity , as the maximum number @xmath30 , for which the probability @xmath33 is still important .
if the detuning of the i - th cavity field , @xmath34 , satisfies the inequality @xmath35 , we are in a region of parameters where the dispersive approximation is valid , and the atom - field interaction can be described by the hamiltonian @xcite @xmath36 with @xmath37 .
the ramsey zone , which rotates the internal state of the atom , is modeled by the hamiltonian @xmath38 where the hamiltonian @xmath39 depends on @xmath40 the vacuum rabi frequency of the ramsey zone and @xmath41 , the squared root of the mean number of photons of the coherent state sustained in the ramsey zone .
the operator @xmath42 is given by @xmath43 .
in order to solve the dynamics of the whole system in the interval @xmath44 $ ] it is convenient to find solutions in the intervals @xmath45 $ ] , for @xmath46 .
for example , in the first interval we can write @xmath47 where the evolution superoperator @xmath48 acts on @xmath49 , the initial state of the total system .
the function @xmath50 stresses that the last equation is valid only in the interval @xmath25 $ ] .
the state of the system at the end of one interval is the initial state for the next interval .
thus , we can write , in general @xmath51 employing lie algebraic techniques @xcite , as shown in appendix [ sec : evolution ] , we cast the evolution superoperator in the @xmath52 $ ] interval as @xmath53 where @xmath54 is the unitary operator which solves the hamiltonian dynamics @xmath55 in that time interval , and @xmath56 is a superoperator given below .
we employ the dot - convention which establishes how ( super)operators act on other ( super)operators : the ( super)operator to the right of the superoperator with dots will occupy the place of the dots .
for example , @xmath57 the dissipative part of the evolution superoperator reads @xmath58 where the operator @xmath59 ( @xmath60 ; @xmath61 ) @xmath62 depends on the parameters @xmath63 , @xmath64 .
the field operators @xmath65 and @xmath66 are given explicitly by @xmath67 the unitary contribution to the evolution superoperator can be cast as @xmath68/\hbar } \tilde{u}_k(t , t_k)\end{aligned}\ ] ] where @xmath69 is the hamiltonian evolution operator in the interaction picture .
hence , @xmath70 is the identity in the intervals of free flight ; @xmath71/\hbar } , \quad k=0,4,\end{aligned}\ ] ] while the atom crosses the high - q cavities and @xmath72,\end{aligned}\ ] ] while the atom crosses the ramsey zone , assumed to be resonant with the atomic transition .
the initial state of the two - cavities - atom system that we consider , which is relatively easy to be prepared with current experimental techniques , is a separable one : the atom in a superposition of its two relevant states and the fields in coherent states , @xmath73 , where @xmath74 and @xmath75 ( @xmath76 ) is a coherent state for the first ( second ) cavity . while the atom crosses the first cavity , the state of the system is @xmath77 where @xmath78 and @xmath79,\\ \nonumber f_x(t)= \frac{{\gamma}_1}{{\gamma}_1+i{\omega}_1}(1-e^{-2({\gamma}_1+i{\omega}_1)t})-(1-e^{-2{\gamma}_1t}).\end{gathered}\ ] ] we have chosen @xmath80 since the atom - first cavity field density operator belongs to a @xmath81 subspace of the total available hilbert space , it corresponds to a couple of qubits , whose entanglement can be quantified by means of wootters concurrence @xcite . in our case
@xmath82 clearly , one of the factors on which concurrence depends , is a function of the nonorthogonality of the states @xmath83 and @xmath84 , the more different the states the larger the concurrence . although the other presents a more complex dependence , its main contribution is of the form @xmath85 that is , it principally depends on the initial number of photons and the product of the decoherence rate and the time spent in the cavity . in fig .
[ fig : a1 ] we show the time evolution of the concurrence between the atom and the first cavity for different values of initial mean number of photons and different dissipation rates , the time interval was chosen between @xmath86 and @xmath87 @xmath88 .
as the initial average photon number is increased , the maximum entanglement lasts for longer times in a lossless cavity .
however , at the same time , for a lossy cavity , at larger values of @xmath89 , atom and cavity field separate faster . 0.5 , ( b ) @xmath901 , ( c ) @xmath902 and for different values of the dissipation rate . in the second column
a blow - up of the first two - oscillations is shown .
for @xmath86 and @xmath87 @xmath88.,scaledwidth=48.0% ] a graphic way to see this is by looking at the phase space representation of the field @xcite .
the physics of this part of the process seems easy to understand : without dissipation , the key element for entanglement , the interaction , is proportional to the field intensity .
however , for coherent states , the role of the interaction is to split the field into a component correlated to the excited atomic state whose label rotates at angular velocity @xmath91 and a component correlated to the ground atomic state whose label rotates at angular velocity @xmath92 ( see eq .
( [ eq : alphas ] ) ) .
it is easier to concentrate on the angular velocity differences , and to plot the labels in a phase space picture ( fig .
[ fig : phasespace ] ) .
it is clear that every @xmath93 seconds both labels coincide , hence the state becomes separable .
this process remains unaffected in the presence of decoherence . from eq .
( [ eq : conc ] ) it is clear that entanglement decreases as the overlap increases , or , in other words , entanglement decreases with the phase space distance between the field states correlated with the excited and ground states , distance which is given by the length of the vertical line that join those states . for larger mean number of photons ,
the phase space distance is large most of the time , except near the separation times ; for small mean number of photons this distance never grows appreciable , neither does the entanglement .
( color online ) phase - space evolution of @xmath94 and @xmath94 in a rotating frame with frequency @xmath95 .
the initial mean number of photons is @xmath96 .
dashed line shows the dissipationless case and solid lines represents the dissipation one with @xmath97 @xmath98 and dispersive frequency @xmath99 @xmath98 .
vertical distance of the trajectories at same time is proportional to entanglement .
the line @xmath100(@xmath101 ) corresponds to a time of larger ( smaller ) entanglement.,scaledwidth=48.0% ] in the presence of dissipation the characteristic time for decoherence is inversely proportional to the average photon number .
we have already pointed out that a factor which decays with the characteristic time of decoherence enters in the expression for concurrence . in fig . [
fig : phasespace ] one observes that the part of the field entangled with @xmath13 ( @xmath12 ) rotates clockwise ( counterclockwise ) and its radius shrinks with dissipation .
the assymetry of the entanglement curves shown in the fig .
[ fig : a1 ] can be understood in terms of the phase space distance , as was discussed above , and marked on fig .
[ fig : phasespace ] with vertical lines . as the atom crosses the second cavity , the expressions for the density operator calculated analytically become much more cumbersome and are omitted .
therefore , concurrence is calculated only numerically . from here on we make a more numerical approach .
taking into account typical values of current experiments @xcite we consider the following fixed parameters : the angular dispersive frequencies @xmath102 @xmath98 , the atomic transition frequency and the ramsey zone frequency @xmath103 @xmath98 , the detuning @xmath104 @xmath98 and rabi frequencies @xmath105 @xmath98 .
the flight times through the cavities are also fixed : the atom spents 30 @xmath88 in each cavity and 10 @xmath88 between cavities .
the dissipation rates for the cavities @xmath106 and @xmath107 are varied according to @xmath108 and @xmath109 , where @xmath110 . in order not to break the dispersive approximation
, we can not increase too much the mean number of photons . to estimate the validity of the dispersive approximation we use @xmath111 .
for the values of the parameters chosen above , we are constrained to coherent states satisfying @xmath112 we thus choose , for the mean number of photons , the values @xmath113 .
we begin by considering the first cavity to be ideal and the second to range between ideal ( q=0 ) and strongly dissipative ( q=1 ) . in fig .
[ fig : ideal1 ] we plot the concurrences for all three pairs of subsystems . for ideal cavities , and initial coherent states with @xmath114 , the final state
contains quantum correlations between all pair of subsystems . if the second cavity is strongly dissipative only the atom and the first field show relevant quantum correlations .
( color online ) atom - field1 ( a ) atom - field2 ( b ) field1-field2 ( c ) concurrence for ideal first cavity and different dissipation rates for the second one . ] if we consider the opposite case , that of a ideal second cavity , and first cavity ranging between ideal and strongly dissipative , entanglement behavior is complementary to the previous case ( fig . [
fig : ideal2 ] ) : only the atom - second cavity field are important for a strongly dissipative first cavity .
( color online ) atom - field1 ( a ) atom - field2 ( b ) field1-field2 ( c ) concurrence for ideal second cavity and different dissipation rates for the first one . ]
we conclude that the most interesting physical effects from the point of view of multipartite entanglement happen for decoherence rates of about one order of magnitude smaller than the dispersive frequency .
this conclusion holds even if we change the mean number of photons in the cavities , as shown in fig .
[ fig : transf ] .
all these results shows that the local independent interactions between the atom and the cavity fields has as consequence the generation of non - local field - field entanglement .
( color online ) field1-field2 concurrence .
maximum values are obtained for ( nearly ) ideal cavities and minimum values for strongly dissipative cavities.,scaledwidth=48.0% ] as noticed before , increasing the mean photon number can lead to smaller atom - cavity ( quantum ) correlations ( see fig .
[ fig : allb ] ) .
in fact , atom - first field concurrence decreases as the initial mean number of photons in second cavity increases from one to four .
also , if we begin with coherent states with mean photon numbers larger than one , even for not so large values of the decoherence rate , we observe sudden death of the atom - first field entanglement ( see fig .
[ fig : allb ] , first row ) . as shown in figures [ fig : ideal1]-[fig : ideal2 ] and [ fig : allb ] , the entanglement between all pairs of subsystems can not increase simultaneously , suggesting some kind of _ entanglement conservation law _ , related to the coffman - kundu - wootters ( ckw ) monogamy inequality @xcite .
( color online ) atom - field1 , atom - field2 and field1-field2 concurrence for an ideal first cavity , @xmath115 and different values of @xmath116 and @xmath117 . ]
it is interesting to notice that as the initial mean number of photons increase so does the field - field entanglement , for ( nearly ) ideal cavities , and short times .
as time goes on , this rule does not hold anymore , and there seems to be an optimal value of the mean number of photons to produce the higher field - field entanglement for a given interaction time ( see fig .
[ fig : allb ] , for g=0 and q=0 , 0.5 ) . in summary ,
we have considered a realistic cavity qed setup , assumed to be operated on the dispersive regime , and have solved analytically its dynamics , in the sense of finding the evolution superoperator .
we have shown that for typical experimental values of the cavity dissipation rates , smaller than the dispersive frequencies , and times of the order of 100 @xmath88 , not only the final state generally exhibits atom - field and field - field entanglement , but that the effect of the dissipation is not trivial , in the sense that the pairwise concurrences are not monotonic functions of the dissipation rates .
the authors thanks h. vinck , m. c. nemes and j. g. gonalves de oliveira junior , for very useful discussions .
k.m.f.r . acknowledge partial support to dib - unal .
it is easy to show that the evolution superoperator @xmath118 in the interval @xmath52 $ ] , @xmath119 , given by @xmath120+\frac{\delta_{k,2}}{i\hbar}[h_r(\tau),\cdot ] \right),\end{aligned}\ ] ] where @xmath121 is the time - ordering operator , can be written as due to the following commutation relations @xmath122 & = 0 = [ \cdot h_0 , \mathcal{l}_d],\\ [ h_r ( t ) \cdot , \mathcal{l}_d ] & = 0 = [ \cdot h_r ( t ) , \mathcal{l}_d].\end{aligned}\ ] ] in the intervals @xmath25 $ ] and @xmath26 $ ] the evolution superoperator @xmath123 can be written as @xmath124 with @xmath125,\\ \mathcal{l}_{n } = \sum_i \left ( 2\gamma_i \mathcal{j}_i -\frac{2\gamma_i+i \omega_{ik } \left({\sigma_z}\cdot-\cdot { \sigma_z}\right)}{2 } \left ( \mathcal{m}_i+\mathcal{p}_i \right ) \right),\end{aligned}\ ] ] where @xmath63 , @xmath64 and the operators @xmath126 are defined in .
taking into account that the atomic operator commute with the field operators , we write @xmath127 where the commutation relations @xmath128 = -2\mathcal{j}_j\delta_{ij } , \quad i , j=1,2,\end{aligned}\ ] ] and disentangling operators techniques @xcite were used .
it is clear that the coefficients @xmath129 contain a term proportional to @xmath130 and one proportional to @xmath131 , which can be separated .
we can write therefore @xmath132 grouping the first term of with the terms above for zero dissipation ( @xmath133 ) we recover the hamiltonian dynamics . on the other hand the last line of corresponds to the purely dissipative dynamics ( with zero hamiltonian ) . after employing the explicit expressions for @xmath134
, the evolution superoperator can be written as .
h. walther , b. t. h. varcoe , b .-
englert and t. becker .
_ cavity quantum electrodynamics _ , rep .
69 , 1325 ( 2006 ) m. brune , e. hagley , j. dreyer , x. matre , a. maali , c. wunderlich , j.m .
raimond and s. haroche .
_ observing the progressive decoherence of the meter in a quantum measurement _ , phys .
77 , 4887 ( 1996 ) .
s. gleyzes , s. kuhr , c. guerlin , j. bernu , s. delglise , u. busk hoff , m. brune , j .-
raimond and s. haroche .
_ quantum jumps of light recording the birth and death of a photon in a cavity _ , nature 446 , 297 ( 2007 ) .
j. i. kim , k. m. fonseca romero , a. m. horiguti , l. davidovich , m. c. nemes and a. f. r. de toledo piza .
_ classical behavior with small quantum numbers : the physics of ramsey interferometry of rydberg atoms _ , phys .
82 , 4737 ( 1999 ) . | we propose an experimental setup , feasible with present day technology , involving two high - quality - factor cavities , one ramsey zone and a two - level atom which interacts with them .
the dynamics in the cavities is modeled by a dissipative dispersive jaynes - cummings model .
analytical expresions for the evolution superoperator in each part of the proposed setup and for the concurrence between the atom and the first cavity are given , and the pairwise entanglement of the subsystems is calculated numerically .
we consider initial separable states given by the tensor product of a superposition of the two atomic relevant states and coherent states for the fields .
we show that for realistic model parameters ( including dissipation ) , when the dissipation rates are smaller than the dispersive frequencies ( the ratio between the squared vacuum rabi frequencies and the detunings ) , the final state of the cavity fields remains entangled . |
let @xmath4 , @xmath5 , be contractions on @xmath6 , i.e. , @xmath7 for all @xmath8 with @xmath9 .
here @xmath10 stands for the usual euclidean norm , but this norm may be replaced by any other norm on @xmath6 . it is well known @xcite that there exists a unique non - empty compact set @xmath11 such that @xmath12 let @xmath13 denote the set of @xmath0 matrices with real entries .
we will assume that @xmath14 where @xmath15 , called _ digits _ , and @xmath16 then @xmath17 is called a _ self - affine set _ or a _ self - affine fractal _ , and can be viewed as the invariant set or the attractor of the ( affine ) _ iterated function system _ ( ifs ) @xmath18 ( in the terminology of dynamical systems ) .
let @xmath19 be the set of @xmath0 integer matrices .
further , if @xmath20 and @xmath21 , it is called an
_ integral self - affine set _ and we will primarily consider such sets in this paper . if , additionally , @xmath22 and the integral self - affine set @xmath17 has positive lebesgue measure , then @xmath17 is called an _
integral self - affine tile_. we sometimes write @xmath23 for @xmath17 to stress the dependence on @xmath24 and @xmath25 . for such tiles ,
the positivity of the lebesgue measure is equivalent to having nonempty interior @xcite .
there is a demand to develop analysis on fractal spaces , in order to deal with physical phenomena like heat and electricity flow in disordered media , vibrations of fractal materials and turbulence in fluids . without a better understanding of the topology of fractals
, this seems to be a difficult task .
there is a growing literature on the formalization and representation of topological questions ; see @xcite for a survey of the field .
one of the interesting aspects of the self - affine sets is the _ connectedness _ , which roughly means the attractor can not be written as a disjoint union of two pieces .
this property is important in computer vision and remote sensing @xcite .
we mention that connected self - affine fractals are curves ; thus , they are sometimes referred to as self - affine curves @xcite .
there is some motivation for studying connected self - affine tiles because they are related to number systems , wavelets , torus maps .
recently , there have been intensive investigations on the topic by kirat and lau @xcite , akiyama and thuswaldner @xcite , ngai and tang @xcite and luo et al .
@xcite . in this note
, we consider planar integral self - affine fractals obtained from @xmath26 integer matrices with reducible characteristic polynomials , and report our findings on their connectedness .
however , our considerations can be generalized to higher dimensions . as for the organization of the paper , in section [ simple ]
, we deal with special cases and state some simple , but unconventional techniques to check the connectedness . in section [ general ] ,
we study the neighbor sets of self - affine fractals .
usually , connectedness criteria were given by using a `` graph '' with vertices in @xmath25 @xcite . in this section
, we present graph - independent techniques to check the connectedness or disconnectedness . throughout the paper
, @xmath27 is a contraction .
let @xmath28 denote the number of elements in @xmath25 .
we first recall a known result .
[ collinear1 ] @xcite suppose @xmath29 $ ] with @xmath30 , and @xmath31 with @xmath32 then @xmath33 is a connected tile if and only if , up to a translation , @xmath34 for some @xmath35 . as one may notice @xmath36 and @xmath25 are not arbitrary in proposition [ collinear1 ] since @xmath30 and @xmath37 .
by using the approach in @xcite , we can remove such restrictions .
for that purpose , we consider the convex hull of @xmath17 and denote it by @xmath38 .
also let @xmath39 then we have the following .
[ collinear2 ] let @xmath40 with @xmath41 and @xmath42 , where @xmath43 and @xmath44 is the identity matrix .
then @xmath33 is connected if and only if @xmath45 .
[ fig1 ] a digit set @xmath25 as in proposition [ collinear2 ] is called a _ collinear
_ digit set .
it is easy to check the condition @xmath45 in the proposition because @xmath38 is a closed interval .
also , note that if @xmath46 , then @xmath33 is connected for any digit set .
a famous example of this type is the sierpi@xmath47ski tile ( see figure [ fig1 ] ) , for which @xmath48 and @xmath49 , d_2=\left [ \begin{array}{cc } 1 \\ 0 \end{array } \right ] , d_3=\left [ \begin{array}{cc } 0 \\ 1 \end{array } \right ] , d_{4}=\left [ \begin{array}{cc } -1 \\ -1 \end{array } \right ] \}$ ] .
the disconnectedness of @xmath33 was studied in @xcite . here
we want to mention another unconventional sufficient condition for disconnectedness . in the rest of the paper , we study attractors @xmath33 in the plane such that @xmath50 has a reducible characteristic polynomial . from @xcite , we know that such matrices are conjugate to one of the following lower triangular matrices @xmath51 } , \ \ \
\ \mbox{where $ |n| \geq |m|$ , and $ t=0 $ or $ t=1 $ } .\ ] ] we also let @xmath52 } : 0\leq i \leq |n| -1 , \ 0\leq j
\leq |m| -1\}.\ ] ] the attractors of the next proposition can be considered as a generalization of sierpi@xmath47ski carpets @xcite .
let @xmath53 be the _ singular value dimension _ of @xmath17 ( see @xcite ) .
we call a collinear digit set @xmath25 with @xmath54 is an eigenvector of @xmath24 _ eigen - collinear_. in that case , @xmath17 is a subset of a line segment . by using corollary 5 in @xcite , we obtain the following . [ disconnectedness ] assume that @xmath24 is as in ( [ matrix ] ) , @xmath55 , and @xmath25 is not eigen - collinear . then @xmath33 is disconnected if @xmath56 , where @xmath57 and @xmath58 is the number of @xmath59 so that @xmath60}\in d$ ] for some @xmath61 .
it is easy to check the sufficient condition for the attractors @xmath33 in proposition [ disconnectedness ] because , in that case , @xmath62
in this section , we will present a practical way of checking the connectedness of @xmath33 with @xmath24 as in ( [ matrix ] ) and @xmath55 . note that it is enough to consider the case @xmath63 , since @xmath64 .
let @xmath65\})$ ] , which we call the _ neighbor set _ of @xmath17 .
set @xmath66 , \ \ \
a_2=\tiny \left[\begin{array}{cc } 0 \\
m-1 \end{array } \right ] , \ \ \
a_3=\tiny \left[\begin{array}{cc } n-1 \\
m-1 \end{array } \right ] } , \ \ \
b_1=\tiny \left[\begin{array}{cc } n-1 \\ 0 \end{array } \right],\ ] ] @xmath67 , \ \
e_2=\tiny \left[\begin{array}{cc } 0 \\ 1 \end{array } \right ] , \ \
e_3=\tiny \left[\begin{array}{cc } 1
\\ 1 \end{array } \right ] , \ \
e_4=\tiny \left[\begin{array}{cc } 1 \\ -1 \end{array } \right ] } .\ ] ] for a digit set @xmath25 , an _
s - chain _ ( in @xmath25 ) is a finite sequence @xmath76 of @xmath77 vectors in @xmath25 such that @xmath78 for @xmath79 then we can put the connectedness criterion in @xcite into the following form . note that if @xmath80 after re - indexing , it is possible that @xmath81 for @xmath82 .
two examples are given in figure [ fig2 ] .
for the first fractal of figure [ fig2 ] , we have @xmath83 and @xmath84,d_2=\left [ \begin{array}{cc } 2 \\ 1 \end{array } \right ] , d_3=\left [ \begin{array}{cc } -1 \\ 1 \end{array } \right ] , d_4=\left [ \begin{array}{cc } 1 \\ 3 \end{array } \right ] , d_5=\left [ \begin{array}{cc } 2 \\ 0 \end{array } \right ] , d_6=\left [ \begin{array}{cc } 2 \\ 2 \end{array } \right ] , d_7=\left [ \begin{array}{cc } -2 \\ 2 \end{array } \right ] , d_8=\left [ \begin{array}{cc } -1 \\ 3 \end{array }
\right ] \}. $ ] in view of proposition [ neighbors1 ] and proposition [ neighbors2 ] , proposition [ connectedness ] is quite feasible because the connectedness can be decided by a simple inspection of @xmath25 using @xmath85 in propositions [ neighbors1]-[neighbors2 ] .
that is , we get a graph - independent way of checking the connectedness .
for the second fractal on the right , @xmath86 and @xmath49 , d_2=\left [ \begin{array}{cc } 1 \\ 1 \end{array } \right ] , d_3=\left [ \begin{array}{cc } 2 \\ 2 \end{array } \right ] , d_{4}=\left [ \begin{array}{cc } 3 \\ 3 \end{array } \right ] , d_{5}=\left [ \begin{array}{cc } 2 \\ 1 \end{array } \right ] , d_{6}=\left [ \begin{array}{cc } 1 \\ 2 \end{array } \right ] , d_{7}=\left [ \begin{array}{cc } 0 \\ 3 \end{array } \right ] , d_{8}=\left [ \begin{array}{cc } 3 \\ 0 \end{array } \right ] \}$ ] . for the general case @xmath87 , we have the following trivial proposition , which again can be used together with proposition [ connectedness ] .
let @xmath88 @xmath89 note that it is possible that @xmath90 or @xmath91 .
t. k. dey , h. edelsbrunner , and s. guha , _ computational topology _ in advances in discrete and computational geometry , b. chazelle , j. e. goodman and r. pollack , eds .
, contemp .
, vol 223 , american mathematical society , providence , ri , 1999 , pp . 109 - 143 . | somehow , the revised version of our paper @xcite does not appear on journals home page . here
we present the revised version altered to reflect the corrections and/or additions to that paper . in this note
, we consider self - affine attractors that are generated by an integer expanding @xmath0 matrix ( i.e. , all of its eigenvalues have moduli @xmath1 ) and a finite set of vectors in @xmath2 .
we concentrate on the problem of connectedness for @xmath3 .
although , there has been intensive study on the topic recently , this problem is not settled even in the one - dimensional case .
we focus on some basic attractors , which have not been studied fully , and characterize connectedness
. * keywords . * self - affine attractors , self - affine tiles , connectedness . |
in many scenarios of physics beyond the standard model ( sm ) , the top quark provides a preferred search window due to its large coupling to the physics responsible for the electro - weak symmetry breaking . a fascinating possibility
then is that the top quark properties exhibit deviations from their predictions within the sm .
the leading such contributions are encoded in the anomalous top couplings to sm gauge fields affecting the top anomalous ( chromo)-electric and ( chromo)-magnetic dipole moments @xcite , flavor changing neutral current ( fcnc ) couplings ( @xmath9 , where @xmath10 ) @xcite , as well as anomalous charged current interactions ( @xmath11 ) .
precision studies of top quark properties are currently underway at the lhc and tevatron colliders and will complete our knowledge of possible new physics ( np ) effecting the properties and interactions of the sm particles .
in particular , both tevatron @xcite and lhc @xcite experiments have recently presented precise measurements , sensitive to the top quark @xmath12 boson
@xmath13 quark vertex ( @xmath8 ) .
recently , we have reconsidered the @xmath14 helicity amplitudes in presence of anomalous @xmath8 interactions , including corrections due to perturbative qcd effects @xcite , and found that some of the anomalous couplings are already being significantly constrained by the cdf and d measurements . in light of the first lhc results being able to provide constraints on possible anomalous @xmath8 couplings , a more recent study @xcite has combined bounds coming from @xmath14 helicity amplitudes , angular asymmetries as well as single top production observables . on the other hand
, the @xmath8 vertex also plays an important role in @xmath15 physics , since fcnc processes involving @xmath13 quarks receive , within the sm , dominant contributions from loops involving a top quark and a @xmath12 boson . during the recent years
, the @xmath15 factories , cdf and d experiments at the tevatron as well as lhcb at the lhc have proceeded to probe @xmath1 oscillation observables , as well as rare fcnc mediated @xmath15 decays with ever increasing precision .
the first @xmath15 physics observable to be studied for the effects of the anomalous @xmath8 vertices was the @xmath16 decay rate @xcite .
the authors considered a complete basis of dimension six operators , free of tree - level fcncs . to simplify the analysis , they assumed alignment of the operator flavor structures with the sm ckm matrix , as well as the reality of the appropriate wilson coefficients .
recently @xcite , we have explored the possible presence of anomalous @xmath11 interactions in @xmath1 oscillations within a model independent approach based on the assumptions of minimal flavor violation ( mfv ) @xcite , which extends the basis of operator flavor structures considered in @xcite .
we have found that the @xmath17 mixing amplitudes can receive large contributions and that some of the possible operators can provide significant new sources of cp violation . in term , based on the recent global fits to neutral meson oscillation observables @xcite , we have derived preferred ranges for the anomalous couplings , some of them being more constraining than both @xmath16 as well as direct measurements . inspired by the fast progress in the field we extend our analysis of the anomalous @xmath11 vertices to rare ( semi)leptonic @xmath15 decays .
again we rely on mfv in order to reduce the number of possible operator flavor structures , and we review the framework in sec .
after performing the one - loop matching of our operator basis onto the low energy effective hamiltonian describing @xmath18 fcnc processes , we obtain corrections to the relevant wilson coefficients in sec .
we proceed in sec .
iv by calculating the effects in the inclusive @xmath19 and @xmath20 decays , cross - checking the existing calculation @xcite in the former mode . in order to derive bounds on both the real and imaginary parts of the appropriate wilson coefficients we include the experimental results not only for the decay rates but also for the cp asymmetry in @xmath16 . after performing a global fit of the wilson coefficients , we derive predictions for several rare @xmath15 meson processes like @xmath6 , the forward - backward asymmetry ( fba ) in @xmath3 and the branching ratios for @xmath7 in sec . v. finally we conclude in sec . vi .
we adopt the framework used in our previous work @xcite , committing to an effective theory , described by the lagrangian @xmath21 where @xmath22 is the sm part , @xmath23 is the scale of np and @xmath24 are dimension - six operators , invariant under sm gauge transformations and consisting of sm fields .
we assume that at the scale @xmath25 the sm fields with up to two higgs doublets are the only propagating degrees of freedom , that the electroweak symmetry is only broken by the vacuum expectation values of these two scalars and that operators up to dimension six give the most relevant contributions to the observables we consider .
such an approach is appropriate to summarize weak scale effects of np at @xmath26 , where the new heavy degrees of freedom have been integrated out .
our operator basis consists of all dimension - six operators that generate charged current quark interactions with the @xmath12 , but do not induce fcncs at the tree - level . since we restrict our discussion to mfv scenarios , lagrangian ( [ eq : lagr ] ) has to be formally invariant under the sm flavor group @xmath27 where @xmath28 stand for quark doublets and up and down type quark singlets respectively .
mfv requires that the only @xmath29 symmetry breaking spurionic fields in the theory are the up and down quark yukawa matrices @xmath30 , formally transforming under @xmath29 as @xmath31 and @xmath32 respectively .
we identify four relevant quark bilinears with distinct transformation properties under @xmath29 : @xmath33 , @xmath34 , @xmath35 and @xmath36 transforming as @xmath37 , @xmath38 , @xmath39 and @xmath40 respectively . using these
, we can construct the most general @xmath29 invariant quark bilinear flavor structures as @xmath41 where @xmath42 are arbitrary polynomials of @xmath43 and/or @xmath44 , transforming as @xmath38 . in order to identify the relevant flavor structures in terms of physical parameters , we can without the loss of generality
consider @xmath30 condensate values in a basis in which @xmath45 is diagonal : @xmath46 and @xmath47 , where we have introduced separate up- and down - type higgs condensates @xmath48 , while @xmath49 is the sm ckm matrix .
we also write @xmath50 in this basis in terms of quark mass eigenstates @xmath51 , where @xmath52
subscripts denote chirality projectors @xmath53 .
the final specification of the operator basis has been attentively described in ref .
it consists of seven distinct operators @xmath54 \big(\phi_d^\dagger\tau^a{\mathrm{i}}d_{\mu}\phi_d\big ) \hspace{-0.1cm}-\hspace{-0.1cm}[\bar q'_3\gamma^{\mu}q'_3]\big(\phi_d^\dagger{\mathrm{i}}d_{\mu}\phi_d\big),\\ { \nonumber}\mathcal q'_{ll}&=&[\bar q_3\tau^a\gamma^{\mu}q_3 ] \big(\phi_d^\dagger\tau^a{\mathrm{i}}d_{\mu}\phi_d\big ) \hspace{-0.1cm}-\hspace{-0.1cm}[\bar q_3\gamma^{\mu}q_3]\big(\phi_d^\dagger{\mathrm{i}}d_{\mu}\phi_d\big),\\ { \nonumber}\mathcal q^{\prime\prime}_{ll}&=&v_{tb}^{*}\big\{[\bar q'_3\tau^a\gamma^{\mu}q_3 ] \big(\phi_d^\dagger\tau^a{\mathrm{i}}d_{\mu}\phi_d\big ) \hspace{-0.1 cm } -[\bar q'_3\gamma^{\mu}q_3]\big(\phi_d^\dagger{\mathrm{i}}d_{\mu}\phi_d\big)\big\}\,,\\ { \nonumber}\mathcal q_{rr}&= & v_{tb } [ \bar{t}_r\gamma^{\mu}b_r ] \big(\phi_u^\dagger{\mathrm{i}}d_{\mu}\phi_d\big ) \ , , { \nonumber}\\ { \nonumber}\mathcal q_{lrb } & = & [ \bar q_3 \sigma^{\mu\nu}\tau^a b_r]\phi_d w_{\mu\nu}^a \,,\\ { \nonumber}\mathcal q_{lrt } & = & [ \bar q'_3 \sigma^{\mu\nu}\tau^a t_r]{\phi_u}w_{\mu\nu}^a \,,\\ \mathcal q'_{lrt } & = & v_{tb}^ { * } [ \bar q_3 \sigma^{\mu\nu}\tau^a t_r]{\phi_u}w_{\mu\nu}^a \ , , \label{eq : ops1}\end{aligned}\ ] ] where we have introduced the left - handed @xmath55 doublets @xmath56 with @xmath57 flavor indices .
furthermore we have @xmath58 and finally @xmath59/2 $ ] , while @xmath60 are the up- and down - type higgs fields ( in the sm @xmath61 ) .
as already mentioned in the introduction , our operator basis coincides with the one used in ref .
@xcite , expanded by the three primed operators .
furthermore we do not make the operators hermitian , hence effects of operators @xmath62 are accompanied by @xmath63 and will be kept track of separately . for completeness
we note that in scenarios , where the @xmath64 expansion is perturbative , the operators @xmath65 and @xmath66 are expected to dominate , with @xmath67 and @xmath68 being suppressed by a single power of @xmath69 and the operators @xmath70 and @xmath71 being supressed by @xmath72 .
to establish how operators ( [ eq : ops1 ] ) affect @xmath73 and @xmath74 transitions , we match our effective theory ( [ eq : lagr ] ) , to the low energy effective theory described by @xmath75 + \frac{4 g_f}{\sqrt{2}}v_{tb}v_{ts}^*\big[\sum_{i=3}^{10 } c_{i}{\cal o}_i + c_{\nu\bar{\nu}}{\cal o}_{\nu\bar{\nu}}\big]\ , , \label{eq : loweff}\end{aligned}\ ] ] where the first term consists of kinetic terms of the light sm particles as well as their qcd and qed interactions .
the relevant operators read @xmath76 on the other hand , since they are not that crucial for our analysis , we omit the definition of the four - quark operators @xmath77 which can be found for example in ref .
@xcite .
we perform the matching of the operators in to by integrating out the top quark and electroweak bosons at leading order ( lo ) in qcd .
we perform our calculation in a general @xmath78 gauge for the weak interactions , giving us the opportunity to check the @xmath79 cancelation in the final results .
the drawback is the appearance of would - be goldstone bosons in both sm as well np contributions . as a consequence of formulating the effective theory in a gauge and mfv invariant manner
, our operators contribute to processes of interest with grater intricacy than just a mere modification of the @xmath8 vertex . generic penguin and box diagrams with anomalous couplings
are shown in fig .
[ fig : feyns11 ] and fig .
[ fig : feyns3 ] . exact diagrams for a specific @xmath24 can be reconstructed using feynman rules given in the appendix [ sec : app1 ] . in all calculations we neglect the @xmath80 quark and lepton masses .
when dealing with @xmath81 transitions , we expand the amplitudes up to second order in external momenta and keep first order @xmath82 terms .
this allows us to generate operators @xmath83 as well as photonic contributions to @xmath84 . on the other hand , in the effective @xmath85 and box diagram contributions presented in fig .
[ fig : feyns3 ] , which both contribute to @xmath86 , the external momenta can be completely neglected .
we are interested in np contributions to the observables at order @xmath87 and thus only need to consider single operator insertions . as the result of the matching procedure
we obtain deviations from the sm values for the wilson coefficients , i.e. @xmath88 , which we parametrize as @xmath89 where @xmath90 , @xmath91 is the matching scale and @xmath92 are defined as @xmath93 analytical expressions for @xmath94 are given in the appendix [ sec : app2 ] .
we note , that the matching procedure for operator @xmath95 stands out compared to the other operators .
the charged current structure of this operator resembles that of the sm operator @xmath96 .
consequently wilson coefficients @xmath97 are changed in a trivial way @xmath98 .
the change of the remaining wilson coefficients @xmath99 , matching of which involves the @xmath100 boson , is however not of this form .
some of the diagrams in fig .
[ fig : feyns11 ] are uv divergent .
we remove these divergences using the @xmath101 prescription leading to remnant @xmath102 terms .
we shall quantify the matching scale dependence of our results and consequently their sensitivity to the uv completion of the effective theory by varying the scale between @xmath103 and @xmath104 .
since uv renormalization is necessary , our operator basis indeed needs to be extended to include operators that can serve as the appropriate counter - terms . within the employed mfv framework examples of these operators read @xmath105\phi_d w^a_{\mu\nu}\,,\hspace{0.5 cm } { \cal q}_2=\big[\bar{q}\gamma^{\mu}a_{qq}q\big]\big[\bar{\ell}\gamma_{\mu}\ell\big]\ ,
, \nonumber\\ { \cal q}_3=\big[\bar{q}\gamma^{\mu}\tau^a a^{{\prime}}_{qq}q\big]\big[\phi_d^{\dagger}\tau^a{\mathrm{i}}d_{\mu}\phi_d\big ] + \big[\bar{q}\gamma^{\mu } a^{{\prime}}_{qq}q\big]\big[\phi_d^{\dagger}{\mathrm{i}}d_{\mu}\phi_d\big]\ , .
\label{eq : counterterms}\end{aligned}\ ] ] the operator @xmath106 produces a counter - term for divergences in @xmath107 , while @xmath108 provide counter - terms for divergent parts of @xmath109 .
the operator @xmath110 generates a tree - level @xmath111 vertex .
the sets of flavor matrices needed to match the structures of divergencies generated by the various operators in are @xmath112 in our numerical analysis , we will selectively set contributions of certain operators to be nonzero .
consequently in the following , we will drop the implicit ( tree - level ) contributions of the operators in to @xmath113 , as these have been already investigated and constrained in the existing literature @xcite .
all @xmath94 are found to be @xmath79 independent and a crosscheck with results from ref .
@xcite is possible for some of them .
we confirm their original results for all the operators except @xmath68 , while an updated version of @xcite confirms our result also for this operator .
.numerical values of functions @xmath114 and @xmath115 at @xmath116 .
numerical values used for the input parameters are @xmath117 gev , @xmath118 , @xmath119 gev , @xmath120 gev , @xmath121 .
all @xmath122 values correspond to matching at lo in qcd . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] the first column shows bounds obtained from @xmath123 mixing as analyzed in ref .
@xcite , while the last column corresponds to combined bounds from all three observables . for
the later we also present the results when the matching scale is set to @xmath104 to check the scale dependence of our results .
we can see that the bounds obtained change significantly only in the case of @xmath124 where lowering the scale to @xmath104 loosens the bounds by almost an order of magnitude .
we have also checked that the @xmath125 bounds agree nicely with those obtained in ref .
@xcite .
we have shown in ref .
@xcite that imaginary parts of primed wilson coefficients can affect the cp violating phase in @xmath126 mixing and nonzero values were found to improve the global fit of @xcite . to constrain imaginary parts of the remaining four operators , which do not contribute with new phases in @xmath126 mixing , we consider the direct cp asymmetry in @xmath125 for which the current world average experimental value reads @xcite @xmath127 based on the recent analysis of this observable in ref .
@xcite we obtain the following semi - numerical formula @xmath128\mathrm{re}[\delta c_7(2m_w ) ] + \big[0.012(\tilde{\lambda}_{17}^u- \tilde{\lambda}_{17}^c)+0.002\big]\mathrm{re}[\delta c_8(2m_w)]\nonumber\\ & + & \big[-0.256 + 0.264 \tilde{\lambda}_{78}-0.023 \tilde{\lambda}_{17}^u-2.799 \tilde{\lambda}_{17}^c\big]\mathrm{im}[\delta c_7(2m_w)]\nonumber\\ & + & \big[-0.668 \tilde{\lambda}_{17}^c-0.005 \tilde{\lambda}_{17}^u-0.563 \tilde{\lambda}_{78}+0.135\big]\mathrm{im}[\delta c_8(2m_w)]\,.\end{aligned}\ ] ] the estimated intervals for hadronic parameters @xmath129 , @xmath130 and @xmath131 as specified in ref .
@xcite dominate the theoretical uncertainty making it sufficient to use a lo qcd analysis in the perturbative regime .
thus , in addition to the numerical parameters specified in @xcite , we have used the lo qcd running for @xmath132 in this observable . performing a combined analysis of all considered bounds on the real and imaginary parts of individual @xmath133 in which we marginalize the hadronic parameters entering @xmath134 within the allowed intervals , we can obtain the allowed regions in the complex plain of @xmath135,{\rm im}[\kappa_i])$ ] . and
@xmath136 c.l . allowed regions in the @xmath137 ( solid ) and @xmath138 ( dashed ) complex plain .
the constraints are dominated by @xmath139 $ ] and @xmath134 . ]
as already argued , the imaginary part of @xmath140 does not contribute to the @xmath113 and thus remains unconstrained .
it also turns out that due to the large hadronic uncertainties , the imaginary parts of @xmath141 , @xmath142 remain largely unconstrained by @xmath143 and @xmath126 mixing observables still provide the strongest constraints @xcite ( except for @xmath144 $ ] which again remains unconstrained ) . on the contrary ,
constraints on the imaginary parts of @xmath145 and @xmath124 reach per - cent level , as can be seen in fig .
[ fig : cp ]
. finally we note that in absence of the long - distance effects on np contributions considered in @xcite , @xmath143 would exhibit an even greater sensitivity to the imaginary parts of @xmath132 @xcite , thus we consider our derived bounds as conservative .
anomalous @xmath8 couplings can be studied directly at colliders , namely in the main top quark decay channel ( @xmath146 ) as well as in the single top production since in the sm both processes proceed through weak interactions .
contrary to indirect studies of @xmath8 couplings in @xmath15 physics , the most general parametrization of the @xmath8 vertex of the form @xmath147tw_{\mu}\,,\end{aligned}\ ] ] is sufficient to a very good approximation , where @xmath148 .
it is straightforward to identify the anomalous couplings @xmath149 and @xmath150 as generated by our seven operators in eq .
( [ eq : ops1 ] ) @xmath151 enabling us to compare recent direct bounds obtained from the tevatron data @xcite with our indirect constraints .
the comparison is shown in fig .
[ fig:2d1 ]
. obtained from @xcite . also shown
is the latest combined tevatron measurement @xcite and allowed intervals for @xmath152 as given in tab .
[ tab : bounds ] . *
middle * : @xmath136 c.l . allowed regions in the @xmath145 - @xmath140 ( solid ) , @xmath153 ( dashed ) , @xmath154 ( dotted ) plane .
matching scale is set to @xmath116 .
* right * : @xmath136 c.l . allowed regions in the @xmath124 - @xmath155 ( solid ) , @xmath156 ( dashed ) plane . matching scale
is set to @xmath116 ( full regions ) and @xmath104 ( hatched regions ) .
thick black lines correspond to bands of @xmath136 c.l .
allowed regions given by direct constraints from ref . @xcite . ]
the first plot shows the longitudinal helicity fraction @xmath157 as a function of the real parts of the anomalous couplings @xmath158 obtained from @xcite and the latest combined measurements from tevatron with the @xmath136 c.l .
band @xcite .
we find the combined indirect constraints to be more stringent than direct bounds coming from helicity fraction measurements .
for other operators this conclusion is even more pronounced . having more than one observable at disposal we can also consider pairs of operators contributing simultaneously and obtain @xmath136 c.l .
allowed regions in the corresponding planes as shown in the second and third plot of fig .
[ fig:2d1 ] .
the full and hatched shapes are obtained from indirect bounds while thick black line marks the border of the region obtained from helicity fraction and single top production analysis of ref .
in particular both vertical lines in the second graph are due to single top production measurements , as helicity fractions are insensitive to a change in the left - handed couplings .
also , the bottom thick line in the third graph is due to constraints from single top production , since as indicated in the first graph , the negative values of @xmath158 are better constrained indirectly than from helicity fractions .
it is apparent that indirect constraints on the real parts of @xmath145 and @xmath124 are at present much more stringent than direct constraints .
having derived bounds on anomalous @xmath92 couplings , it is interesting to study to what extent these can still affect other rare @xmath15 decay observables . analyzing one operator at a time we set the matching scale to @xmath103 . turning again to the semi - numerical formulae given in ref .
@xcite , we first consider the branching ratio @xmath159 $ ] for which cdf s latest analysis yields @xcite @xmath160<3.9\times 10^{-8}\,,\hspace{0.5 cm } \text{at $ 90\%$ c.l.}\,.\end{aligned}\ ] ] and the differential forward - backward asymmetry @xmath161 in the @xmath162 decay , for which the latest measurement of lhcb has recently been presented in ref .
@xcite . finally , following ref .
@xcite we analyze the allowed effects of @xmath133 on the branching ratios @xmath163 , which are expected to become experimentally accessible at the super - b factories @xcite .
intervals given in tab .
[ tab : bounds ] .
line in the @xmath156 bar marks the value obtained when the coupling is set to the central fitted value given by the @xmath126 mixing analysis .
we also show the sm predictions ( black ) with @xmath164 theoretical uncertainty band ( dotted ) and for the muonic decay channel the lower end of the experimental @xmath165 c.l .
interval from @xcite . ]
we present our findings in figs . [
fig : predict1 ] and [ fig : predict2 ] .
the effects of anomalous couplings @xmath92 on all branching ratios are similar .
there is a slight tenancy of anomalous @xmath92 couplings to increase the predictions compared to the sm values at the level of the present theoretical uncertainties , with the exception of @xmath145 and @xmath124 of which effects are negligible . in particular
, none of the contributions can accommodate the recent cdf measurement of @xmath166 $ ] at the @xmath136 c.l .
, while a possible future measurement at the level of the sm could significantly constrain the @xmath167 and @xmath141 contributions .
furthermore we find that the forward - backward asymmetry @xmath161 can still be somewhat effected by @xmath154 and @xmath145 , for which we present the bands obtained when varied within the @xmath136 c.l .
intervals in fig .
[ fig : predict2 ] . while not sensitive at the moment , in the near future , improved measurements by the lhcb experiment
could possibly probe such effects . on the other hand , the contributions of other anomalous couplings all fall within the theoretical uncertainty bands around the sm predicted curve .
we have investigated contributions of anomalous @xmath8 couplings in flavor changing neutral current mediated @xmath18 processes within an effective field theory framework assuming minimal flavor violation . having computed contributions to the inclusive @xmath4 decay rate to order @xmath87 , and combining them with the modifications of the @xmath5 and also @xmath1 observables , we have determined the indirect bounds on the real and imaginary parts of the anomalous @xmath8 couplings . for most of the considered effective operators , these indirect bounds are at present much stronger than the direct constraints coming from the @xmath146 helicity fraction measurements , angular asymmetries and single top production at the tevatron and the lhc . in particular , we are able for the first time to constrain the imaginary parts of most of the anomalous couplings already at order @xmath87 . taking into account these bounds ,
we have predicted the still allowed effects of the anomalous @xmath8 interactions on the branching ratio of the @xmath6 decay , the forward - backward asymmetry in @xmath3 , as well as the branching ratios of @xmath7 decays . the better knowledge of these and other recently proposed @xcite observables in the future could further constrain some of the anomalous couplings .
we thank bohdan grzadkowski and mikolaj misiak for useful correspondence and for positive verification of our corrected result for the function @xmath168 in eq .
( 12 ) of ref .
@xcite . j.f.k . acknowledges insightful conversations with jure zupan , joaquim matias , sbastien descotes - genon , and gino isidori .
this work is supported in part by the slovenian research agency , by the national science foundation under grant no .
1066293 and the hospitality of the aspen center for physics .
the relevant feynman rules with anomalous couplings are shown in tab . [
tab : feyns ] . we have used the following abbreviations @xmath169 where @xmath57 denote flavor indices .
m2.0cml|m2.0cml & @xmath170p_{r , l}$ ] & & @xmath171 + & @xmath172p_{r , l}\end{array}$ ] & & @xmath173p_{r , l}$ ] + & @xmath174 & & @xmath175 + & @xmath176 & & @xmath177 + & @xmath178 & & @xmath179 +
in this appendix we present analytical expressions for functions @xmath114 and @xmath115 defined in eq .
( [ eq : fs ] ) . for shorter notation
we further decompose @xmath180 below we give all nonzero contributions @xmath181\label{eq : rr}\\ f_7^{(lrb)}&=&\frac{m_w}{m_b}\big[-\frac{x}{2}\log \frac{m_w^2}{\mu^2}+\frac{6x^3 - 31x^2 + 19x}{12(x-1)^2}+\frac{-3x^4 + 16x^3 - 12x^2 + 2x}{6(x-1)^3}\log x\big ] \label{eq : lrb}\\ f_7^{(lrt)}&=&\frac{m_t}{m_w}\big[\frac{1}{8}\log\frac{m_w^2}{\mu^2}+\frac{-9x^3 + 63x^2 - 61x+19}{48(x-1)^3}+\frac{3x^4 - 12x^3 - 9x^2 + 20x-8}{24(x-1)^4}\log x\big]\\ \tilde{f}_7^{(lrt)}&=&\tilde{f}_7^{(lrt\prime)}=\frac{m_t}{m_w}\big[\frac{-3x^3 + 17x^2 - 4x-4}{24(x-1)^3}+\frac{2x-3x^2}{4(x-1)^4}\log x\big]\\ f_7^{(lrt\prime)}&=&\frac{m_t}{m_w}|v_{tb}|^2\big[\frac{-x^2-x}{8(x-1)^2}+\frac{x^2\log x}{4(x-1)^3}\big]\\ f_8^{(ll)}&=&\tilde{f}_8^{(ll)}=f_8^{(ll{\prime}{\prime})}=\frac{5 x^3 - 9 x^2 + 30 x-8}{24 ( x-1)^3}-\frac{3 x^2 \log x}{4 ( x-1)^4}\\ f_8^{(ll{\prime})}&=&\frac{-x^3 + 5x^2 + 2x}{8 ( x-1)^3}-\frac{3 x^2 \log x}{4 ( x-1)^4}\\ f_8^{(rr)}&=&\frac{m_t}{m_b}\big[\frac{-x^2-x-4}{4 ( x-1)^2}+\frac{3 x \log x}{2 ( x-1)^3}\big]\\ f_8^{(lrb)}&=&\frac{m_w}{m_b}\big[\frac{x^2 + 5 x}{4 ( x-1)^2}+\frac{2 x^3 - 6 x^2+x}{2 ( x-1)^3}\log x\big]\\ f_8^{(lrt)}&=&\frac{m_t}{m_w}\big[\frac{3 x^2 - 13 x+4}{8 ( x-1)^3}+\frac{5 x-2 } { 4 ( x-1)^4}\log x\big]\\ \tilde{f}_8^{(lrt)}&=&\tilde{f}_8^{(lrt{\prime})}=\frac{m_t}{m_w}\big[\frac{x^2 - 5 x-2}{8 ( x-1)^3}+\frac{3 x \log ( x)}{4 ( x-1)^4}\big]\\ g^{(ll)}&=&\tilde{g}^{(ll)}=(-x-\frac{4}{3})\log\frac{m_w^2}{\mu^2}+\frac{250x^3 - 384x^2 + 39x+77}{108 ( x-1)^3}\\ & + & \frac{-18x^5 + 48x^4 - 102x^3 + 135x^2 - 68x+8}{18(x-1)^4}\log x\\ g^{(ll{\prime})}&=&(\frac{4}{9}-\frac{x}{2})\log\frac{m_w^2}{\mu^2}+\frac{125x^3 - 253x^2 + 138x -16}{36(x-1)^3}\\ & + & \frac{-9x^5 + 12x^4 - 48x^3 + 99x^2 - 59x+8}{18(x-1)^4}\log x - |v_{tb}|^2\frac{x}{2}\nonumber\\ \tilde{g}^{(ll{\prime})}&=&\tilde{h}^{(ll{\prime})}=\tilde{f}^{(ll{\prime})}_{\nu\bar{\nu}}=-\frac{x}{2}\log\frac{m_w^2}{\mu^2}+\frac{x}{2}(1-\log x -|v_{tb}|^2)\\ g^{(ll{\prime}{\prime})}&=&-\big(\frac{4}{3}+\frac{x}{2}+|v_{tb}|^2\frac{x}{2 } \big)\log\frac{m_w^2}{\mu^2 } + \frac{250x^3 - 384x^2 + 39x + 77}{108(x-1)^3}\\ & + & \frac{-9x^5 + 12x^4 - 48x^3 + 99x^2 - 59x+8}{18(x-1)^4}\log x - |v_{tb}|^2\frac{x}{2}\log x \nonumber \\\end{aligned}\ ] ] @xmath182\\ g^{(lrt{\prime})}&=&\frac{m_t}{m_w}|v_{tb}|^2\big[\frac{x^2 + 3x-2}{8(x-1)^2}+\frac{x-2x^2}{4(x-1)^3}\log x\big]\\ \tilde{g}^{(lrt{\prime})}&=&\frac{m_t}{m_w}\bigg[\frac{-54x^3 + 59x^2 + 35x-34}{36(x-1)^3}+\frac{15x^3 - 27x^2 + 10x+1}{6(x-1)^4}\log x\\ & + & |v_{tb}|^2\big[\frac{x^2 + 3x-2}{8(x-1)^2}+\frac{x-2x^2}{4(x-1)^3}\log x\big]\bigg]\\ h^{(ll)}&=&\tilde{h}^{(ll)}=-(x+\frac{3}{2})\log\frac{m_w^2}{\mu^2}+\frac{11x-5}{4(x-1)}+\frac{-2x^3+x^2 - 2x}{2(x-1)^2}\log x\\ h^{(ll{\prime})}&=&-\frac{x}{2}\log\frac{m_w^2}{\mu^2}+\frac{3x}{2(x-1)}-\frac{x^3+x^2+x}{2(x-1)^2}\log x-|v_{tb}|^2\frac{x}{2 } \\
x-|v_{tb}|^2 \frac{x}{2}\log x\\ h^{(lrt)}&=&\tilde{h}^{(lrt)}=\tilde{h}^{(lrt{\prime})}=\frac{m_t}{m_w}\big[-\frac{3x}{2(x-1)}+\frac{3x\log x}{2(x-1)^2}\big]\\ k^{(ll)}&=&\tilde{k}^{(ll)}=h^{(ll)}-\frac{3}{(x-1)}+\frac{3x\log x}{(x-1)^2}\\ k^{(ll{\prime})}&=&h^{(ll{\prime})}-\frac{3x}{(x-1)}+\frac{3x\log x}{(x-1)^2}\\ k^{(ll{\prime}{\prime})}&=&h^{(ll{\prime}{\prime})}-\frac{3}{(x-1)}+\frac{3x\log x}{(x-1)^2}\\ k^{(lrt)}&=&\tilde{k}^{(lrt)}=\tilde{k}^{(lrt{\prime})}=h^{(lrt)}+\frac{3}{x-1}-\frac{3x\log x}{(x-1)^2 } \ ] ] t. aaltonen _ et al . _ [ the cdf collaboration ] , phys . rev .
lett . * 105 * , 042002 ( 2010 ) .
[ arxiv:1003.0224 [ hep - ex ] ] ; v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
rev . * d83 * , 032009 ( 2011 ) .
[ arxiv:1011.6549 [ hep - ex ] ] .
b. grzadkowski , m. misiak , phys .
* d78 * ( 2008 ) 077501 .
[ arxiv:0802.1413 [ hep - ph ] ] .
j. drobnak , s. fajfer , j. f. kamenik , phys .
lett . *
b701 * ( 2011 ) 234 - 239 .
[ arxiv:1102.4347 [ hep - ph ] ] .
g. dambrosio , g. f. giudice , g. isidori , a. strumia , nucl .
phys . * b645 * , 155 - 187 ( 2002 ) .
[ hep - ph/0207036 ] ; a. l. kagan , g. perez , t. volansky , j. zupan , phys .
rev . * d80 * , 076002 ( 2009 ) .
[ arxiv:0903.1794 [ hep - ph ] ] .
z. ligeti , m. papucci , g. perez , j. zupan , phys .
* 105 * , 131601 ( 2010 ) .
[ arxiv:1006.0432 [ hep - ph ] ] ; a. lenz , u. nierste , j. charles , s. descotes - genon , a. jantsch , c. kaufhold , h. lacker , s. monteil _ et al .
_ , phys .
* d83 * ( 2011 ) 036004 .
[ arxiv:1008.1593 [ hep - ph ] ] .
g. buchalla , a. j. buras , m. e. lautenbacher , rev .
* 68 * ( 1996 ) 1125 - 1144 .
[ hep - ph/9512380 ] .
t. hurth , g. isidori , j. f. kamenik , f. mescia , nucl .
* b808 * , 326 - 346 ( 2009 ) .
[ arxiv:0807.5039 [ hep - ph ] ] .
t. huber , e. lunghi , m. misiak , d. wyler , nucl .
* b740 * ( 2006 ) 105 - 137 .
[ hep - ph/0512066 ] .
d. asner _
[ heavy flavor averaging group collaboration ] , [ arxiv:1010.1589 [ hep - ex ] ] t. huber , t. hurth , e. lunghi , nucl . phys .
* b802 * ( 2008 ) 40 - 62 .
[ arxiv:0712.3009 [ hep - ph ] ] .
a. crivellin , l. mercolli , [ arxiv:1106.5499 [ hep - ph ] ] .
s. descotes - genon , d. ghosh , j. matias , m. ramon , jhep * 1106 * ( 2011 ) 099 .
[ arxiv:1104.3342 [ hep - ph ] ] k. g. chetyrkin , m. misiak , m. munz , phys .
lett . * b400 * , 206 - 219 ( 1997 ) .
[ arxiv : hep - ph/9612313 [ hep - ph ] ] . c. bobeth , m. misiak , j. urban , nucl .
phys . * b574 * , 291 - 330 ( 2000 ) .
[ hep - ph/9910220 ] ; j. a. gracey , phys .
lett . * b488 * , 175 - 181 ( 2000 ) .
[ hep - ph/0007171 ] ; p. gambino , m. gorbahn , u. haisch , nucl .
phys . * b673 * , 238 - 262 ( 2003 ) .
[ hep - ph/0306079 ] ; m. gorbahn , u. haisch , m. misiak , phys .
lett . *
95 * , 102004 ( 2005 ) .
[ hep - ph/0504194 ] .
a. j. buras , m. misiak , m. munz , s. pokorski , nucl .
phys . * b424 * , 374 - 398 ( 1994 ) .
[ hep - ph/9311345 ] .
a. freitas , u. haisch , phys .
rev . * d77 * , 093008 ( 2008 ) .
[ arxiv:0801.4346 [ hep - ph ] ] .
t. aushev , w. bartel , a. bondar , j. brodzicka , t. e. browder , p. chang , y. chao , k. f. chen _ et al .
_ , [ arxiv:1002.5012 [ hep - ex ] ] ; b. oleary _ et al . _
[ superb collaboration ] , [ arxiv:1008.1541 [ hep - ex ] ] . c. bobeth , g. hiller , g. piranishvili , jhep * 0712 * , 040 ( 2007 ) .
[ arxiv:0709.4174 [ hep - ph ] ] ; c. bobeth , g. hiller , g. piranishvili , jhep * 0807 * , 106 ( 2008 ) .
[ arxiv:0805.2525 [ hep - ph ] ] ; c. bobeth , g. hiller , d. van dyk , jhep * 1007 * , 098 ( 2010 ) .
[ arxiv:1006.5013 [ hep - ph ] ] ; c. bobeth , g. hiller , d. van dyk , jhep * 1107 * , 067 ( 2011 ) .
[ arxiv:1105.0376 [ hep - ph ] ] ; c. bobeth , g. hiller , d. van dyk , [ arxiv:1105.2659 [ hep - ph ] ] ; u. egede , t. hurth , j. matias , m. ramon , w. reece , jhep * 0811 * , 032 ( 2008 ) .
[ arxiv:0807.2589 [ hep - ph ] ] ; u. egede , t. hurth , j. matias , m. ramon , w. reece , jhep * 1010 * , 056 ( 2010 ) .
[ arxiv:1005.0571 [ hep - ph ] ] ; w. altmannshofer , p. ball , a. bharucha , a. j. buras , d. m. straub , m. wick , jhep * 0901 * , 019 ( 2009 ) .
[ arxiv:0811.1214 [ hep - ph ] ] ; d. becirevic , e. schneider , [ arxiv:1106.3283 [ hep - ph ] ] ; | precision studies of top quark properties are currently underway at the lhc and tevatron colliders with the prospect of probing anomalous @xmath0 interactions . in the mean time
, recent experimental results for the @xmath1 oscillation observables , the branching ratio @xmath2 , as well as the forward - backward asymmetry in @xmath3 , accompanied by the accurate theoretical predictions for the relevant observables obtained within the sm motivate a combined study of these observables in the presence of anomalous @xmath0 vertices .
we investigate contributions of such anomalous couplings to the @xmath4 decay mode , and combining them with the modifications of the @xmath1 , @xmath5 observables , we determine indirect bounds on the real and imaginary parts of the anomalous @xmath0 interactions .
we find these to be mostly superior to present direct constraints coming from top decay and production measurements at the lhc and tevatron .
finally , we predict the allowed effects in the branching ratios of the @xmath6 and @xmath7 , as well as the forward - backward asymmetry in @xmath3 . we find that improved knowledge of these observables in the future could further constrain some of the anomalous @xmath8 interactions . |
recently , four new charmed mesons , @xmath0 , @xmath1 , @xmath2 and @xmath3 , were observed by babar collaboration @xcite .
the @xmath15 and @xmath16 with neutral charge were first found in the @xmath17 channel .
then their isospin partners @xmath18 and @xmath19 were observed in @xmath20 as well
. further analysis of the @xmath21 invariant mass spectrum confirmed the @xmath15 .
furthermore , two additional new charmed mesons , @xmath22 and @xmath23 , were found in the @xmath21 channel .
the measured branching ratio fractions are @xmath24 the other observed results are summarized in tab . [ wfs ] . to determine the spin - parity @xmath25 of these newly observed charmed mesons , the babar collaboration also analyzed their helicity distributions .
these newly observed charmed mesons make great progress in the establishment of the charmed meson spectroscopy . from the pdg book @xcite
, it is seen that only six low - lying states , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 and @xmath31 , have been established .
the higher excitations , @xmath32 and @xmath33 waves , are still absent .
thus , the find of @xmath0 , @xmath1 , @xmath2 and @xmath3 provides us a good opportunity to establish the missing @xmath32 and @xmath33 states .
the @xmath22 may be identified as the radial excitation of the @xmath34 ( i.e. @xmath14 ) @xcite , for its quark model predicted mass @xmath35 gev @xcite , helicity distribution ( @xmath36 ) @xcite and dominated decay mode @xmath37 consist with the observations .
the @xmath1 is observed in both @xmath38 and @xmath37 channels , thus , its possible @xmath25 are @xmath39 and @xmath40 in the @xmath32 and @xmath33 states .
the babar analysis of helicity distribution ( @xmath41 ) @xcite also indicates that the @xmath1 may be @xmath39 or @xmath40 assignments @xcite .
the typical quark model predicted mass of @xmath4 is @xmath42 gev @xcite , which is much larger than that of @xmath1 .
thus , the @xmath1 as the @xmath5 assignment should be excluded
. the @xmath2 and @xmath3 may be good candidates of @xmath26 wave states for their masses are close to those of @xmath26 waves predicted in various quark models @xcite .
since the @xmath23 is observed in @xmath21 channel , its possible @xmath25 are @xmath39 , @xmath43 and @xmath40 .
the helicity distribution of the @xmath44 and @xmath40 assignments is a simple @xmath45 distribution @xcite , which is inconsistent with the babar observation that the @xmath23 does not show a simple helicity distribution @xcite .
although the mass of @xmath3 is very close to that of @xmath2 , they may be two different states for their mass and width values differ by 2.6@xmath46 and 1.5@xmath46 , respectively @xcite .
the observation of @xmath3 in @xmath38 channel indicates it may be a candidate of @xmath12 or @xmath4 . to distinguish the different candidates for these newly observed charmed mesons , in this work
, we study their strong decay properties in a constituent quark model , which has been developed and successfully used to deal with the strong decays of heavy - light mesons and charmed baryons @xcite .
very recently , the strong decays of the @xmath0 , @xmath1 and @xmath3 were studied by liu _
_ in a @xmath47 model @xcite . for the @xmath0 and @xmath1 , the main @xmath47 model predictions are compatible with our quark model predictions . in @xcite ,
two candidates are suggested for the @xmath3 .
they are the mixed state via @xmath13-@xmath12 mixing and @xmath4 , respectively . in our predictions , only the @xmath4 is the favored assignment to @xmath3 .
the paper is organized as follows . in the subsequent section
, a brief review of the model is given .
the numerical results are presented and discussed in sec .
finally , a summary is given in sec .
[ suma ] .
.summary of the experimental results . [ cols="^,^,^,^,^,^,^",options="header " , ] in this work
we have studied the strong decay properties of the newly observed @xmath0 , @xmath1 , @xmath2 and @xmath3 by babar collaboration in a constituent chiral quark model .
these newly observed charmed mesons provide us a chance to establish a more completed @xmath26 meson spectroscopy , which has been shown in tab .
[ dds ] . for comparison
, the @xmath48 meson spectroscopy is also included .
we have found that @xmath22 as the @xmath14 is still questionable .
the predicted narrow width of @xmath14 is inconsistent with the observation , although its decay modes , helicity distributions and theoretical predicted mass satisfy this classification .
given the poor statistics of @xmath22 , its decay width may be overestimated by experimentalists .
we expect them to observe it in both @xmath37 and @xmath49 channels .
the @xmath1 can be identified as the low mass mixed state @xmath50 via the @xmath13-@xmath12 mixing .
this mixed state is also predicted in the @xmath48 meson spectroscopy , which corresponds to the @xmath51 @xcite . in our
mixing scheme the high mass partner @xmath52 may be too broad to be observed in @xmath26 meson spectroscopy , while it might be found in @xmath48 spectroscopy @xcite .
to understand the nature of @xmath1 further , we suggest to observe it in @xmath53 , @xmath54 , @xmath55 and @xmath56 channels .
the @xmath3 is most likely to be the @xmath57 .
its decays are governed by @xmath38 and @xmath37 , which can naturally explain why @xmath3 is first observed in @xmath38 channel .
the predicted ratio is @xmath58 .
the @xmath3 as high mass partner of @xmath1 via the @xmath13-@xmath12 mixing was also suggested in @xcite , where the predicted partial decay width of @xmath38 is tiny .
further experimental measurement of the ratios @xmath59 , @xmath60 and @xmath61 should be able to disentangle its properties .
the @xmath2 and @xmath3 might be two different charmed meson states .
the @xmath2 is favorably interpreted as the high mass mixed state @xmath6 ( @xmath43 ) via the @xmath8-@xmath9 mixing .
its low - mass partner @xmath62 may be too broad to be observed in experiments .
to confirm the @xmath2 , the decay channels @xmath53 , @xmath49 , @xmath63 and @xmath54 are suggested to be observed in future experiments .
finally , we should mention that in our previous work @xcite , we predicted that @xmath64 might correspond to two largely overlapping resonances , one resonance is likely to be the @xmath4 [ denoted by @xmath65 and the other resonance seems to be the mixed state between @xmath9 and @xmath8 [ denoted by @xmath66 . combining the study of the @xmath2 and @xmath3 in present work , we easily conclude that both @xmath3 and @xmath67 are most likely to be the @xmath4 , while both @xmath2 and @xmath68 might be classified as the mixed state @xmath6 with almost the same mixing angle . to test our predictions and clarify the controversial situation of @xmath64 @xcite
, we suggest to analyze the helicity distribution of @xmath69 in experiments .
if the helicity distribution is in proportion to @xmath70 , the @xmath64 should be two largely overlapping resonances .
the author would like to thank professor xiang liu for drawing his attention to the babar collaboration s new data .
this work is supported by the national natural science foundation of china ( grant nos .
10775145 and 11075051 ) .
j. benitez _
et al . _ , in ichep2010 , paris , july 23 , 2010 ; p.
del amo sanchez _ et al .
_ [ the babar collaboration ] , phys . rev .
d * 82 * , 111101 ( 2010 ) . c. amsler
_ et al . _
[ particle data group ] , phys .
b * 667 * , 1 ( 2008 ) .
s. godfrey and n. isgur , phys .
d * 32 * , 189 ( 1985 ) .
x. h. zhong and q. zhao , phys .
d * 77 * , 074008 ( 2008 ) .
z. f. sun , j. s. yu , x. liu and t. matsuki , phys .
d * 82 * , 111501 ( 2010 ) .
a. manohar and h. georgi , nucl .
b * 234 * , 189 ( 1984 ) .
z. p. li , phys .
d * 50 * , 5639 ( 1994 ) .
q. zhao , z. p. li and c. bennhold , phys .
c * 58 * , 2393 ( 1998 ) ; phys .
b * 436 * , 42 ( 1998 ) .
q. zhao , phys .
c * 63 * , 025203 ( 2001 ) .
q. zhao , j. s. al - khalili and c. bennhold , phys .
c * 64 * , 052201 ( 2001 ) .
i. hleiqawi _ et al . _ [ clas collaboration ] , phys .
c * 75 * , 042201 ( 2007 ) [ erratum - ibid .
c * 76 * , 039905 ( 2007 ) ] .
m. nanova _ et al .
_ [ cbelsa / taps collaboration ] , eur .
j. a * 35 * , 333 ( 2008 ) .
p. colangelo , f. de fazio , s. nicotri and m. rizzi , phys . rev .
d * 77 * , 014012 ( 2008 ) .
f. e. close , c. e. thomas , o. lakhina and e. s. swanson , phys .
b * 647 * , 159 ( 2007 ) .
d. m. li and b. ma , phys .
d * 81 * , 014021 ( 2010 ) .
f. e. close and e. s. swanson , phys .
d * 72 * , 094004 ( 2005 ) .
zhi - gang wang , arxiv:1009.3605 [ hep - ph ] . | the strong decay properties of the newly observed @xmath0 , @xmath1 , @xmath2 and @xmath3 are studied in a constituent quark model .
it is predicted that the @xmath3 and @xmath2 seem to be two overlapping resonances .
the @xmath3 could be identified as the @xmath4 with @xmath5 , while the @xmath2 is most likely to be the high - mass mixed state @xmath6 ( @xmath7 ) via the @xmath8-@xmath9 mixing .
the @xmath1 favors the low - mass mixed state @xmath10 @xmath11 via the @xmath12-@xmath13 mixing .
the @xmath0 as the @xmath14 assignment bears controversies for its too broad width given in experiments . |
in many cases , the results of analytical calculation of feynman diagrams can be represented as combinations of hypergeometric functions .
however , the problem of constructing the @xmath0-expansion of hypergeometric functions within dimensional regularization @xcite ( where @xmath6 is the space - time dimension ) is not completely solved .
rather often , results for the terms of the @xmath0-expansion can be expressed in terms of polylogarithms @xcite as well as nielsen polylogarithms @xcite .
recently it was demonstrated that new types of functions , harmonic polylogarithms @xcite and multiple polylogarithms @xcite , appear in multiloop calculations .
furthermore , the nested sums @xcite were proposed as a generalization of multiple polylogarithms , and the @xmath0-expansion for a large class of hypergeometric function was constructed @xcite . in this paper
we study the _ multiple inverse binomial sums _ defined as ) , but with @xmath7 in the numerator , rather than in the denominator . ]
@xmath8^{i_1}\ldots [ s_{a_p}(j\!-\!1)]^{i_p}\ ; [ s_{b_1}(2j\!-\!1)]^{j_1}\ldots [ s_{b_q}(2j\!-\!1)]^{j_q } , \label{binsum}\ ] ] where @xmath9 is the harmonic sum and @xmath10 will always mean @xmath11 and @xmath12 , respectively , even we do not mention this explicitly . when there are no sums of the type @xmath13 or @xmath14 on the r.h.s . of eq
( [ binsum ] ) , we shall put a `` @xmath15 '' sign instead of the indices @xmath16 or @xmath17 of @xmath18 , respectively . ] and @xmath19 is an arbitrary argument . in what follows , we will also use @xmath20 as the argument of the occurring hypergeometric functions . for sums of the type ( [ binsum ] ) ,
the _ weight _ @xmath21 can be defined as @xmath22 , whereas the _ depth _ can be associated with the sum @xmath23 .
all multiple binomial sums ( [ binsum ] ) can be presented in terms of function @xmath24 and its derivatives by means of the following relation @xmath25 , \qquad k>1,\ ] ] where @xmath26 is the @xmath27-th derivative of the @xmath28-function . in particular , for @xmath29 we have @xmath30 , where @xmath31 is euler s constant .
the sums ( [ binsum ] ) appear in the calculation of massive feynman diagrams within several different approaches : for instance , as solutions of differential equations for feynman amplitudes @xcite , through a nave @xmath0-expansion of hypergeometric functions within mellin
barnes technique @xcite , or in the framework of recently proposed algebraic approach @xcite .
physical applications include the one- , two- and three - loop massive feynman diagrams with two massive cuts @xcite .
the case @xmath32 corresponds to the single - scale propagator - type diagrams @xcite .
although there are many publications concerning harmonic series @xcite , only a limited number of results are available for the inverse binomial sums .
some particular results for @xmath33 can be extracted from @xcite .
the sums with @xmath34 are expressible in terms of an `` odd '' basis @xcite , whereas the case @xmath35 corresponds to an `` even '' basis @xcite .
several new results for a generalization of the sums ( [ binsum ] ) are presented in @xcite .
however , only for special type of sums the analytical results are available @xcite , @xmath36 where @xmath37 , @xmath38 @xmath39 is the generalized log - sine function @xcite .
some examples when such functions occur in the @xmath0-expansion of feynman diagrams can be found in refs .
usually , the occurring angles ( [ def_theta ] ) , possess certain geometrical meaning @xcite .
the main aim of the present publication is the analytical calculation of inverse binomial sums ( [ binsum ] ) , including some relevant examples of physically important feynman diagrams .
the paper is organized as follows . in section 2 , employing the connection with the @xmath0-expansion of hypergeometric functions , we obtain analytical results for sums of the type ( [ binsum ] ) , valid for @xmath40 ( @xmath41 ) .
mainly the sums of the _ weights _ @xmath42 and @xmath43 are considered . in section
3 the analytical continuation to other values of @xmath19 is constructed .
section 4 contains some applications of our results related to the @xmath0-expansion of feynman diagrams , mainly two - loop master integrals . in appendix
a we briefly summarize the relevant properties of the harmonic polylogarithms of complex arguments and related functions . in appendix
b we show how certain identities between hypergeometric functions can be used to establish relations between the corresponding sums . in appendix
c we have collected explicit results for inverse binomial sums of lower weights . in appendix
d relations between binomial , harmonic and inverse binomial sums are explored , and analytical results for multiple binomial sums up to weight * 3 * are presented .
appendix e contains a realistic example of a physical application of the considered integrals .
( remember to change when appendices are ready ! )
here , our analysis is based on comparing two representations of hypergeometric function whose parameters depend on @xmath45 .
one of them is the series representation in terms of the harmonic sums , whereas the second one is the exact result in terms of the functions related to the polylogarithms .
in particular , we consider the @xmath46 hypergeometric function of a special type , @xmath47 where @xmath48 , @xmath49 is the pochhammer symbol , and we have used the duplication formula @xmath50 . to perform the @xmath0-expansion we use the well - known representation @xmath51 \ ; , \label{pochhamer}\ ] ] which yields @xmath52 \nonumber \\ & & + { \varepsilon}^2 \bigl [ 2 b^2 \left ( \bar{s}_2 + \bar{s}_1 ^ 2 \right ) - 2 b ( a_1+b ) s_1 \bar{s}_1 - { { \textstyle{\frac{1}{2 } } } } ( a_2+b^2 ) s_2 + { { \textstyle{\frac{1}{2 } } } } ( a_1+b)^2 s_1 ^ 2 \bigr ] \nonumber \\ & & + { \varepsilon}^3 \bigl [ { { \textstyle{\frac{1}{6 } } } } ( a_1+b ) ^3 s_1 ^ 3 - b ( a_1+b)^2 s_1 ^ 2 \bar{s}_1 - { { \textstyle{\frac{1}{2 } } } } ( a_1+b)(a_2+b^2)s_1 s_2 \nonumber \\ & & \hspace{10 mm } + 2 b^2 ( a_1+b ) s_1 \left ( \bar{s}_2 + \bar{s}_1 ^ 2 \right ) + b ( a_2+b^2 ) s_2 \bar{s}_1 + { { \textstyle{\frac{1}{6 } } } } ( 2 b^3 - a_1 ^ 3 + 3 a_1 a_2 ) s_3 \nonumber \\ & & \hspace{10 mm } - { { \textstyle{\frac{4}{3 } } } } b^3 \left ( 2 \bar{s}_3 + 3 \bar{s}_2
\bar{s}_1 + \bar{s}_1 ^ 3 \right ) \bigr ] + { \cal o } ( { \varepsilon}^4 ) \biggr\ } \ ; , \label{hyper : f21}\end{aligned}\ ] ] where @xmath53 note that the coefficient of the @xmath54 term can be represented as @xmath55 where we have introduced the following combinations of harmonic sums : @xmath56 to completely define the @xmath54 order of the expansion of @xmath44 , we need results for six combinations of sums , according to the number of independent combinations of the parameters @xmath57 and @xmath58 .
let us first consider the case @xmath59 ( @xmath60 ) . in this region the following parametrization can be used : @xmath61 ( @xmath62 ) , where @xmath63 in the rest of this paper , we will use the short - hand notation @xmath64 for a few special cases , the @xmath0-expansion of the @xmath44 functions is known .
first of all , the following relation holds @xcite : @xmath65 its expansion contains only the even powers of @xmath0 . using it
, it is easy to get @xmath66 etc .
then , there are some cases @xcite when an arbitrary term of the @xmath0-expansion can be calculated in terms of log - sine functions , @xmath67 \ ; , \label{basis_ii } \\ _
2f_1 \left(\begin{array}{c| } 1 , 1+{\varepsilon}\\ { { \textstyle{\frac{3}{2}}}}+{\varepsilon}\end{array } ~\sin^2 { { \textstyle{\frac{\theta}{2 } } } } \right ) & = & -\frac{(1 + 2{\varepsilon})}{\sin\theta ( 2 \sin{{\textstyle{\frac{\theta}{2}}}})^{2{\varepsilon } } } \sum_{j=0}^{\infty } \frac{(2{\varepsilon})^j}{j ! } { { \mbox{ls}}_{j+1}\left(\theta\right ) } \ ; , \\ _
2f_1 \left(\begin{array}{c| } 1 , 1+{\varepsilon}\\ { { \textstyle{\frac{3}{2}}}}+{{\textstyle{\frac{1}{2}}}}{\varepsilon}\end{array } ~\sin^2 { { \textstyle{\frac{\theta}{2 } } } } \right ) & = & -\frac{1+{\varepsilon}}{(2\sin\theta)^{1+{\varepsilon } } } \sum_{j=0}^{\infty } \frac{{\varepsilon}^j}{j ! } { { \mbox{ls}}_{j+1}\left(2\theta\right ) } \ ; .\end{aligned}\ ] ] moreover , for a more general case an integral representation can be obtained @xcite , @xmath68 any order of the @xmath0-expansion of eq .
( [ basis_iii ] ) can be expressed in terms of the generalized `` log - sine - cosine '' function @xmath69 whose properties are listed in appendix b of ref .
@xcite , @xmath70 up to the level @xmath71 , only one independent function appears , @xmath72 , that can not be expressed in terms of ordinary log - sine integrals . using the definition ( [ lsc ] )
, one can obtain the following representation : @xmath73 the integral occurring in eq .
( [ lsc23_via_int ] ) can be expressed in terms of the inverse tangent integrals ( see in ref .
@xcite ) , @xmath74 = \frac{1}{2 { \rm i } } \bigl[{{\mbox{li}}_{n}\left ( { \rm i}z\right ) } - { { \mbox{li}}_{n}\left(- { \rm i}z\right)}\bigr ] \ ; , \qquad { { \mbox{ti}}_{n}\left(z\right ) } = \int\limits_0^z \frac{{\rm d}x}{x}\ ; { { \mbox{ti}}_{n-1}\left(x\right ) } \ ; .\ ] ] for @xmath75 , the following integral representation ( see , e.g. , in ref .
@xcite ) is useful : @xmath76 in this way , we arrive at the following representation of @xmath72 in terms of the inverse tangent integrals ( [ ti_n ] ) : @xmath77 note that @xmath78 can be reduced to @xmath72 by using eq .
( a.27 ) of ref .
@xcite . using eqs .
( [ basis_ii])([basis_iii ] ) , we obtain the following results for the weight-*2 * and weight-*3 * sums , in addition to eq .
( [ s2 ] ) : @xmath79 \ ; , \label{s1a } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } \bar{s}_1 & = & \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl [ 2 { { \mbox{ls}}_{2}\left(\pi- \theta\right ) } + { { \mbox{ls}}_{2}\left(\theta\right ) } + \theta l_\theta - 2 \theta l_\theta \bigr ] \ ; , \label{s1bar } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } s_1 ^ 2 & = & 4 \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl [ { { \mbox{ls}}_{3}\left(\pi\!-\!\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } - 2 { { \mbox{ls}}_{2}\left(\pi\!-\!\theta\right ) } l_\theta + \theta l^2_\theta + { { \textstyle{\frac{1}{24 } } } } \theta^3 \bigr ] , \hspace*{7 mm } \label{s1s1 } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } s_1 \bar{s}_1 & = & \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl\ { 5 \left [ { { \mbox{ls}}_{3}\left(\pi\!-\!\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] - { { \mbox{ls}}_{3}\left(\theta\right ) } + { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{3}\left(2\theta\right ) } - 2 { { \mbox{ls}}_{2}\left(\theta\right ) } l_\theta \nonumber \\ & & + 2 { { \mbox{ls}}_{2}\left(\pi\!-\!\theta\right ) } l_\theta - 8 { { \mbox{ls}}_{2}\left(\pi\!-\!\theta\right ) } l_\theta - 2 \theta l_\theta l_\theta + 4 \theta l_\theta^2 + { { \textstyle{\frac{1}{12 } } } } \theta^3 \bigr\ } , \hspace*{7 mm } \label{s1s1bar } \\ \!\
\sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } \left ( \bar{s}_2 + \bar{s}_1 ^ 2 \right ) \!\ !
& = & \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl\ { 6 \left [ { { \mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] - 3 { { \mbox{ls}}_{3}\left(\theta\right ) } + 4 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta \nonumber \\ & & - 8 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta + 2 { { \mbox{ls}}_{2}\left(\theta\right ) } l_\theta + { { \mbox{ls}}_{3}\left(2\theta\right ) } - 4 { { \mbox{ls}}_{2}\left(\theta\right ) } l_\theta \nonumber \\ & & + \theta l^2_\theta - 4 \theta l_\theta l_\theta + 4 \theta l^2_\theta + { { \textstyle{\frac{1}{12 } } } } \theta^3 \bigr\ } \ ; , \label{s2s1bar}\end{aligned}\ ] ] where @xmath80 . in fact ,
all @xmath81 are expressible in terms of the @xmath82-function @xcite . for the weight * 4 *
, the results for some combinations of sums involved in ( [ combinations ] ) can also be expressed in terms of log - sine functions , using eqs .
( [ basis_ii])([basis_iii ] ) , @xmath83 l_\theta \!+\ ! 3 { { \mbox{ls}}_{2}\left(\pi \!-\ ! \theta\right ) } l^2_\theta \!-\ ! \theta l^3_\theta \bigr\ } \ ; , \label{combination:1 } \\ & & \hspace*{-10 mm } \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } { \cal c}_1 = \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl\ { { { \textstyle{\frac{2}{3 } } } } { { \mbox{ls}}_{4}\left(\theta\right ) } - { { \textstyle{\frac{1}{3 } } } } { { \mbox{ls}}_{4}\left ( 2\theta\right ) } - { { \textstyle{\frac{14}{3 } } } } \left [ { { \mbox{ls}}_{4}\left ( \pi - \theta\right ) } - { { \mbox{ls}}_{4}\left ( \pi\right ) } \right ] \nonumber \\ & & + 14 \left [ { { \mbox{ls}}_{3}\left ( \pi - \theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] l_\theta - 2 { { \mbox{ls}}_{3}\left(\theta\right ) } l_\theta + { { \mbox{ls}}_{3}\left(2 \theta\right ) } \left [ l_\theta + l_\theta \right ] - 14 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta^2 \nonumber \\ & & + 2 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta^2 - 2 { { \mbox{ls}}_{2}\left(2 \theta\right ) } l_\theta l_\theta - { { \mbox{ls}}_{2}\left(2 \theta\right ) } l_\theta^2 - 2 \theta l_\theta l_\theta^2 - 2 \theta l_\theta^2 l_\theta + 4 \theta l_\theta^3 \bigr\ } \ ; , \label{combination : c1 } \\ & & \hspace*{-10 mm } \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } { \cal c}_2 = \tan{{\textstyle{\frac{\theta}{2 } } } } \bigl\ { { { \textstyle{\frac{1}{3 } } } } { { \mbox{ls}}_{4}\left(2 \theta\right ) } - { { \mbox{ls}}_{4}\left(\theta\right ) } + 4 \left [ { { \mbox{ls}}_{4}\left(\pi - \theta\right ) } - { { \mbox{ls}}_{4}\left(\pi\right ) } \right ] - { { \mbox{ls}}_{3}\left(2 \theta\right)}\left ( l_\theta + l_\theta \right ) \nonumber \\ & & - 12 \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] l_\theta
+ 3 { { \mbox{ls}}_{3}\left(\theta\right ) } l_\theta + 12 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta^2 - 3 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta^2 \nonumber \\ & & - { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{2}\left(2\theta\right ) } l_\theta^2 + { { \mbox{ls}}_{2}\left(2\theta\right ) } l_\theta^2 + 2 { { \mbox{ls}}_{2}\left(2\theta\right ) } l_\theta l_\theta + 2 \theta l_\theta^2 l_\theta + 2 \theta l_\theta l_\theta^2 - { { \textstyle{\frac{1}{3 } } } } \theta l_\theta^3 - { { \textstyle{\frac{10}{3 } } } } \theta l_\theta^3 \bigr\ } \ ; , \label{combination : c2}\end{aligned}\ ] ] where @xmath84 and the combinations of the harmonic sums @xmath85 are defined in ( [ combinations2 ] ) . finally ,
using ( [ basis_iii ] ) the result for the sum involving @xmath86 can be expressed in terms of the @xmath87-function ( [ lsc - ti ] ) , @xmath88 \nonumber \\ & & + 2 \left [ { { \mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] l_\theta - 8 \left [ { { \mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] l_\theta - { { \mbox{ls}}_{3}\left(2 \theta\right ) } l_\theta + 2 { { \mbox{ls}}_{3}\left(\theta\right ) } l_\theta \nonumber \\ & & + 8 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta^2 - 4 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta l_\theta + { { \mbox{ls}}_{2}\left(2 \theta\right ) } l_\theta^2 - 2 \theta l_\theta^3 + 2 \theta l_\theta^2 l_\theta \bigr\ } \ ; .\end{aligned}\ ] ] some of the results ( [ s1a])([last4 ] ) can be written in a slightly different form by means of relations @xcite @xmath89 however , in this way one can not obtain results for the two remaining combinations in ( [ combinations ] ) , involving @xmath90 , since in eq .
( [ basis_iii ] ) we always have @xmath91 .
an interesting relation between these two functions is obtained in appendix b. it should also be noted that for higher hypergeometric functions other combinations of sums may arise at this level ( see below ) .
let us consider the hypergeometric function of the following type : @xmath92 where @xmath93 . using the representation ( [ pochhamer ] )
its @xmath0-expansion can be reduced to inverse binomial sums ( [ binsum ] ) .
the original hypergeometric function ( [ pfq ] ) can be written as ( see details in appendix b of ref .
@xcite ) @xmath94 where @xmath48 , @xmath95 \nonumber \\ & = & 1 - { \varepsilon}\left ( w_1 j^{-1 } + t_1 s_1 + 2 u_1 \bar{s}_1 \right ) + { \varepsilon}^2 \bigl [ { { \textstyle{\frac{1}{2 } } } } j^{-2}\left ( w_2 + w_1 ^ 2 \right ) + w_1 j^{-1 } \left ( t_1 s_1 + 2 u_1 \bar{s}_1 \right ) \nonumber \\ & & + 2 t_1 u_1 s_1 \bar{s}_1 + { { \textstyle{\frac{1}{2 } } } } t_2 s_2 + { { \textstyle{\frac{1}{2 } } } } t_1 ^ 2 s_1 ^ 2 + 2 \left ( u_2 \bar{s}_2 + u_1 ^ 2 \bar{s}_1 ^ 2 \right ) \bigr ] \nonumber \\ & & - { \varepsilon}^3 \bigl\ { { { \textstyle{\frac{1}{6 } } } } j^{-3 } \left(2 w_3 + 3 w_1 w_2 + w_1 ^ 3 \right ) + { { \textstyle{\frac{1}{2 } } } } j^{-2 } \left ( w_2 + w_1 ^ 2\right ) \left ( t_1 s_1 + 2 u_1 \bar{s}_1 \right ) \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } w_1 j^{-1 } \left [ t_2 s_2 + t_1 ^ 2 s_1 ^ 2 + 4 t_1 u_1 s_1 \bar{s}_1 + 4 \left ( u_2 \bar{s}_2 + u_1 ^ 2 \bar{s}_1 ^ 2 \right ) \right ] \nonumber \\ & & + { { \textstyle{\frac{1}{6 } } } } t_1 ^ 3 s_1 ^ 3 + t_1 ^ 2 u_1 s_1 ^ 2 \bar{s}_1 + { { \textstyle{\frac{1}{2 } } } } t_1 t_2 s_1 s_2 + t_2 u_1 s_2 \bar{s}_1 + { { \textstyle{\frac{1}{3 } } } } t_3 s_3 \nonumber \\ & & + 2 t_1 s_1 \left ( u_2 \bar{s}_2 + u_1 ^ 2 \bar{s}_1 ^ 2 \right ) + { { \textstyle{\frac{4}{3 } } } } \left ( u_1 ^ 3 \bar{s}_1 ^ 3 + 3 u_1 u_2 \bar{s}_1 \bar{s}_2 + 2 u_3 \bar{s}_3 \right ) \bigr\ } + { \cal o}({\varepsilon}^4)\ ; , \label{expansion}\end{aligned}\ ] ] and we introduced the constants @xmath96 where the summations extend over all possible values of the parameters in eq .
( [ pfq ] ) .
we can see that for the general values of the parameters of the @xmath97 function ( [ pfq ] ) we need all the occurring sums separately . however , for many applications , it is sufficient to consider the case @xmath98 only . in this case ,
@xmath99 and @xmath100 . substituting this into eq .
( [ expansion ] ) , we see that @xmath101 and @xmath102 would only appear in combinations @xmath103 and @xmath104 . furthermore , using the notations ( [ combinations2 ] ) , eq .
( [ expansion ] ) in the case @xmath98 can be presented as @xmath105 \nonumber \\ & & - { \varepsilon}^3 \bigl\ { { { \textstyle{\frac{1}{6 } } } } j^{-3 } \left(2 w_3 + 3 w_1 w_2 + w_1 ^ 3 \right ) + { { \textstyle{\frac{1}{2 } } } } j^{-2 } \left ( w_2 + w_1 ^
2\right ) \left ( t_1 s_1 + 2 f_1 \bar{s}_1 \right ) \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } w_1 j^{-1 } \left [ t_2
s_2 + t_1 ^ 2 s_1 ^ 2 + 4 f_1 t_1 s_1 \bar{s}_1 + 4 f_1 ^ 2 \left ( \bar{s}_2 + \bar{s}_1 ^ 2 \right ) \right ] \nonumber \\ & & + { { \textstyle{\frac{1}{6 } } } } t_1 \left ( t_1 ^ 2 + 3 f_1 t_1 + 3 f_1 ^ 2 \right ) s_1 ^ 3 + { { \textstyle{\frac{1}{2 } } } } t_1 \left ( t_2 - f_1 t_1 -f_1 ^ 2 \right ) s_1 s_2 + { { \textstyle{\frac{1}{3 } } } } t_3 s_3 \nonumber \\ & & + f_1 \left ( t_2 + t_1 ^ 2 + 2 f_1 t_1 + 2 f_1 ^ 2 \right ) s_2 \bar{s}_1 + 2 f_1 ^ 2 t_1 { \cal c}_1 - 4 f_1 ^ 3 { \cal c}_2 \nonumber \\ & & - { { \textstyle{\frac{1}{2 } } } } f_1 ( t_1 + 2 f_1 ) ( t_1 + 3 f_1 ) { \cal c}_3 \bigr\ } + { \cal o}({\varepsilon}^4)\ ; . \label{expansion2}\end{aligned}\ ] ] eq .
( [ expansion ] ) also shows that we may need the sums with higher values of @xmath106 ( the power of @xmath107 ) , which would come from @xmath108 in eq .
( [ pfq2 ] ) and extra powers of @xmath107 in eq .
( [ expansion ] ) .
note that in the case @xmath109 , when we have the same number of @xmath110 and @xmath111 , we have @xmath107 factor in the sum in eq .
( [ pfq2 ] ) . moreover , when the parameters @xmath112 and @xmath113 are at all missing , we get no extra factors of @xmath107 in eq .
( [ expansion ] ) , for all @xmath114 would vanish . for the sums of the type ( [ binsum ] ) with an arbitrary integer power @xmath115 and @xmath116 , the following one - fold
integral representation @xcite is useful : @xmath117^{c-2 } \sum_{j=1}^\infty \frac{\left ( 4 \sin^2{{\textstyle{\frac{\phi}{2 } } } } \right)^{j}}{j } \frac{f(j)}{\left ( 2j\atop j\right ) } \ ; , \label{higher_c}\end{aligned}\ ] ] where @xmath118 stands for an arbitrary combination of the harmonic sums . in this manner ,
the original sum of the type ( [ binsum ] ) with an arbitrary positive integer parameter @xmath115 is reduced to a one - fold integral representation containing a sum with @xmath119 in the integrand . using the representation ( [ higher_c ] ) we can generalize our results ( [ s2 ] ) , ( [ s4 ] ) , ( [ s1a])([last4 ] ) to the case of an arbitrary integer @xmath120 .
however , only for special types of sums these results are expressible in terms of generalized log - sine functions ( [ def_ls ] ) : @xmath121 where @xmath37 , @xmath116 , and the angle @xmath122 is defined in eq .
( [ def_theta ] ) .
further results can be extracted from integral representation ( [ higher_c ] ) by using the integration rules for clausen s function @xcite and the following relation : @xmath123 in this way , we obtain @xmath124 - 4\left [ { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } \right]^2 + { { \textstyle{\frac{1}{24 } } } } \theta^4 \ ; , \label{s1s1c2 } \\
\sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{j^2 } s_1 \bar{s}_1 & = & 5 \theta \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] - \theta { { \mbox{ls}}_{3}\left(\theta\right ) } + { { \textstyle{\frac{1}{2 } } } } \theta { { \mbox{ls}}_{3}\left(2 \theta\right ) } \nonumber \\ & & - 4 \left[{{\mbox{ls}}_{2}\left(\pi-\theta\right ) } \right]^2 + { { \textstyle{\frac{1}{48 } } } } \theta^4 - 2 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } { { \mbox{ls}}_{2}\left(\theta\right ) } \ ; , \label{s1s1barc2 } \\
\sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{j^2 } \left ( \bar{s}_1 ^ 2 + \bar{s}_2 \right ) & = & 6 \theta \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] - 3 \theta { { \mbox{ls}}_{3}\left(\theta\right ) } + \theta { { \mbox{ls}}_{3}\left(2 \theta\right ) } \nonumber \\ & & - 4 \left[{{\mbox{ls}}_{2}\left(\pi-\theta\right ) } \right]^2 - \left[{{\mbox{ls}}_{2}\left(\theta\right ) } \right]^2 + { { \textstyle{\frac{1}{48 } } } } \theta^4 - 4 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } { { \mbox{ls}}_{2}\left(\theta\right ) } \ ; . \label{s2bar_2}\end{aligned}\ ] ] however , other sums are not expressible in terms of the generalized log - sine functions .
one needs to introduce a new function one may introduce the generalized glashier function @xmath125 , see eq .
( [ gl4(theta,1 ) ] ) in appendix a. ] , @xmath126 which obeys the following symmetry property : @xmath127 where @xmath128 the following sums of the weight * 4 * are expressible in terms of the function @xmath129 : @xmath130 - { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{4}^{(1)}\left(2 \theta\right ) } + 2 { { \mbox{ls}}_{4}^{(1)}\left(\theta\right ) } - 4 \theta { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta \nonumber \\ & & - 2 \pi \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] + 8 \left [ { { \mbox{cl}}_{3}\left(\pi-\theta\right ) } - { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] l_\theta + 4 \phi(\pi-\theta ) - 4 \phi(\pi ) \ ; , \hspace*{6 mm } \label{s1_3 } \\ & & \hspace*{-10 mm } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{j^3 } \bar{s}_1 = 2 \left [ { { \mbox{ls}}_{4}^{(1)}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{4}^{(1)}\left(\pi\right ) } \right ] - { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{4}^{(1)}\left(2 \theta\right ) } + 4 { { \mbox{ls}}_{4}^{(1)}\left(\theta\right ) } + \left [ { { \mbox{ls}}_{2}\left(\theta\right ) } \right]^2 \nonumber \\ & & - 2 \pi \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] + 8 \left [ { { \mbox{cl}}_{3}\left(\pi-\theta\right ) } - { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] l_\theta - 4 \left [ { { \mbox{cl}}_{3}\left(\theta\right ) } - \zeta_3 \right ] l_\theta \nonumber \\ & & - 2 \theta { { \mbox{ls}}_{2}\left(\theta\right ) } l_\theta - 4 \theta { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta + 4 \phi(\pi-\theta ) - 4 \phi(\pi ) \ ; , \label{s1bar_3}\end{aligned}\ ] ] where we have also used the following integral : @xmath131 \nonumber \\ & & - { { \textstyle{\frac{1}{2}}}}\pi \left[{{\mbox{ls}}_{3}\left(\pi - \theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] \ ; . \nonumber \end{aligned}\ ] ]
let us note that @xmath129 can be related to the real part of a certain harmonic polylogarithm @xcite of complex argument , @xmath132 , \ ] ] where ( for details , see appendix a ) @xmath133 the function @xmath134 has a branch cut starting at the point @xmath135 ( @xmath136 ) , which is subtracted by logarithmic terms in ( [ phih ] ) , so that @xmath129 is a smooth function of variable @xmath122 . for further investigation of the sums of the type ( [ binsum ] ) , a recursive approach appears to be useful .
some ideas applied below are described in the book @xcite .
let us rewrite eq .
( [ binsum ] ) in the following form : @xmath137 where @xmath138 and @xmath139 denote the collective sets of indices , whereas @xmath140 is the coefficient of @xmath141 in eq .
( [ binsum ] ) , i.e. the product of binomial sums divided by @xmath142 .
the idea is to find a recurrence relation , with respect to @xmath143 , for the coefficients @xmath140 , and then transform it into a differential equation for the _ generating _ function @xmath144 . in this way
, the problem of summing the series would be reduced to solving a differential equation . using the explicit form of @xmath140 given in eq .
( [ binsum ] ) , the recurrence relation can be written in the following form : @xmath145 where the explicit form of the `` remainder '' @xmath146 is given by @xmath147^{i_k } \prod\limits_{l=1}^q \left [ s_{b_l}(2j-1 ) + ( 2j)^{-b_l } + ( 2j+1)^{-b_l } \right]^{j_l } \nonumber \\ & & - \prod\limits_{k=1}^p \left [ s_{a_k}(j-1 ) \right]^{i_k } \prod\limits_{l=1}^q \left [ s_{b_l}(2j-1 ) \right]^{j_l } \ ; . \label{rab}\end{aligned}\ ] ] in other words , it contains all contributions generated by @xmath148 , @xmath149 and @xmath150 which appear because of the shift of the index @xmath143 . multiplying both sides of eq .
( [ rec : relation ] ) by @xmath141 , summing from 1 to infinity , and using the fact that any extra power of @xmath143 corresponds to the derivative @xmath151 , we arrive at the following differential equation for the generating function @xmath144 : @xmath152 where @xmath153 and @xmath154 . using eq .
( [ rab ] ) we obtain @xmath155 \ ; , \label{r_b1 } \\
r_{a_1,a_2;-}(u ) & = & \sum_{j=1}^\infty \frac{u^j}{\left ( 2j \atop j\right ) } \biggl [ \frac{s_{a_1}}{j^{a_2 } } + \frac{s_{a_2}}{j^{a_1 } } + \frac{1}{j^{a_1+a_2 } } \biggr ] \ ; , \label{r_a1a2 } \\
r_{a_1;b_1}(u ) & = & \sum_{j=1}^\infty \frac{u^j}{\left ( 2j \atop j\right ) } \biggl [ \frac{s_{a_1}}{(2j)^{b_1 } } \!+\ !
\frac{s_{a_1}}{(2j+1)^{b_1 } } \!+\ ! \frac{\bar{s}_{b_1}}{j^{a_1 } } \!+\ ! \frac{1}{j^{a_1}(2j)^{b_1 } } \!+\ ! \frac{1}{j^{a_1}(2j+1)^{b_1 } } \biggr ] \ ; , \label{r_a1b1 } \\
r_{-;b_1,b_2}(u ) & = & \sum_{j=1}^\infty \frac{u^j}{\left ( 2j \atop j\right ) } \biggl [ \frac{\bar{s}_{b_1}}{(2j)^{b_2 } } \!+\ ! \frac{\bar{s}_{b_2}}{(2j)^{b_1 } } \!+\ ! \frac{\bar{s}_{b_1}}{(2j+1)^{b_2 } } \!+\ ! \frac{\bar{s}_{b_2}}{(2j+1)^{b_1 } } \!+\ ! \frac{1}{(2j)^{b_1+b_2 } } \nonumber \\ & & \hspace{20 mm } \!+\ ! \frac{1}{j^{b_1}(2j+1)^{b_2 } } \!+\ ! \frac{1}{j^{b_2}(2j+1)^{b_1 } } \!+\ ! \frac{1}{(2j+1)^{b_1+b_2 } } \biggr ] \ ; , \label{r_b1b2 } \\
r_{a_1,a_2,a_3;-}(u ) & = & \sum_{j=1}^\infty \frac{u^j}{\left ( 2j \atop j\right ) } \biggl [ \frac{s_{a_1}s_{a_2}}{j^{a_3 } } \!+\ ! \frac{s_{a_1}s_{a_3}}{j^{a_2 } } \!+\ ! \frac{s_{a_2}s_{a_3}}{j^{a_1 } } \!+\ ! \frac{s_{a_1}}{j^{a_2+a_3 } } \!+\ ! \frac{s_{a_2}}{j^{a_1+a_3 } } \!+\ ! \frac{s_{a_3}}{j^{a_1+a_2 } } \!+\ ! \frac{1}{j^{a_1+a_2+a_3 } } \biggr ] .
\nonumber \label{r_a1a2a3 } \\
\end{aligned}\ ] ] in terms of the geometrical variable ( [ def_theta ] ) , @xmath156 the differential equation ( [ diff : i ] ) takes the following form : @xmath157 furthermore , eq .
( [ diff : ii ] ) can be represented as @xmath158 where @xmath159 with @xmath160 .
in particular , for @xmath119 and @xmath161 we have , respectively , @xmath162 \ ; , \label{uniq } \\ \sigma_{a;b;2}(u_{\theta } ) & = & \int\limits_0^{\theta } { \rm d}\phi \sigma_{a;b}(\phi ) = \theta \sigma_{a;b}(\theta ) - { { \textstyle{\frac{1}{2}}}}\delta_{p0}\theta^2 - \int\limits_0^{\theta } \phi { \rm d}\phi\ ; r_{a , b}(u_{\phi})\ ; .\end{aligned}\ ] ] introducing @xmath163 , eq . ( [ basisia ] )
can be rewritten as @xmath164 which is nothing but the differential form of the relation ( [ higher_c ] ) .
the iterative solution of eq .
( [ basisib ] ) is @xmath165 where @xmath166 .
in particular , this solution allows us to formulate and prove the following important _
statement_. if for some @xmath27 the derivative @xmath167 is expressible only in terms of the powers of @xmath122 and @xmath168 , then the sum @xmath169 can be presented in terms of the generalized log - sine functions . moreover , according to a statement proven in appendix a.1 of ref .
@xcite , the analytic continuation of any generalized log - sine function @xmath170 can be expressed in terms of nielsen polylogarithms .
let us now explain how this general method works for specific sums of interest .
consider @xmath171 where @xmath172 is defined in eq .
( [ r_a1 ] ) and the result can be extracted from eq .
( [ kv ] ) : @xmath173 + \theta^2 l_{\theta } \ ; , \label{j^3}\ ] ] integrating over @xmath174 we obtain @xmath175 for another sum , @xmath176 , we get @xmath177 where @xmath178 is defined in eq .
( [ r_a1a2 ] ) . using eqs .
( [ s1c2 ] ) , ( [ s2 ] ) and ( [ j^3 ] ) to calculate @xmath178 , and integrating over @xmath174 , we get @xmath179 for the sum @xmath180 , this approach is also applicable . in this case
we have @xmath181 with @xmath182 defined in eq .
( [ r_a1a2a3 ] ) . using eqs .
( [ s1s1 ] ) , ( [ s1c2 ] ) and ( [ j^3 ] ) to calculate @xmath182 , and integrating over @xmath174 , we get @xmath183 + 8 \left [ { { \mbox{ls}}_{4}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{4}\left(\pi\right ) } \right ] - \theta^2 { { \mbox{cl}}_{2}\left(\theta\right ) }
\nonumber \\ & & + 3 \theta^2 { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } + 24 { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } l^2_\theta - \theta^3 l_\theta - 8 \theta l^3_\theta + 7 \zeta_3 \theta \bigr\ } \ ; .
\label{s1s1s1c1}\end{aligned}\ ] ] combining eqs .
( [ s3c1 ] ) , ( [ s1s2c1 ] ) and ( [ s1s1s1c1 ] ) , we successfully reproduce eq ( [ combination:1 ] ) . as a further example ,
let us consider the sum @xmath184 .
we obtain @xmath185 where @xmath186 corresponds to eq .
( [ r_a1b1 ] ) , with some of the contributing sums having @xmath187 in the denominator .
let us denote them as @xmath188 to calculate such sums , it is convenient to apply the differential operator that lowers the value of @xmath189 , @xmath190 \leftrightarrow \frac{2}{\cos{{{\textstyle{\frac{\theta}{2 } } } } } } \frac{{\rm d}}{{\rm d}\theta } \left [ \sin{{{\textstyle{\frac{\theta}{2 } } } } } \ ; { \widetilde{\sigma}}_{a;b;c;d}(u_{\theta } ) \right ] = { \widetilde{\sigma}}_{a;b;c;d-1}(u_{\theta } ) \ ; . \label{diff_2j}\ ] ] in particular , for @xmath191 the sum on the r.h.s .
is the standard sum ( [ binsum ] ) which is supposed to be known .
then the result for @xmath192 can be obtained by integration , @xmath193 using the following decomposition @xmath194 we are able to rewrite @xmath195 as a linear combination of the sums , with one of the last two indices equal to zero : @xmath196 for the calculation of the @xmath197 ( with @xmath198 ) we will apply d - times the procedure ( [ diff_2j ] ) @xmath199 = \left ( \tan{{\textstyle{\frac{\theta}{2 } } } } \frac{{\rm d}}{{\rm d}\theta}\right ) { \sigma}_{a;b;1;0}(u_{\theta } ) \ ; .\ ] ] this equation can be easily integrated for some particular cases @xmath200 - 2 \ ; , \\
\sum_{j=1}^{\infty } \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{(2j+1 ) } s_2 & = & \frac{\theta^3}{6\sin \theta } - 2 \theta \cot{{{\textstyle{\frac{\theta}{2 } } } } } - { { \textstyle{\frac{1}{2 } } } } \theta^2 + 4 \ ; , \\
\sum_{j=1}^{\infty } \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{(2j+1 ) } \bar{s}_1 & = & \frac{1}{\sin \theta } \left [ 2 { { \mbox{ls}}_{2}\left(\pi\!-\!\theta\right ) } \!-\ ! 2 \theta l_\theta \!+\ ! { { \mbox{ls}}_{2}\left(\theta\right ) } \!+\ !
\theta l_\theta \right ] \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \theta \cot{{{\textstyle{\frac{\theta}{2 } } } } } - \frac{2}{\sin{{{\textstyle{\frac{\theta}{2 } } } } } } \ ; { { \mbox{ti}}_{2}\left(\tan{{{\textstyle{\frac{\theta}{4}}}}}\right ) } - 1 \ ; , \\
\sum_{j=1}^{\infty } \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{(2j+1 ) } s_1 ^ 2 & = & - { { \textstyle{\frac{1}{2 } } } } \theta^2 + 4 \cot { { \textstyle{\frac{\theta}{2 } } } } \left [ { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } - \theta l_\theta
\right ] - 4 + 2 \theta \cot { { \textstyle{\frac{\theta}{2 } } } } + \frac{\theta^3}{6\sin \theta } \nonumber \\ & & + \frac{4}{\sin \theta } \left [ { { \mbox{ls}}_{3}\left(\pi \!-\ ! \theta\right ) } \!-\ ! { { \mbox{ls}}_{3}\left(\pi\right ) } \!+\ !
\theta l^2_\theta \!-\ ! 2 { { \mbox{ls}}_{2}\left(\pi-\theta\right ) } l_\theta \right ] \ ; , \\ \sum_{j=1}^{\infty } \frac{1}{\left ( 2j \atop j\right ) } \frac{u^j}{(2j+1 ) } s_1 s_2 & = & 8 \!-\ !
\theta^2 l_\theta \!+\ ! 2 \theta { { \mbox{cl}}_{2}\left(\pi\!-\!\theta\right ) } \!-\ ! 2 \theta { { \mbox{cl}}_{2}\left(\theta\right ) } \!-\ ! 4 { { \mbox{cl}}_{3}\left(\pi\!-\!\theta\right ) } \!-\ ! 2 { { \mbox{cl}}_{3}\left(\theta\right ) } \!-\ ! \zeta_3 \nonumber \\ & & + \cot { { \textstyle{\frac{\theta}{2 } } } } \left [ { { \textstyle{\frac{1}{6 } } } } \theta^3 \!-\ ! 4 \theta \!+\ ! 4 \theta l_\theta \!-\ ! 4 { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } \right ] \nonumber \\ & & + \frac{1}{\sin \theta } \bigl\ { \theta^2 \left [ { { \mbox{cl}}_{2}\left(\pi \!-\ ! \theta\right ) } \!-\ ! { { \mbox{cl}}_{2}\left(\theta\right ) } \right ] \!-\ ! 4 \theta \left [ { { \mbox{cl}}_{3}\left(\pi \!-\ ! \theta\right ) } + { { \mbox{cl}}_{3}\left(\theta\right ) } \right ] \nonumber \\ & & \hspace{15 mm } - 8 { { \mbox{cl}}_{4}\left(\pi \!-\ ! \theta\right ) } \!+\ ! 6 { { \mbox{cl}}_{4}\left(\theta\right ) } \!+\ ! \theta \zeta_3 \!-\ ! { { \textstyle{\frac{1}{3 } } } } \theta^3 l_\theta \bigr\ } \ ; , \end{aligned}\ ] ] where @xmath75 is the inverse tangent integral ( [ ti_n ] ) . together with eqs .
( [ s2 ] ) and ( [ s1barc2 ] ) , this provides us the result for @xmath186 , @xmath201 finally , using eq .
( [ uniq ] ) and integrating over @xmath174 , we arrive at @xmath202 using the results for the sums with @xmath119 and applying eq .
( [ higher_c ] ) , we obtain the following results for the case @xmath161 : @xmath203 we note that the results for other sums obtained in the previous sections can be also reproduced by this method .
moreover , it is possible to consider not only the combinations @xmath204 that correspond to the @xmath98 case of eq .
( [ expansion ] ) ( given in eq .
( [ expansion2 ] ) ) , but also the sums containing @xmath101 or @xmath205 separately ( which appear for @xmath206 ) . as an example , let us consider the sum @xmath207 \ ; , \label{halfangle1}\ ] ] where @xmath208 corresponds to the case ( [ r_b1 ] ) .
substituting @xmath209 into eq .
( [ halfangle1 ] ) and integrating over @xmath174 , we arrive at @xmath210 moreover , upon applying eq .
( [ higher_c ] ) we can perform another integration and obtain the result for @xmath161 , @xmath211 ^ 2 + { { \textstyle{\frac{1}{96}}}}\theta^4 \ ; .\ ] ] using the connections between @xmath212 , @xmath213 and the log - sine functions ( see eq .
( 17 ) on p. 292 and eq .
( 44 ) on p. 298 of ref .
@xcite ) , we can get other representations , @xmath214 \nonumber \\ & & + 2 \left [ { { \mbox{ls}}_{3}\left(\pi - { { \textstyle{\frac{\theta}{2}}}}\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] - 2 { { \mbox{ls}}_{3}\left({{\textstyle{\frac{\theta}{2}}}}\right ) } + { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{3}\left(\theta\right ) } \bigr \ } \ ; , \label{s2barc1 } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } \bar{s}_2 & \!\!=\!\ ! & 2 \theta \left [ { { \mbox{ls}}_{3}\left(\pi-{{\textstyle{\frac{\theta}{2}}}}\right)}-{{\mbox{ls}}_{3}\left({{\textstyle{\frac{\theta}{2}}}}\right ) } \right ] + { { \textstyle{\frac{1}{2 } } } } \theta { { \mbox{ls}}_{3}\left(\theta\right ) } -2 \left [ { { \mbox{ls}}_{2}\left(\pi-{{\textstyle{\frac{\theta}{2}}}}\right)}+{{\mbox{ls}}_{2}\left({{\textstyle{\frac{\theta}{2}}}}\right ) } \right]^2 \nonumber \\ & & + { { \textstyle{\frac{1}{6}}}}\pi^3 \theta + { { \textstyle{\frac{1}{96}}}}\theta^4 \ ; . \label{s2barc2}\end{aligned}\ ] ] note the appearance of log - sine functions of arguments @xmath215 and @xmath216 .
to obtain results valid in other regions of variable @xmath19 ( for @xmath217 and @xmath218 ) , we will construct the proper analytical continuation of the expressions presented in the previous section . for generalized log - sine integrals
it is described in @xcite .
let us introduce a new variable @xmath219 where the choice of the sign @xmath220 is related to the causal `` + i0 '' prescription for the propagators . for completeness
, we also present the inverse relations , @xmath221 and also expressions for @xmath222 in terms of @xmath122 , @xmath223 in terms of this variable @xmath224 , the analytic continuation of all generalized log - sine integrals can be expressed in terms of nielsen polylogarithms , whereas for the function @xmath129 we get @xmath225 \nonumber \\ & & - { { \textstyle{\frac{1}{4 } } } } \left[{{\mbox{li}}_{4}\left(y\right ) } + { { \mbox{li}}_{4}\left(y^{-1}\right ) } \right ] - { { \textstyle{\frac{1}{2 } } } } \left [ \ln(1+y ) - { { \textstyle{\frac{1}{2 } } } } \ln y \right ] \left[{{\mbox{li}}_{3}\left(y\right ) } + { { \mbox{li}}_{3}\left(y^{-1}\right ) } \right ] \;,\end{aligned}\ ] ] with @xmath134 defined in eq . ( [ h(-1001 ) ] )
. for the cases involving the inverse tangent integrals @xmath226 and @xmath227 , the analytic continuation is straightforward ( see eq .
( [ ti_n ] ) ) , @xmath228 \ ; , \qquad \;\ ; \omega = \frac{1-y}{1+y } = - { \rm i } \sigma \tan{{\textstyle{\frac{\theta}{2 } } } } \;.
\label{def_omega } \\ { { \mbox{ti}}_{n}\left(\tan{{\textstyle{\frac{\theta}{4}}}}\right ) } & = & - \frac{\sigma}{2 { \rm i } } \left [ { { \mbox{li}}_{n}\left(\omega_s\right ) } - { { \mbox{li}}_{n}\left ( - \omega_s\right ) } \right ] \ ; , \qquad \omega_s = \frac{1-\sqrt{y}}{1+\sqrt{y } } = - { \rm i } \sigma \tan{{\textstyle{\frac{\theta}{4 } } } } \;. \label{def_omega_s}\end{aligned}\ ] ] below we list the most complicated results , corresponding to the analytical continuation of the results obtained in section 2 : @xmath229 ^ 2 \nonumber \\ & & \hspace{20 mm } - { { \textstyle{\frac{1}{24 } } } } \ln^4 y + 4 \zeta_2 { { \mbox{li}}_{2}\left(-y\right ) } + \zeta_2 \ln^2 y + 4 \zeta_3 \ln y + { { \textstyle{\frac{5}{2 } } } } \zeta_4 \ ; , \\ & & \hspace*{-10 mm } \label{an_s1sb1_2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_1 \bar{s}_1 = - 10 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y + { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln y - 2 { { \mbox{s}}_{1,2\!}\left(y\right ) } \ln y + 4 \left[{{\mbox{li}}_{2}\left(-y\right ) } \right]^2 \nonumber \\ & & - 2 { { \mbox{li}}_{2}\left(y\right ) } { { \mbox{li}}_{2}\left(-y\right ) } + 3 { { \mbox{li}}_{3}\left(-y\right ) } \ln y - { { \mbox{li}}_{3}\left(y\right ) } \ln y - { { \textstyle{\frac{3}{2 } } } } { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 y + { { \textstyle{\frac{1}{2 } } } } { { \mbox{li}}_{2}\left(y\right ) } \ln^2 y \nonumber \\ & & - { { \textstyle{\frac{1}{48 } } } } \ln^4 y + 6 \zeta_2 { { \mbox{li}}_{2}\left(-y\right ) } - \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } + { { \textstyle{\frac{5}{4 } } } } \zeta_2 \ln^2 y + { { \textstyle{\frac{11}{2 } } } } \zeta_3 \ln y + 5 \zeta_4 \ ; , \\ & & \hspace*{-10 mm } \label{an_sb2_2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } \left ( \bar{s}_2 + \bar{s}_1 ^ 2 \right ) = - 12 { { \mbox{s}}_{1,2\!}\left ( -y\right ) } \ln y + 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln y - 6 { { \mbox{s}}_{1,2\!}\left(y\right ) } \ln y + 4 \left[{{\mbox{li}}_{2}\left(-y\right ) } \right]^2 \nonumber \\ & & + \left[{{\mbox{li}}_{2}\left(y\right ) } \right]^2 - 4 { { \mbox{li}}_{2}\left(-y\right ) } { { \mbox{li}}_{2}\left(y\right ) } + 2 { { \mbox{li}}_{3}\left(-y\right ) } \ln y - { { \mbox{li}}_{3}\left(y\right ) } \ln y - { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 y \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } { { \mbox{li}}_{2}\left(y\right ) } \ln^2 y + 8 \zeta_2 { { \mbox{li}}_{2}\left(-y\right ) } - 4 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } + \zeta_2 \ln^2 y + 8 \zeta_3 \ln y + 10 \zeta_4 \ ; , \\ & & \hspace*{-10 mm } \label{an_c1 } \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } { \cal c}_1 = \frac{1-y}{1+y } \bigl [ 3 { { \mbox{li}}_{4}\left(-y\right ) } - 3 { { \mbox{li}}_{4}\left(y\right ) } + 28 { { \mbox{s}}_{1,3\!}\left(-y\right ) } - 2 { { \mbox{s}}_{1,3\!}\left(y^2\right ) } + 4 { { \mbox{s}}_{1,3\!}\left(y\right ) } \nonumber \\ & & - 14 { { \mbox{s}}_{2,2\!}\left(-y\right ) } + { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 2{{\mbox{s}}_{2,2\!}\left(y\right ) } + 28 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln(1+y ) - 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln(1-y ) \nonumber \\ & & - 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln(1+y ) + 4 { { \mbox{s}}_{1,2\!}\left(y\right ) } \ln(1-y ) + 4 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1-y ) + 2 { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1-y ) \nonumber \\ & & - 10 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1+y ) + 4 { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1+y ) - 2 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 ( 1-y ) \nonumber \\ & & - 2 { { \mbox{li}}_{2}\left(y^2\right ) } \ln ( 1+y ) \ln ( 1-y ) + 12 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 ( 1+y ) - 2 { { \mbox{li}}_{2}\left(y\right ) } \ln^2 ( 1+y ) \nonumber \\ & & - 2 \ln y \ln^2 ( 1-y ) \ln ( 1+y ) - 2 \ln y \ln ( 1-y ) \ln^2 ( 1+y ) + 4 \ln y \ln^3 ( 1+y ) \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \ln^2 y \ln^2 ( 1-y ) + 2 \ln^2 y \ln ( 1-y ) \ln ( 1+y ) - { { \textstyle{\frac{5}{2 } } } } \ln^2 y \ln^2 ( 1+y ) - { { \textstyle{\frac{1}{2 } } } } \ln^3 y \ln ( 1-y ) \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \ln^3 y \ln ( 1+y ) + { { \textstyle{\frac{45}{8 } } } } \zeta_4 - \zeta_3 \ln ( 1-y ) - 13 \zeta_3 \ln ( 1+y ) + 7 \zeta_3 \ln y + { { \textstyle{\frac{9}{4 } } } } \zeta_2 \ln^2 y \nonumber \\ & & - \zeta_2 \ln^2 ( 1-y ) + 2 \zeta_2 \ln ( 1-y ) \ln(1+y ) + 8 \zeta_2 \ln^2 ( 1+y ) - 9 \zeta_2 \ln y \ln ( 1+y ) \bigr ] \ ; , \\ & & \hspace*{-10 mm } \label{an_c2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j\atop j\right ) } \frac{u^j}{j } { \cal c}_2 = \frac{1-y}{1+y } \bigl [ 2 { { \mbox{s}}_{1,3\!}\left(y^2\right ) } - 24 { { \mbox{s}}_{1,3\!}\left(-y\right ) } - 6 { { \mbox{s}}_{1,3\!}\left(y\right ) } + 12 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - { { \mbox{s}}_{2,2\!}\left(y^2\right ) } \nonumber \\ & & + 3 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 2 { { \mbox{li}}_{4}\left(-y\right ) } + { { \textstyle{\frac{5}{2 } } } } { { \mbox{li}}_{4}\left(y\right ) } - 24 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln(1+y ) - 6 { { \mbox{s}}_{1,2\!}\left(y\right ) } \ln ( 1-y ) \nonumber \\ & & - { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1-y ) + 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln ( 1-y ) + 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln ( 1+y ) - 4 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1-y ) \nonumber \\ & & + 8 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1+y ) - 4 { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1+y ) + 2 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 ( 1-y ) - { { \mbox{li}}_{2}\left(y\right ) } \ln^2 ( 1-y ) \nonumber \\ & & + 4 { { \mbox{li}}_{2}\left(-y\right ) } \ln ( 1-y ) \ln ( 1+y ) + 4 { { \mbox{li}}_{2}\left(y\right ) } \ln ( 1-y ) \ln ( 1+y ) - 10 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 ( 1+y ) \nonumber \\ & & + 2 { { \mbox{li}}_{2}\left(y\right ) } \ln^2 ( 1+y ) - { { \textstyle{\frac{1}{3 } } } } \ln y \ln^3 ( 1-y ) + 2 \ln y \ln^2 ( 1-y ) \ln ( 1+y ) \nonumber \\ & & + 2 \ln y \ln ( 1-y ) \ln^2 ( 1+y ) - { { \textstyle{\frac{10}{3 } } } } \ln y \ln^3 ( 1+y ) - { { \textstyle{\frac{1}{4 } } } } \ln^2 y \ln ^2 ( 1-y ) \nonumber \\ & & - 2 \ln^2 y \ln ( 1-y ) \ln ( 1+y ) + 2 \ln^2 y \ln^2 ( 1+y ) + { { \textstyle{\frac{5}{12 } } } } \ln^3 y \ln ( 1-y ) - { { \textstyle{\frac{1}{3 } } } } \ln^3 y \ln ( 1+y ) \nonumber \\ & & - { { \textstyle{\frac{1}{96 } } } } \ln^4 y - { { \textstyle{\frac{9}{4 } } } } \zeta_4 + 2 \zeta_3 \ln ( 1-y ) + 11 \zeta_3 \ln ( 1+y ) - { { \textstyle{\frac{13}{2 } } } } \zeta_3 \ln y - 7 \zeta_2 \ln^2 ( 1+y ) \nonumber \\ & & + 2 \zeta_2 \ln^2 ( 1-y ) - 2 \zeta_2 \ln ( 1-y ) \ln ( 1+y ) - { { \textstyle{\frac{7}{4 } } } } \zeta_2 \ln^2 y - \zeta_2 \ln y \ln ( 1-y ) \nonumber \\ & & + 8 \zeta_2 \ln y \ln ( 1+y ) \bigr ] \ ; , \\ & & \hspace*{-10 mm } \label{an_s1_3 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^3 } s_1 = 4 h_{-1,0,0,1}(-y ) + { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 4 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 4 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - 6 { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & - 2 { { \mbox{li}}_{4}\left(y\right ) } + 4 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y + 4 { { \mbox{s}}_{1,2\!}\left ( y\right ) } \ln y - 2 { { \mbox{s}}_{1,2\!}\left ( y^2\right ) } \ln(y ) + 4 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1-y ) \nonumber \\ & & + 2 { { \mbox{li}}_{3}\left(-y\right ) } \ln y + 2 { { \mbox{li}}_{3}\left(y\right ) } \ln y - { { \mbox{li}}_{2}\left(y\right ) } \ln^2 y - 4 { { \mbox{li}}_{2}\left(-y\right ) } \ln y \ln ( 1-y ) \nonumber \\ & & - { { \textstyle{\frac{1}{3 } } } } \ln^3 y \ln ( 1-y ) + { { \textstyle{\frac{1}{24 } } } } \ln^4 y + 2 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } - { { \textstyle{\frac{1}{2 } } } } \zeta_2 \ln^2 y + 2 \zeta_2 \ln y \ln ( 1-y ) \nonumber \\ & & + 6 \zeta_3 \ln ( 1-y ) - 3 \zeta_3 \ln y - 4 \zeta_4 \ ; , \\ & & \hspace*{-10 mm } \label{an_sb1_3 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^3 } \bar{s}_1 = 4 h_{-1,0,0,1}(-y ) + { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 8 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 4 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - 6 { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & + 2 { { \mbox{li}}_{4}\left(y\right ) } - \left [ { { \mbox{li}}_{2}\left(y\right ) } \right]^2 + 4 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y + 8 { { \mbox{s}}_{1,2\!}\left ( y\right ) } \ln y - 2 { { \mbox{s}}_{1,2\!}\left ( y^2\right ) } \ln y + { { \textstyle{\frac{1}{48 } } } } \ln^4 y \nonumber \\ & & + 4 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1-y ) - 4 { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1-y ) + 2 { { \mbox{li}}_{3}\left(-y\right ) } \ln y - 4 { { \mbox{li}}_{2}\left(-y\right ) } \ln y \ln ( 1-y ) \nonumber \\ & & + 2 { { \mbox{li}}_{2}\left(y\right ) } \ln y \ln ( 1-y ) - { { \textstyle{\frac{1}{2 } } } } { { \mbox{li}}_{2}\left(y\right ) } \ln^2 y - { { \textstyle{\frac{1}{6 } } } } \ln^3 y \ln ( 1-y ) + 4 \zeta_2 \ln y \ln ( 1-y ) \nonumber \\ & & - \zeta_2 \ln^2 y + 10 \zeta_3 \ln ( 1-y ) - 5 \zeta_3 \ln y + 4 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } - { { \textstyle{\frac{19}{2 } } } } \zeta_4 \ ; , \\ & & \hspace*{-10 mm } \label{an_s1 ^ 3_1 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_1 ^ 3 = \frac{1-y}{1+y } \bigl [ - 48 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln ( 1+y ) - 48 { { \mbox{s}}_{1,3\!}\left(-y\right ) } + 24 { { \mbox{s}}_{2,2\!}\left(-y\right ) } \nonumber \\ & & - 12 \zeta_2 \ln^2 ( 1+y ) - 24 \ln^2 ( 1+y ) { { \mbox{li}}_{2}\left(-y\right ) } + 24 \zeta_3 \ln ( 1+y ) + 24 \ln(1+y ) { { \mbox{li}}_{3}\left(-y\right ) } \nonumber \\ & & - 8 \ln y \ln^3 ( 1+y ) + 12 \zeta_2 \ln y \ln ( 1+y ) + 6 \ln^2 y \ln^2 ( 1+y ) - \ln^3 y \ln(1+y ) \nonumber \\ & & + { { \textstyle{\frac{1}{24 } } } } \ln^4 y - { { \textstyle{\frac{3}{2 } } } } \zeta_2 \ln^2 y + 3 \ln^2 y { { \mbox{li}}_{2}\left(-y\right ) } + \ln^2 y { { \mbox{li}}_{2}\left(y\right ) } - 5 \zeta_3 \ln y - 12 \ln y { { \mbox{li}}_{3}\left(-y\right ) } \nonumber \\ & & - 4 \ln y { { \mbox{li}}_{3}\left(y\right ) } + { { \textstyle{\frac{3}{2 } } } } \zeta_4 + 12 { { \mbox{li}}_{4}\left(-y\right ) } + 6 { { \mbox{li}}_{4}\left(y\right ) } \bigr ] \ ; , \\ & & \hspace*{-10 mm } \label{an_s3_2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_3 = - 12 { { \mbox{li}}_{5}\left(y\right ) } + 6 \ln y { { \mbox{li}}_{4}\left(y\right ) } - \ln^2 y { { \mbox{li}}_{3}\left(y\right ) } - { { \textstyle{\frac{1}{120 } } } } \ln^5 y \nonumber \\ & & + \zeta_3 \ln^2 y + 6 \zeta_4 \ln y + 12 \zeta_5 \ ; , \\ & & \hspace*{-10 mm } \label{an_s1s2_2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_1 s_2 = - 12 { { \mbox{li}}_{5}\left(y\right ) } - 16 { { \mbox{li}}_{5}\left(-y\right ) } + 6 \ln y { { \mbox{li}}_{4}\left(y\right ) } + 8 \ln y { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & - \ln^2 y { { \mbox{li}}_{3}\left(y\right ) } - 2 \ln^2 y { { \mbox{li}}_{3}\left(-y\right ) } + { { \textstyle{\frac{1}{3 } } } } \ln^3 y { { \mbox{li}}_{2}\left(-y\right ) } + { { \textstyle{\frac{1}{120 } } } }
\ln^ 5 y - { { \textstyle{\frac{1}{6 } } } } \zeta_2 \ln^3 y \nonumber \\ & & - { { \textstyle{\frac{1}{2 } } } } \zeta_3 \ln^2 y - \zeta_4 \ln y - 3 \zeta_5 \ ; , \\ & & \hspace*{-10 mm } \label{an_s2sb1_2 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_2 \bar{s}_1 = 2 { { \mbox{li}}_{5}\left(y\right ) } - 16 { { \mbox{li}}_{5}\left(-y\right ) } - \ln y { { \mbox{li}}_{4}\left(y\right ) } + 8 \ln y { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \ln^2 y { { \mbox{li}}_{3}\left(y\right ) } - 2 \ln^2 y { { \mbox{li}}_{3}\left(-y\right ) } + { { \textstyle{\frac{1}{3 } } } } \ln^3 y { { \mbox{li}}_{2}\left(-y\right ) } - { { \textstyle{\frac{1}{6 } } } } \ln^3 y { { \mbox{li}}_{2}\left(y\right ) } + { { \textstyle{\frac{1}{240 } } } } \ln^ 5 y \nonumber \\ & & - { { \textstyle{\frac{1}{3 } } } } \zeta_2 \ln^3 y - 2 \zeta_3 \ln^2 y - 8 \zeta_4 \ln y - 17 \zeta_5 \;,\end{aligned}\ ] ] where we have used the relations ( [ hrel1 ] ) and ( [ hrel2 ] ) . for the results involving @xmath72 we can use eq .
( [ lsc - ti ] ) to express them in terms of @xmath230 function , and then employ eq .
( [ def_omega ] ) .
for example , starting from eq .
( [ s1s1sb1_1 ] ) we obtain @xmath231 \nonumber \\ & & + 2 \ln ( 1-y ) \left [ { { \mbox{s}}_{1,2\!}\left(y^2\right ) } - 4 { { \mbox{s}}_{1,2\!}\left(y\right ) } - 2 { { \mbox{li}}_{3}\left(-y\right ) } \right ] - 8 \ln y { { \mbox{li}}_{3}\left(-y\right ) } \nonumber \\ & & + 2 { { \mbox{li}}_{2}\left(-y\right ) } \left [ \ln^2 ( 1-y ) + 2 \ln ( 1-y ) \ln ( 1+y ) - 11 \ln^2(1+y ) + \ln^2 y \right ] \nonumber \\ & & + 2 { { \mbox{li}}_{2}\left(y\right ) } \left [ \ln^2 ( 1+y ) - \ln^2 ( 1-y ) + 2 \ln ( 1-y ) \ln ( 1+y ) \right ] + { { \textstyle{\frac{1}{48 } } } } \ln^4 y \nonumber \\ & & + 2 \ln y \left [ \ln ( 1+y ) \ln^2 ( 1-y ) + \ln ( 1-y ) \ln^2 ( 1+y ) - { { \textstyle{\frac{1}{3 } } } } \ln^3 ( 1-y ) - { { \textstyle{\frac{11}{3 } } } } \ln^3 ( 1+y ) \right ] \nonumber \\ & & - 2 \ln^2 y \ln ( 1-y ) \ln ( 1+y ) + 5 \ln^2 y \ln^2 ( 1+y ) + { { \textstyle{\frac{1}{6 } } } } \ln^3 y \ln ( 1-y ) - { { \textstyle{\frac{2}{3 } } } } \ln^3 y \ln ( 1+y ) \nonumber \\ & & \!+\ ! \zeta_2
\left [ 3 \ln^2 ( 1\!-\!y ) \!-\ ! 2 \ln ( 1\!-\!y ) \ln ( 1\!+\!y ) \!-\ ! 13 \ln^2 ( 1\!+\!y ) \!-\ ! 2 \ln y \ln ( 1\!-\!y ) \!+\ ! 14 \ln y \ln ( 1\!+\!y ) \!-\ ! { { \textstyle{\frac{3}{2 } } } } \ln^2 y \right ] \nonumber \\ & & \!+\ ! \zeta_3 \left [ 3 \ln ( 1-y ) - 6 \ln y + 23 \ln ( 1+y ) \right ] + { { \textstyle{\frac{33}{4 } } } } \zeta_4 \biggr\ } \;,\end{aligned}\ ] ] where @xmath232 , see eq .
( [ def_omega ] ) .
we note that the analytic continuation of @xmath72 can also be obtained directly , using the integral representation ( [ lsc ] ) .
this procedure is described in appendix a , eqs .
( [ begin_an_lsc])([ac_of_lsc ] ) . the analytic continuation of the sum involving @xmath101 can be presented in terms of the variable @xmath233 given in eq .
( [ def_omega_s ] ) , @xmath234 + 2 \bigl [ { { \mbox{li}}_{2}\left(\omega_s\right ) } - { { \mbox{li}}_{2}\left(-\omega_s\right ) } \bigr]^2 + { { \textstyle{\frac{1}{96 } } } } \ln^4 y \;.\ ] ] the results for lower values of @xmath106 can be deduced using @xmath235 for example , @xmath236 the results for these two sums can be extracted from ref .
further results for the sums are collected in appendix c.
below we present results for the @xmath0-expansion of one- and two - loop master integrals shown in fig . 1 .
in the rest of this paper we use the notation @xmath237 .
let us consider a one - loop triangle diagram with @xmath238 , @xmath239 , with an arbitrary ( off - shell ) value of @xmath240 .
such diagrams occur , for example , in higgs decay into two photons or two gluons via a massive quark loop . following the notation of ref .
@xcite , we will denote this integral ( with unit powers of propagators ) as @xmath241 . according to eq .
( 40 ) of ref .
@xcite , the result in an arbitrary space - time dimension @xmath6 is @xmath242 for this integral a number of one - fold integral representations are available , see eqs .
( 3.10)(3.11 ) in ref . @xcite .
expanding in @xmath0 we get @xmath243 .
\hspace*{9mm}\end{aligned}\ ] ] substituting results for the occurring inverse binomial sums , we obtain @xmath244 \nonumber \\ & & + { \varepsilon}^2 \bigl [ 2 \left[{{\mbox{li}}_{2}\left(-y\right ) } \right]^2 + 2 { { \mbox{li}}_{3}\left(-y\right ) } \ln y - 4 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y - { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 y \nonumber \\ & & + 2 \zeta_2 { { \mbox{li}}_{2}\left(-y\right ) } + { { \textstyle{\frac{1}{2}}}}\zeta_2 \ln^2 y + 2 \zeta_3 \ln y - { { \textstyle{\frac{1}{24 } } } } \ln^4 y + { { \textstyle{\frac{5}{4 } } } } \zeta_4 \bigl ] + { \cal o}({\varepsilon}^3 ) \bigl\ } \ ; .
\label{hgg1}\end{aligned}\ ] ] these results correspond to the analytic continuation of eqs . ( 3.13 ) , ( 3.14 ) and ( 3.16 ) from ref . @xcite .
the @xmath54 term is eq .
( [ j3sums ] ) contains the same combination of sums as @xmath245 in eq .
( [ combinations2 ] ) . however , the sum in eq .
( [ j3sums ] ) contains @xmath246 in the denominator , i.e. , it corresponds to the case @xmath161 . using the result ( [ combination:1 ] ) ( with @xmath119 ) together with ( [ higher_c ] ) , we obtain a one - fold integral representation for the @xmath54-term in eq .
( [ j3sums ] ) . analyzing it
, we see that we get only one new non - trivial integral , @xmath247 while all other terms can be expressed in terms of known functions . for @xmath248 the integral ( [ nontrivial5 ] )
is connected with the new element @xmath249 of the _ odd _ basis , whereas for @xmath250 it is related to the new element @xmath251 of the _ even _ basis , see section 3.3 in ref .
@xcite and eqs .
( 14)(15 ) in ref .
@xcite . at the same time , the @xmath252-term of eq .
( [ j3sums ] ) ( and all other even powers of @xmath0 ) can be calculated in terms of log - sine functions , so that their analytic continuation can be expressed in terms of nielsen polylogarithms .
namely , for the @xmath252-term , combining eqs .
( 3.14 ) , ( 3.19 ) from ref .
@xcite and eq .
( [ s4 g ] ) , we obtain the result for the sum @xmath253 with @xmath161 . using eq .
( [ derivative ] ) , we also obtain the result for the case @xmath119 .
this integral is a good illustration of the application of general expressions given in section 3 .
the off - shell result for this integral in arbitrary dimension was presented in @xcite ( where it was called @xmath255 , see eq .
( 22 ) of @xcite ) . for unit powers of propagators ,
the result reads . ]
@xmath256 expanding in @xmath0 , we obtain @xmath257 + { \cal o } ( { \varepsilon}^2)\biggr \ } \nonumber\\ & & \hspace*{-10 mm } = 6 { { \mbox{li}}_{3}\left(y\right ) } - 6 { { \mbox{li}}_{2}\left(y\right ) } \ln y -2 \ln^2 y \ln ( 1-y ) - 6 \zeta_3 \nonumber\\ & & \hspace*{-10 mm } + { \varepsilon}\bigl\ { 28 h_{-1,0,0,1}(-y ) + 7 { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 28 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - 16 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 42 { { \mbox{li}}_{4}\left(-y\right ) } - 26 { { \mbox{li}}_{4}\left(y\right ) } \nonumber\\ & & \hspace*{-10 mm } + 4 \left[{{\mbox{li}}_{2}\left(y\right ) } \right]^2 + 16 { { \mbox{s}}_{1,2\!}\left(y\right ) } \ln y - 14 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \ln y + 28 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y + 18 { { \mbox{li}}_{3}\left(-y\right ) } \ln y \nonumber\\ & & \hspace*{-10 mm } + 20 { { \mbox{li}}_{3}\left(y\right ) } \ln y + 20 { { \mbox{li}}_{3}\left(-y\right ) } \ln ( 1-y ) + 12 { { \mbox{li}}_{3}\left(y\right ) } \ln ( 1-y ) - 2 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 y - 9 { { \mbox{li}}_{2}\left(y\right ) } \ln^2 y \nonumber\\ & & \hspace*{-10 mm } - 24 { { \mbox{li}}_{2}\left(-y\right ) } \ln y \ln ( 1-y ) - 4 { { \mbox{li}}_{2}\left(y\right ) } \ln y \ln ( 1-y ) - 2 \ln^3 y \ln ( 1-y ) + 6 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } \nonumber\\ & & \hspace*{-10 mm } + 4 \zeta_2 \ln y \ln ( 1-y ) - 12 \zeta_3 \ln y + 24 \zeta_3 \ln ( 1-y ) - 9 \zeta_4 \bigr\ }
+ { \cal o } ( { \varepsilon}^2 ) \ ; . \label{f10101}\end{aligned}\ ] ] the result for the finite part coincides with @xcite , whereas the result for the @xmath0-term is new .
let us consider sunset - type diagrams with two equal masses and one zero mass ( see fig . 1 ) .
the off - shell result for the sunset - type integral @xmath258 with arbitrary powers of propagators has been obtained in refs .
@xcite by using the mellin
barnes technique @xcite : @xmath259 for simplicity , in the definition of the integral @xmath258 we will omit the arguments @xmath260 and @xmath261 @xmath262 let us remind that for the integrals @xmath258 with different integer values of @xmath263 and @xmath264 there are two master integrals @xcite of this type , @xmath265 and @xmath266 .
however , two other independent combinations of the integrals of this type happen to be more suitable for constructing the @xmath0-expansion , @xmath267 and @xmath268 ( see also in ref .
the latter combination corresponds to the integral @xmath265 in @xmath269 dimensions @xcite . in ref .
@xcite @xmath270 has been used as the second integral . to construct the @xmath0-expansion of the integrals @xmath265 and
@xmath266 up to order @xmath271 , the integral @xmath267 should also be expanded up to @xmath271 , whereas @xmath272 $ ] or @xmath270 up to the order @xmath0 only . in appendix
c we give an example of a realistic calculation which demonstrates the required orders of the @xmath0-expansion for the integrals involved .
for @xmath267 we have obtained @xmath273 \nonumber \\ & & + { \varepsilon}^2 \biggl [ { { \textstyle{\frac{1}{j^2 } } } } + { { \textstyle{\frac{5}{j}}}}s_1 - { { \textstyle{\frac{2}{j } } } } \bar{s}_1 + { { \textstyle{\frac{25}{2 } } } } s_1 ^ 2 - 10 s_1 \bar{s}_1 - { { \textstyle{\frac{5}{2 } } } } s_2 + 2 \bar{s}_1 ^ 2 + 2 \bar{s}_2 \biggr ] \nonumber \\ & & + { \varepsilon}^3 \left [ { { \textstyle{\frac{1}{j^3 } } } } + { { \textstyle{\frac{5}{j^2}}}}s_1 - { { \textstyle{\frac{2}{j^2 } } } } \bar{s}_1 + { { \textstyle{\frac{25}{2 j } } } } s_1 ^ 2 - { { \textstyle{\frac{10}{j } } } } s_1 \bar{s}_1 - { { \textstyle{\frac{5}{2 j } } } } s_2 + { { \textstyle{\frac{2}{j } } } } \left ( \bar{s}_1 ^ 2 + \bar{s}_2 \right ) + { { \textstyle{\frac{125}{6 } } } } s_1 ^ 3 - 25 s_1 ^ 2 \bar{s}_1 \right . \nonumber \\ & & \left . - { { \textstyle{\frac{25}{2 } } } } s_1 s_2 + { { \textstyle{\frac{11}{3 } } } } s_3 + 5 s_2 \bar{s}_1 + 10 s_1 \biggl ( \bar{s}_1 ^ 2 + \bar{s}_2 \right ) - { { \textstyle{\frac{4}{3 } } } } \left ( \bar{s}_1 ^ 3 + 3 \bar{s}_1 \bar{s}_2 + 2 \bar{s}_3
\biggr ) \right ] + { \cal o } ( { \varepsilon}^4 ) \biggr\ } \nonumber \\ & & \hspace*{-5 mm } = 2 \frac{(m^2)^{-2{\varepsilon}}}{p^2 } \left ( 1-y \right)^ { 2 { \varepsilon } } y^ { 2{\varepsilon } } \biggl\ { - { { \textstyle{\frac{1}{2 } } } } \ln^2 y \nonumber \\ & & + { \varepsilon}\bigl [ { { \textstyle{\frac{1}{2 } } } } \ln^3 y + \zeta_2 \ln y -6 \ln y { { \mbox{li}}_{2}\left(-y\right ) } -4 \ln y { { \mbox{li}}_{2}\left(y\right ) } + 3 \zeta_3 + 12 { { \mbox{li}}_{3}\left(-y\right ) } + 6 { { \mbox{li}}_{3}\left(y\right ) } \bigr ] \nonumber \\ & & + { \varepsilon}^2 \bigl [ 12 h_{-1,0,0,1 } ( -y ) -12 \ln y { { \mbox{s}}_{1,2\!}\left(-y\right ) } -12 \ln y { { \mbox{s}}_{1,2\!}\left(y^2\right ) } + 8 \ln y { { \mbox{s}}_{1,2\!}\left(y\right ) } - 12 { { \mbox{s}}_{2,2\!}\left(-y\right ) } \nonumber \\ & & + 3 { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 12 \ln ( 1-y ) { { \mbox{li}}_{3}\left(-y\right ) } -{{\textstyle{\frac{7}{24 } } } } \ln^4 y - { { \textstyle{\frac{1}{2 } } } } \zeta_2 \ln^2 y + 2 \zeta_3 \ln y + 6 \ln y { { \mbox{li}}_{3}\left(-y\right ) } \nonumber \\ & & + 8 \ln y { { \mbox{li}}_{3}\left(y\right ) } + 6 \zeta_2 { { \mbox{li}}_{2}\left(-y\right ) } + 4 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } + { { \textstyle{\frac{13}{4 } } } } \zeta_4 + 18 \left [ { { \mbox{li}}_{2}\left(-y\right ) } \right]^2 \nonumber \\ & & + 12 { { \mbox{li}}_{2}\left(y\right ) } { { \mbox{li}}_{2}\left(-y\right ) } + 4 \left [ { { \mbox{li}}_{2}\left(y\right ) } \right]^2 - { { \textstyle{\frac{9}{4 } } } } { { \mbox{li}}_{4}\left(y^2\right ) } \bigr ] + { \cal o } ( { \varepsilon}^3 ) \biggr\ } \ ; , \label{j011_122}\end{aligned}\ ] ] where the result for the @xmath271-term is new .
for another combination , we get @xmath274 + { \varepsilon}\left [ { { \textstyle{\frac{25}{2 } } } } s_1 ^ 2 - 10 s_1 \bar{s}_1 - { { \textstyle{\frac{5}{2 } } } } s_2 + 2 \bar{s}_1 ^ 2 + 2 \bar{s}_2 \right ] \nonumber \\ & & \hspace*{10 mm } + { \varepsilon}^2 \left [ { { \textstyle{\frac{125}{6 } } } } s_1 ^ 3 - 25 s_1 ^ 2 \bar{s}_1 - { { \textstyle{\frac{25}{2 } } } } s_1 s_2 + { { \textstyle{\frac{11}{3 } } } } s_3 + 5 s_2 \bar{s}_1 + 10 s_1 \biggl ( \bar{s}_1 ^ 2 + \bar{s}_2 \right ) \nonumber \\ & & \hspace*{20 mm } - { { \textstyle{\frac{4}{3 } } } } \left ( \bar{s}_1 ^ 3 + 3 \bar{s}_1 \bar{s}_2 + 2 \bar{s}_3
\biggr ) \right ] \!+\ !
{ \cal o } ( { \varepsilon}^3 ) \biggr\ } \nonumber \\ & & \hspace*{-10 mm } = - \frac{2(m^2)^{-2{\varepsilon}}}{p^2}\ ; \frac{(1-y)^{1 - 2{\varepsilon}}}{(1+y)^{1 + 6{\varepsilon } } } \biggl\ { \frac{1}{{\varepsilon}}\ln y + 2 \ln^2 y - \zeta_2 - 6 { { \mbox{li}}_{2}\left(-y\right ) } - 2 { { \mbox{li}}_{2}\left(y\right ) } \nonumber \\ & & + { \varepsilon}\bigl [ 24 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \!+\ ! 6 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } \!-\ ! 8 { { \mbox{s}}_{1,2\!}\left(y\right ) } \!+\ ! { { \textstyle{\frac{5}{3 } } } } \ln^3 y \!-\ ! 4 \zeta_2 \ln y \!-\ ! 11 \zeta_3 \!-\ ! 24 { { \mbox{li}}_{3}\left(-y\right ) } \!-\ ! 8 { { \mbox{li}}_{3}\left(y\right ) } \bigr ] \nonumber \\ & & + { \varepsilon}^2 \bigl [ 24 { { \mbox{li}}_{4}\left(\omega\right ) } - 24 { { \mbox{li}}_{4}\left(-\omega\right ) } - 24 { { \mbox{s}}_{1,3\!}\left(y^2\right ) } + 12 { { \mbox{s}}_{2,2\!}\left(y^2\right ) } + 64 { { \mbox{s}}_{1,3\!}\left(y\right ) } \nonumber \\ & & \hspace{10 mm } + 96 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - 32 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 192 { { \mbox{s}}_{1,3\!}\left(-y\right ) } + 52 { { \mbox{li}}_{4}\left(y\right ) } + 48 { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & \hspace{10 mm } - 6 \ln\omega \left [ 2 { { \mbox{s}}_{1,2\!}\left(y^2\right ) } - 8 { { \mbox{s}}_{1,2\!}\left(y\right ) } - 8 { { \mbox{s}}_{1,2\!}\left(-y\right ) } + 7 \zeta_3 \right ] - 72 \ln y { { \mbox{li}}_{3}\left(-y\right ) } - 48 \ln y { { \mbox{li}}_{3}\left(y\right ) } \nonumber \\ & & \hspace{10 mm } + 6 \ln^2 \omega \left [ 2 { { \mbox{li}}_{2}\left(y\right ) } - 2 { { \mbox{li}}_{2}\left(-y\right ) } - 3 \zeta_2 \right ] + 12 \ln^2 y { { \mbox{li}}_{2}\left(y\right ) } + 18 \ln^2 y { { \mbox{li}}_{2}\left(-y\right ) } \nonumber \\ & & \hspace{10 mm } + 4 \ln y \ln^3 \omega + { { \textstyle{\frac{11}{12 } } } } \ln^4 y - 5 \zeta_2 \ln^2 y - 26 \zeta_3 \ln y - 57 \zeta_4 \bigr ] + { \cal o } ( { \varepsilon}^3 ) \biggr\ } \ ; , \label{j011_c}\end{aligned}\ ] ] where @xmath232 , see eq . ( [ def_omega ] ) . the @xmath0-term can be related to that of the result for @xmath270 presented in ref .
@xcite , whereas the result for the @xmath271 term is new .
consider the two - loop vertex - type diagram @xmath275 given in ref .
@xcite from ref .
@xcite corresponds to our @xmath260 , whereas their @xmath276 corresponds to our @xmath19 .
for @xmath275 we keep the normalization used in in ref .
@xcite for two - loop vertices : each loop integral is divided by @xmath277 , where @xmath31 is euler s constant .
note that it is different from the normalization used in ref .
@xcite , where each loop is divided by @xmath278 , where @xmath279 is the scale parameter of dimensional regularization @xcite .
furthermore , in numerical results presented in table 1 of ref .
@xcite an extra common factor @xmath280 was extracted . ]
( see fig . 1 ) , @xmath281 \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \log^2(-u ) - s_1 \log(-u ) - { { \textstyle{\frac{3}{2 } } } } s_2 - { { \textstyle{\frac{15}{2 } } } } s_1 ^ 2 + 4 s_1 \bar{s}_1 + 2 \frac{s_1}{j } + { \cal o } ( { \varepsilon } ) \biggr\ } \ ; .\end{aligned}\ ] ] it corresponds to one of the two - loop contributions to a boson decay into two massless particles , with a massive triangle subloop .
diagrams of such type have been intensively studied for higgs boson production via gluon fusion @xcite . using our approach ,
we obtain @xmath282 \nonumber \\ & & + 8 h_{-1,0,0,1}(-y ) + 2 { { \mbox{s}}_{2,2\!}\left(y^2\right ) } - 8 { { \mbox{s}}_{2,2\!}\left(-y\right ) } - 8 { { \mbox{s}}_{2,2\!}\left(y\right ) } - 4 { { \mbox{li}}_{4}\left(y\right ) } - 12 { { \mbox{li}}_{4}\left(-y\right ) } \nonumber \\ & & - 8 { { \mbox{li}}_{2}\left(y\right ) } { { \mbox{li}}_{2}\left(-y\right ) } - 14 \left [ { { \mbox{li}}_{2}\left(-y\right ) } \right]^2 + 28 { { \mbox{s}}_{1,2\!}\left(-y\right ) } \ln y - 10 { { \mbox{li}}_{3}\left(-y\right ) } \ln y \nonumber \\ & & + 7 { { \mbox{li}}_{2}\left(-y\right ) } \ln^2 y - 4 { { \mbox{li}}_{2}\left(-y\right ) } \ln y \ln ( 1-y ) - \ln^2 y \ln^2 ( 1-y ) - { { \textstyle{\frac{1}{6 } } } } \ln^4 y \nonumber \\ & & + { { \textstyle{\frac{2}{3 } } } } \ln^3 y \ln ( 1-y ) + 2 \zeta_2 \ln y \ln ( 1-y ) - { { \textstyle{\frac{5}{2 } } } } \zeta_2 \ln^2 y + 6 \zeta_3 \ln ( 1-y ) - 11 \zeta_3 \ln y \nonumber \\ & & - 6 { { \mbox{li}}_{2}\left(-y\right ) } \zeta_2 - { { \textstyle{\frac{27}{4 } } } } \zeta_4 + { \cal o } ( { \varepsilon } ) \ ; . \label{hgg2}\end{aligned}\ ] ] here , the result for the finite part is new .
alternatively , results of such type can be obtained in a different way , using a technique based on mellin
barnes contour integrals @xcite .
let us consider another two - loop vertex - type diagram shown in fig . 1 , @xmath283 , where two external momenta are on shell ( @xmath284 ) and we also put @xmath285 .
note that @xmath286 appears as one of the master integrals in ref .
@xcite ( see eqs .
( 102)(106 ) of @xcite ) .
first of all , integration over the momentum of the massless loop ( with powers of the propagators equal to @xmath287 and @xmath288 ) can be easily performed , yielding the corresponding massless propagator to the power @xmath289 , times a well - known factor containing @xmath290 functions .
the resulting integral is nothing but a one - loop triangle function , @xmath291 in the notation of ref .
@xcite , with two external momenta ( @xmath292 and @xmath293 ) on shell and @xmath294 . using the results of ref .
@xcite , one can see that in this on - shell limit the three - point integral @xmath295 reduces to a two - point function @xmath296 with two massive propagators ( with the powers @xmath297 and @xmath298 ) , but with a shifted value of the space - time dimension ( @xmath299 ) , @xmath300 this is a generalization to arbitrary values of @xmath264 of the relation found in ref .
@xcite for @xmath301 .
using eq .
( [ 3to2 ] ) , we arrive at the following result for the diagram @xmath302 ( which we normalize by dividing each loop by @xmath303 ) : @xmath304 where @xmath237 . for @xmath305 and @xmath306 , the parameter @xmath307 in eq .
( [ 3to2 ] ) gets equal to @xmath0 , and the function corresponds to a one - loop two - point function in @xmath308 dimensions ( i.e. , @xmath309 ) , @xmath310 as in the one - loop case ( cf .
( c29)(c30 ) of ref .
@xcite ) , the @xmath44 function can be reduced to the type considered in section 2 by means of one of the kummer relations , @xmath311 all orders of the @xmath0-expansion of the resulting @xmath44 function are given in ( see also in refs .
@xcite ) . for another special case , @xmath312 , the parameter @xmath307 in eq .
( [ 3to2 ] ) is equal to @xmath313 , and the function corresponds to a one - loop two - point function in @xmath314 dimensions . in this case
, we directly get the @xmath44 function of the type ( [ basis_ii ] ) , with @xmath309 .
analytic continuation in terms of the nielsen polylogarithms was discussed in section 2.2 of ref . @xcite . in the case of @xmath286 ,
we get @xmath315 \biggr\ } \ ; , \hspace*{8 mm } \label{k11110_snp}\end{aligned}\ ] ] where the generalized polylogarithms @xmath316 can be expressed in terms of inverse argument by means of the standard formulae given in ref .
@xcite . in this way
, we have obtained results for _ all _ coefficients of the @xmath0-expansion of @xmath286 .
the first three coefficients coincide with those given in eqs .
( 104)(106 ) of ref .
@xcite .
consider a three - loop vacuum diagram with two different masses shown in fig .
( [ propagator ] ) .
such an integral with equal masses ( @xmath318 ) enters as a master integral in avdeev s package @xcite and matad @xcite .
although there is no similar package for three - loop vacuum diagrams with two different masses , it is clear that such configuration ( with unit powers of the propagators ) will remain one of the master integrals in this more general case .
this integral can be calculated by integrating over the momentum @xmath319 the off - shell two - loop diagram @xmath254 ( also shown in fig .
[ propagator ] ) with a massive propagator containing an arbitrary mass @xmath320 ( see eq . ( 4.7 ) in @xcite ) , @xmath321 where @xmath322 .
the result of this integration can be presented in the following form : @xmath323 where , as before , @xmath324 . all hypergeometric functions occurring in eq .
( [ d4_hypergeom ] ) belong to the type considered in section 2 . applying the results for the @xmath0-expansion of these functions , we arrive at @xmath325 \!-\ !
2 \theta { { \mbox{ls}}_{3}\left(\theta\right ) } \!-\ !
6 \left [ { { \mbox{ls}}_{2}\left(\theta\right ) } \right]^2 \!+\ ! { { \textstyle{\frac{1}{12 } } } } \theta^4 \!-\ ! { { \textstyle{\frac{1}{2 } } } } \zeta_2 \theta^2 \!-\ ! 9 \zeta_4 + { \cal o } ( { \varepsilon } ) .
\hspace*{4 mm } \label{d4_theta}\end{aligned}\ ] ] using analytic continuation described in section 3 , we can also present this result in terms of the variable @xmath224 defined in ( [ y<->u ] ) , @xmath326 \nonumber \\ & & + 4 { { \mbox{li}}_{4}\left(y\right ) } - 4 { { \mbox{s}}_{2,2\!}\left(y\right ) } + 6 \left[{{\mbox{li}}_{2}\left(y\right ) } \right]^2 - 4 \ln ( 1-y ) { { \mbox{li}}_{3}\left(y\right ) } + 12 \ln y \ln ( 1-y ) { { \mbox{li}}_{2}\left(y\right ) } \nonumber \\ & & - 3 \ln^2 y { { \mbox{li}}_{2}\left(y\right ) } + 5 \ln^2 y \ln^2 ( 1-y ) - { { \textstyle{\frac{7}{3 } } } } \ln^3 y \ln ( 1-y ) + { { \textstyle{\frac{1}{4 } } } } \ln^4 y \nonumber \\ & & - 12 \zeta_2 { { \mbox{li}}_{2}\left(y\right ) } - 8 \zeta_2 \ln y \ln ( 1-y ) + { { \textstyle{\frac{3}{2 } } } } \zeta_2 \ln^2 y + 4 \zeta_3 \ln ( 1-y ) + 3 \zeta_4 + { \cal o}({\varepsilon } ) .
\label{d4_y}\end{aligned}\ ] ] one should remember however , that in this case the variable @xmath224 becomes complex ( although all imaginary parts should cancel for real masses ) , whereas eq .
( [ d4_theta ] ) is explicitly real . as a non - trivial check on these results we consider two particular values , @xmath327 and @xmath328 . in the first case ( @xmath329 ) , we reproduce the known result ( see in ref .
@xcite ) for the master integral @xmath330 , @xmath331 ^ 2 + { \cal o}({\varepsilon } ) \;.\ ] ] in the second case , @xmath328 ( @xmath332 , @xmath333 ) , the result for the master integral @xmath334 is reproduced ( which was first calculated in ref .
@xcite ) , @xmath335 in a similar manner , analytical results can be deduced for other diagrams with two different mass scales , like @xmath336 and @xmath337 ( for notations , see in ref .
@xcite ) . to our knowledge
, this is the first example of the calculation of the finite part of a three - loop vacuum diagram with six internal lines with two different mass scales . in previous publications
@xcite , only the divergent parts of some three - loop vacuum integrals have been analyzed .
in this paper , we have studied the _ multiple inverse binomial sums _ of the type ( [ binsum ] ) , for arbitrary values of the variable @xmath19 .
our analysis was based on the connection between these sums and terms of the expansion of certain hypergeometric functions with respect to the parameter @xmath0 . using known results for the hypergeometric functions , together with the integral representation ( [ higher_c ] ) , we have obtained a number of new analytical results up to the weight * 4 * , eqs .
( [ s1a])([last4 ] ) , ( [ s1c2])([s2bar_2 ] ) and ( [ s1_3])([s1bar_3 ] ) .
moreover , in some cases like ( [ s2g])([s4 g ] ) the results can be obtained for an arbitrary weight .
constructing analytical continuation of the obtained results , we have expressed them in terms of the generalized polylogarithms ( [ an_s1s1_2])([an_sb1_3 ] ) . in the cases considered , only one new function was needed , in addition to the basis of nielsen polylogarithms .
as such function one can take , e.g. , the harmonic polylogarithm @xmath338 , where the variable @xmath224 is given in eq .
( [ y<->u ] ) .
this approach allowed us to construct some terms of the @xmath0-expansion of the generalized hypergeometric function @xmath339 ( [ expansion ] ) and obtain new analytical results for higher terms of the @xmath0-expansion of some one- and two - loop propagator - type ( see eqs .
( [ f10101 ] ) , ( [ j011_122 ] ) and ( [ j011_c ] ) ) and vertex - type ( see eqs .
( [ hgg1 ] ) and ( [ hgg2 ] ) ) diagrams depending on one dimensionless variable @xmath237 . as a by - product
, we have analytically proven the earlier published results for the three - loop vacuum integrals , the finite part of @xmath340 and the @xmath0-part of @xmath341 and @xmath342 , corresponding to the particular value of @xmath32 ( @xmath343 ) , see sections 4.3 , 4.5 and 4.6 in ref . @xcite and refs .
these integrals enter as master integrals in avdeev s package @xcite and matad @xcite . moreover , the developed technique is useful in the calculation of two - loop vertex - type diagrams @xcite
. * acknowledgements .
* we are grateful to f. jegerlehner , a. kotikov , m. spira , v.a .
smirnov and o.v .
tarasov for useful discussions .
we would like to thank r. delbourgo and j. fleischer for their interest in our work , and j. gluza for checks of some formulae .
we are very thankful to g. passarino and s. uccirati for an independent numerical check of some expressions from sections 3 and 4 which helped to identify typos in earlier versions of eqs .
( 3.11 ) and ( [ hgg2 ] ) .
this research was supported in part by intas - cern grant no .
99 - 0377 and by the australian research council grant
here we collect some properties of the harmonic polylogarithms introduced in @xcite .
let us define a @xmath344-dimensional vector @xmath345 , where @xmath346 is the leftmost component of @xmath347 , while @xmath348 stands for the vector of the remaining @xmath349 components .
the harmonic polylogarithms of weight @xmath344 are then defined as follows : @xmath350 where the three rational fractions @xmath351 are given by @xmath352 the derivatives can be written in a compact form @xmath353 let us put @xmath354 , so that @xmath355 where @xmath356 consider , for example , the harmonic polylogarithm @xmath357 given in eq .
( [ h(-1001 ) ] ) . using the decomposition of @xmath358 into the real and imaginary parts @xcite and integrating by parts , we obtain expressions in terms of clausen s and generalized log - sine functions , @xmath359 \nonumber \\ & & - { \rm i } \left [ { { \mbox{gl}}_{2}\left(\theta\right ) } { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } + { { \textstyle{\frac{1}{2 } } } } ( \pi- \theta ) { { \mbox{cl}}_{3}\left(\pi-\theta\right ) } - { { \textstyle{\frac{1}{2}}}}\pi { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] \ ; , \end{aligned}\ ] ] where @xmath360 and @xmath129 is defined in eq .
( [ def_phi ] ) . in this paper
, we have also used the following relations : @xmath361 and @xmath362 the following two representations of @xmath357 are also useful : @xmath363 .\end{aligned}\ ] ] instead of @xmath129 , one could also introduce another function .
one of such possibilities is to consider a generalization of the glashier function .
let us recall that the real part of @xmath364 can be presented as$ ] ( see in ref .
@xcite ) . ] @xmath365 where @xmath366 and @xmath367 are clausen and glashier functions , respectively .
in particular , @xmath367 is just a polynomial in @xmath122 . a possible non - trivial generalization of @xmath367 ( for even @xmath143 )
could be @xmath368 so that @xmath369 . in particular , @xmath370 where @xmath371 so that @xmath372 using eqs . ( 17 ) and ( 18 ) on p. 293 of @xcite , we can see that the general result ( [ gl2_gen ] ) simplifies in the case @xmath373 ( @xmath374 ) , @xmath375 then , let us consider @xmath376 for @xmath332 we get @xmath377 where @xmath378 is the alternating two - fold euler sum considered in @xcite ( see also ref .
@xcite ) , @xmath379 for general @xmath122 , one can see that @xmath125 is related to the real part of a harmonic polylogarithm , @xmath380 in particular , the function @xmath129 can be presented as @xmath381 \ln\left(2\cos{{\textstyle{\frac{\theta}{2}}}}\right ) , \ ] ] where @xmath382
. using the symmetry property of @xmath129 , eq .
( [ sym_phi ] ) , we get @xmath383 \ln\left(2\cos{{\textstyle{\frac{\theta}{2}}}}\right ) \nonumber \\ & & -\left[{{\mbox{cl}}_{3}\left(\pi-\theta\right)}-{{\mbox{cl}}_{3}\left(\pi\right)}\right ] \ln\left(2\sin{{\textstyle{\frac{\theta}{2}}}}\right ) .
\label{sym_gl4}\end{aligned}\ ] ] for special values of @xmath122 , eq .
( [ sym_gl4 ] ) yields @xmath384 ^ 2 \ ; , \end{aligned}\ ] ] where @xmath385 is the catalan constant .
moreover , with the help of pslq algorithm @xcite one can obtain results for @xmath386 and @xmath387 separately , @xmath388 ^ 2 + { { \textstyle{\frac{1}{4 } } } } \pi { { \mbox{ls}}_{3}\left({{\textstyle{\frac{2\pi}{3}}}}\right ) } - { { \textstyle{\frac{3}{8 } } } } { { \mbox{ls}}_{4}^{(1)}\left({{\textstyle{\frac{2\pi}{3}}}}\right ) } \ ; , \\ { { \mbox{gl}}_{4}\left({{\textstyle{\frac{2\pi}{3 } } } } ; 1\right ) } & = & -{{\textstyle{\frac{241}{216 } } } } \zeta_4 -{{\textstyle{\frac{1}{3}}}}\left [ { { \mbox{cl}}_{2}\left({{\textstyle{\frac{\pi}{3}}}}\right ) } \right]^2 - { { \textstyle{\frac{1}{4 } } } } \pi { { \mbox{ls}}_{3}\left({{\textstyle{\frac{2\pi}{3}}}}\right ) } + { { \textstyle{\frac{3}{8 } } } } { { \mbox{ls}}_{4}^{(1)}\left({{\textstyle{\frac{2\pi}{3}}}}\right ) } \ ; .\end{aligned}\ ] ] we would also like to discuss the analytical continuation of @xmath72 ( see eq .
( [ lsc ] ) ) , @xmath389 if we introduce a variable @xmath390 , we see that @xmath391 and the analytical continuation of @xmath72 is given by @xmath392 \ ; , \nonumber \end{aligned}\ ] ] with @xmath224 defined in ( [ y : def ] ) .
this integral can be calculated by using the following relations @xmath393 \ ; , \\
\int\limits_0^y \frac{{\rm d}z}{z } \ln^2 ( 1\mp z ) \ln ( 1\pm z ) & = & 2 \ln(1\pm y ) { { \mbox{s}}_{1,2\!}\left(\pm y\right ) } - 2 \h_{-1,0,1,1}(\pm y ) , \end{aligned}\ ] ] where @xmath394 is another harmonic polylogarithm @xcite .
we note that @xmath395 @xmath396 therefore , the analytic continuation of @xmath72 reads @xmath397 \nonumber \\ & & + \ln^2{y } \left [ { { \textstyle{\frac{1}{4 } } } } { { \mbox{li}}_{2}\left(y\right)}+{{\textstyle{\frac{1}{2}}}}{{\mbox{li}}_{2}\left(-y\right ) } \right ] + { { \textstyle{\frac{1}{32}}}}\ln^4{y } -{{\textstyle{\frac{3}{16}}}}\zeta_4 -{{\textstyle{\frac{1}{6 } } } } \int\limits_0^y \frac{{\rm d}z}{z}\ ; \ln^3\left(\frac{1-z}{1+z}\right ) \nonumber \\ & & -{\rm i}\sigma\pi \left\ { { { \mbox{s}}_{1,2\!}\left(-y\right ) } - { { \textstyle{\frac{1}{2}}}}{{\mbox{li}}_{3}\left(-y\right ) } + { { \textstyle{\frac{1}{2}}}}\ln{y } { { \mbox{li}}_{2}\left(-y\right ) } + { { \textstyle{\frac{1}{24}}}}\ln^3{y } -{{\textstyle{\frac{1}{2}}}}\zeta_3 \right\ } \ ; .\end{aligned}\ ] ] note that the integral occurring in eqs .
( [ new_h_via_li ] ) and ( [ ac_of_lsc ] ) is directly related to the integral in eq .
( [ lsc23_via_int ] ) , and it can be calculated in terms of polylogarithms , @xmath398 - 6 \ln{\omega } \left [ { { \mbox{li}}_{3}\left(\omega\right ) } - { { \mbox{li}}_{3}\left(-\omega\right ) } \right ] \nonumber \\ & & + 3 \ln^2{\omega } \left [ { { \mbox{li}}_{2}\left(\omega\right ) } - { { \mbox{li}}_{2}\left(-\omega\right ) } \right ] + \ln^3{\omega } \ln{y } \ ; , \hspace*{8mm}\end{aligned}\ ] ] where @xmath232 ( see eq .
( [ def_omega ] ) ) . therefore , @xmath399 reduces to polylogarithms , as well as the analytic continuation of @xmath72 .
we note that the imaginary part on the r.h.s . of eq .
( [ ac_of_lsc ] ) gets cancelled in the results for the corresponding sums .
the following relations are useful for the transformations : @xmath400 \nonumber \\ & & - { { \textstyle{\frac{1}{2 } } } } \ln^2 \omega \ln y + { { \textstyle{\frac{3}{2 } } } } \zeta_2 \ln \omega + { { \textstyle{\frac{7}{4 } } } } \zeta_3 \;. \ ] ]
as we have seen , the @xmath0-expansion of the hypergeometric functions produces series of the type ( [ binsum ] ) .
one can use certain properties of hypergeometric functions to get relations between the sums ( [ binsum ] ) involved in the @xmath0-expansion .
let us consider eq .
( b.18 ) of ref .
@xcite ( which follows from eq .
( 22 ) on p. 498 of @xcite ) , @xmath401 ^ 2 .
\label{less_trivial}\ ] ] it reduces the given @xmath402 function to a square of the @xmath44 function . substituting the @xmath0-expansions of the @xmath402 and @xmath44 functions into ( [ less_trivial ] ) , we obtain the following relations between the sums : @xmath403 ^ 2 \ ; , \\ & & \sigma_{1;-;2}^{1;-}(u ) - \sigma_{-;1;2}^{-;1}(u ) - \sigma_{-;-;3}^{-;-}(u ) = \frac{4-u}{u } \ ; \sigma_{-;-;1}^{-;-}(u ) \left [ \sigma_{1;-;1}^{1;-}(u ) - \sigma_{-;1;1}^{-;1}(u ) \right ] \ ; , \\ & & \sigma_{2;-;2}^{1;-}(u ) = \frac{4-u}{4u } \ ; \sigma_{-;-;1}^{-;-}(u ) \sigma_{2;-;1}^{1;-}(u ) \ ; , \\ & & \sigma_{1;-;2}^{2;-}(u ) - 2 \sigma_{1;1;2}^{1;1}(u ) + \sigma_{-;1;2}^{-;2}(u ) + \sigma_{-;2;2}^{-;1}(u ) -2\sigma_{1;-;3}^{1;-}(u ) + 2\sigma_{-;1;3}^{-;1}(u ) + 2\sigma_{-;-;4}^{-;-}(u ) \nonumber \\ & & \hspace*{10 mm } = \frac{4-u}{u } \bigl [ \sigma_{-;-;1}^{-;-}(u ) \left ( \sigma_{1;-;1}^{2;-}(u ) - 2 \sigma_{1;1;1}^{1;1}(u ) + \sigma_{-;1;1}^{-;2}(u ) + \sigma_{-;2;1}^{-;1}(u ) -{{\textstyle{\frac{3}{8 } } } } \sigma_{2;-;1}^{1;-}(u)\right ) \nonumber \\ & & \hspace*{24 mm } + \left ( \sigma_{1;-;1}^{1;-}(u ) - \sigma_{-;1;1}^{-;1}(u)\right)^2 \bigl ] \ ; , \end{aligned}\ ] ] with @xmath48 , so that @xmath404 another interesting relation , eq .
( b.19 ) of ref .
@xcite ( which follows from eq .
( 20 ) on p. 498 of @xcite ) , reads @xmath405}\ ; { } _ { 3}f_2 \left(\begin{array}{c| } 1+{{\textstyle{\frac{1}{2 } } } } a_1 { \varepsilon } , 1+{{\textstyle{\frac{1}{2 } } } } a_2 { \varepsilon } , 1 \\ { { \textstyle{\frac{3}{2}}}}+ { { \textstyle{\frac{1}{2}}}}(a_1+a_2){\varepsilon } , \ ; 2 \end{array } ~z \right ) \right\ } \ ; . \label{even_less_trivial}\end{aligned}\ ] ] in this way , we get the following relations : @xmath406 + \sigma_{-;-;2}^{-;-}(u ) \left [ \sigma_{1;-;1}^{1;-}(u ) - \sigma_{-;1;1}^{-;1}(u ) \right ] \ ; . \hspace*{7mm}\end{aligned}\ ] ] we have checked that all above equations are satisfied by the explicit results for the sums listed in this paper .
let us also present some relations for higher - order sums : @xmath407 \right .
\nonumber \\ & & \left .
+ \sigma_{-;-;1}^{-;-}(u ) \left [ \sigma_{2;1;1}^{1;1}(u ) - \sigma_{2,1;-;1}^{1,1;-}(u ) + { { \textstyle{\frac{1}{2 } } } } \sigma_{3;-;1}^{1;-}(u ) \right ] \right\ } , \\ \label{higher2 } & & \hspace{-5 mm }
\sigma_{2;-;2}^{2;-}(u ) - \sigma_{4;-;2}^{1;-}(u ) = \frac{4-u}{16u } \left\ { \left [ \sigma_{2;-;1}^{1;-}(u )
\right]^2 + \sigma_{-;-;1}^{-;-}(u ) \left [ \sigma_{2;-;1}^{2;-}(u ) - \sigma_{4;-;1}^{1;-}(u ) \right ] \right\ } \ ; , \nonumber \\
\label{higher3 } & & \hspace{-5 mm } \sigma_{2;-;1}^{2;-}(u ) -\sigma_{4;-;1}^{1;-}(u ) = { { \textstyle{\frac{2}{45 } } } } \sigma_{-;-;1}^{-;-}(u ) \left\ { \left [ \sigma_{-;-;2}^{-;-}(u ) \right]^2 + 3 \sigma_{2;-;2}^{1;-}(u ) \right\ } \ ; , \\ \label{higher4 } & & \hspace{-5 mm } 2 \left[\sigma_{-;-;2}^{-;-}(u ) \right]^2 \left [ \sigma_{1;-;1}^{1;-}(u ) - \sigma_{-;1;1}^{-;1}(u ) \right ] + 4 \sigma_{-;-;2}^{-;-}(u ) \sigma_{-;-;1}^{-;-}(u ) \left [ \sigma_{1;-;2}^{1;-}(u ) - \sigma_{-;1;2}^{-;1}(u ) \right ] \nonumber \\ & & \hspace{-5 mm } + 6 \sigma_{2;-;2}^{1;-}(u ) \left [ \sigma_{1;-;1}^{1;-}(u ) - \sigma_{-;1;1}^{-;1}(u ) \right ] + 6 \sigma_{-;-;1}^{-;-}(u ) \left [ \sigma_{2,1;-;2}^{1,1;-}(u ) - \sigma_{2;1;-}^{1;1}(u ) \right ] \nonumber \\ & & \hspace{-5 mm } + 45 \sigma_{2;1;1}^{2;1}(u ) - 45 \sigma_{2,1;-;1}^{2,1;-}(u ) + 45 \sigma_{4,1;-;1}^{1,1;-}(u ) - 45 \sigma_{4;1;1}^{1;1}(u ) + 93 \sigma_{3,2;-;1}^{1,1;-}(u ) - 93 \sigma_{5;-;1}^{1;-}(u ) \nonumber \\ & & \hspace{-5 mm } + 4 \sigma_{3;-;1}^{1;-}(u ) \sigma_{-;-;2}^{-;-}(u ) - 3 \sigma_{3;-;2}^{1;-}(u ) \sigma_{-;-;1}^{-;-}(u ) = 0 \ ; , \end{aligned}\ ] ] where eqs .
( [ higher1 ] ) and ( [ higher2 ] ) follow from eq .
( [ less_trivial ] ) , while eqs .
( [ higher3 ] ) and ( [ higher4 ] ) follow from eq .
( [ even_less_trivial ] ) . one more relation can be derived from eq .
( 47 ) on p. 456 of @xcite , @xmath408 we can compare the @xmath0-expansions of these @xmath44 functions , considering the coefficients of @xmath409 as functions of @xmath122 . introducing @xmath410 we get the following relations at orders @xmath411 and @xmath412 : @xmath413 at order @xmath271 , comparison of the coefficients of @xmath414 and @xmath415 yields @xmath416 \nonumber \\ & & \hspace*{30 mm } = { { \textstyle{\frac{1}{2 } } } } \cot \theta \sum_{j=1}^{\infty } \frac{1}{\left(2j\atop j\right ) } \frac{{\widetilde{u}}^j}{j } \left [ { { \textstyle{\frac{1}{2}}}}\left(\bar{s}_2+\bar{s}_1 ^ 2\right ) -s_1 \bar{s}_1 -{{\textstyle{\frac{1}{8}}}}s_2 + { { \textstyle{\frac{1}{2}}}}s_1 ^ 2 \right ] \ ; .\end{aligned}\ ] ] all these equations are satisfied by analytic expressions for these sums in terms of @xmath122 from section 2 of this paper
. remember that one needs to substitute @xmath417 as an argument on the r.h.s . at order @xmath54
, we have two independent structures , @xmath418 and @xmath419 .
it is more convenient , however , to compare the coefficients of @xmath418 and @xmath420 .
the first equation coming from the coefficients of @xmath418 yields @xmath421 where @xmath422 are the combinations of the harmonic sums defined in eqs .
( [ combinations2 ] ) .
this equation is also satisfied , if we use analytic expressions for these sums in terms of @xmath122 given in section 2 .
the second equation at order @xmath54 comes from the coefficient of @xmath420 , @xmath423 if we introduce two functions @xmath424 we obtain an interesting relation between them , @xmath425 this relation is satisfied by the explicit results for the sums involved given in eqs .
( [ s3c1 ] ) , ( [ s1s2c1 ] ) , ( [ s2s1barc1 ] ) .
for completeness , in this appendix we collect some results for the _ multiple inverse binomial sums _ of lower weights .
they can be obtained by applying the operator @xmath426 to our results presented in sections 2 and 3 .
these sums occur in lower terms of the @xmath0-expansion of hypergeometric functions given in eq .
( [ expansion ] ) .
first of all , we list explicit results for some particular cases of the general formulae ( [ kv ] ) and ( [ s2 g ] ) , in terms of the angular variable @xmath122 ( [ def_theta ] ) : @xmath427 + \theta^2 l_{\theta}^2 \ ; , \label{j^4 } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_2 & = & { { \textstyle{\frac{1}{24 } } } } \theta^4 \;. \label{s2_2}\end{aligned}\ ] ] then , let us present a few more complicated examples of the results in terms of @xmath122 , @xmath428 + { { \mbox{cl}}_{2}\left(2 \theta\right ) } \left [ 2 l^2_\theta + { { \textstyle{\frac{1}{4 } } } } \theta^2 \right ] \nonumber \\ & & \hspace{15 mm } + 4 l_\theta { { \mbox{ls}}_{3}\left ( \theta\right ) } - 2 l_\theta { { \mbox{ls}}_{3}\left(2 \theta\right ) } + 4 { { \mbox{lsc}}_{2,3\!}\left(\theta\right ) } - { { \textstyle{\frac{1}{2 } } } } \theta^2 { { \mbox{cl}}_{2}\left(\theta\right ) } - { { \mbox{cl}}_{4}\left(\theta\right ) } \nonumber \\ & & \hspace{15 mm } + \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] \left [ 4 l_\theta - 28 l_\theta \right ] - 8 \theta \left [ { { \mbox{cl}}_{3}\left(\pi-\theta\right ) } - { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] \nonumber \\ & & \hspace{15 mm } + 8 \left[{{\mbox{ls}}_{4}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{4}\left(\pi\right)}\right ] - 16 \left[{{\mbox{cl}}_{4}\left(\pi-\theta\right ) } - { { \mbox{cl}}_{4}\left(\pi\right ) } \right ] \nonumber \\ & & \hspace{15 mm } + { { \textstyle{\frac{1}{6 } } } } \theta^3 l_\theta - { { \textstyle{\frac{2}{3 } } } } \theta^3 l_\theta - 8 \theta l^3_\theta + 13 \zeta_3 \theta + 4 \theta l_\theta l^2_\theta \biggr\ } \label{s1s1sb1_1 } \\ & & \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_1 \left [ \bar{s}_1 ^ 2 + \bar{s}_2 \right ] = \tan{{\textstyle{\frac{\theta}{2 } } } } \biggl\ { { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } \left [ 2 \theta^2 + 2 l^2_\theta - 20 l_\theta l_\theta + 32 l^2_\theta \right ] - { { \textstyle{\frac{1}{3 } } } } { { \mbox{ls}}_{4}\left(2\theta\right ) } \nonumber \\ & & \hspace{15 mm } + { { \mbox{cl}}_{2}\left(2\theta\right ) } \left [ { { \textstyle{\frac{1}{4 } } } } \theta^2 - 2 l_\theta l_\theta + 4 l^2_\theta \right ] + { { \mbox{ls}}_{3}\left(2 \theta\right ) } \left [ l_\theta - 4 l_\theta \right ] - 2 { { \mbox{ls}}_{3}\left(\theta\right ) } \left [ l_\theta - 5 l_\theta \right ] \nonumber \\ & & \hspace{15 mm } + \left[{{\mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } \right ] \left [ 10 l_\theta - 32 l_\theta \right ] - 6 \theta \left[{{\mbox{cl}}_{3}\left(\pi-\theta\right ) } - { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] - { { \mbox{cl}}_{4}\left(\theta\right ) } \nonumber \\ & & \hspace{15 mm } + \frac{22}{3 } \left[{{\mbox{ls}}_{4}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{4}\left(\pi\right ) } \right ] + 10 { { \mbox{lsc}}_{2,3\!}\left(\theta\right ) } - 12 { { \mbox{cl}}_{4}\left(\pi-\theta\right ) } + { { \textstyle{\frac{2}{3 } } } } { { \mbox{ls}}_{4}\left ( \theta\right ) } \nonumber \\ & & \hspace{15 mm } - { { \textstyle{\frac{1}{2 } } } } \theta^2 { { \mbox{cl}}_{2}\left(\theta\right ) } + 10 \zeta_3 \theta + { { \textstyle{\frac{1}{6 } } } } \theta^3 l_\theta - { { \textstyle{\frac{1}{2 } } } } \theta^3 l_\theta - 8 \theta l^3_\theta + 8
\theta l^2_\theta l_\theta - 2 \theta l_\theta l^2_\theta \biggr\ } \label{s1s2b_1 } \\ & & \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } \left [ \bar{s}_1 ^ 3 + 3 \bar{s}_1 \bar{s}_2 + 2 \bar{s}_3 \right ] = \tan{{\textstyle{\frac{\theta}{2 } } } } \biggl\ { 9 { { \mbox{cl}}_{2}\left(\pi-\theta\right ) } \left [ { { \textstyle{\frac{1}{4 } } } } \theta^2 + ( 2 l_\theta - l_\theta)^2 \right ] \nonumber \\ & & \hspace{15 mm } + { { \textstyle{\frac{3}{2 } } } } { { \mbox{cl}}_{2}\left(2\theta\right ) } \left [ { { \textstyle{\frac{1}{4 } } } } \theta^2 + ( 2 l_\theta - l_\theta ) ^2 \right ] + 18 { { \mbox{lsc}}_{2,3\!}\left(\theta\right ) } - { { \textstyle{\frac{3}{2 } } } } { { \mbox{cl}}_{4}\left(\theta\right ) } - { { \textstyle{\frac{3}{4 } } } } \theta^2 { { \mbox{cl}}_{2}\left(\theta\right ) } \nonumber \\ & & \hspace{15 mm } - 18 ( 2 l_\theta - l_\theta ) \left [ { { \mbox{ls}}_{3}\left(\pi-\theta\right ) } - { { \mbox{ls}}_{3}\left(\pi\right ) } - { { \textstyle{\frac{1}{2 } } } } { { \mbox{ls}}_{3}\left(\theta\right ) } + { { \textstyle{\frac{1}{6 } } } } { { \mbox{ls}}_{3}\left(2 \theta\right ) } \right ] + { { \textstyle{\frac{21}{2 } } } } \zeta_3 \theta \nonumber \\ & & \hspace{15 mm } \!+\ ! 6 \left [ { { \mbox{ls}}_{4}\left(\pi \!-\ !
\theta\right ) } \!-\ ! { { \mbox{ls}}_{4}\left(\pi\right ) } \right ] \!-\ !
6 \theta \left [ { { \mbox{cl}}_{3}\left(\pi \!-\ !
\theta\right ) } \!-\ ! { { \mbox{cl}}_{3}\left(\pi\right ) } \right ] \!-\ ! { { \mbox{ls}}_{4}\left(2\theta\right ) } \!+\ ! 3 { { \mbox{ls}}_{4}\left(\theta\right ) } \nonumber \\ & & \hspace{15 mm } \!-\ ! 12 { { \mbox{cl}}_{4}\left(\pi \!-\ !
\theta\right ) } \!+\ ! { { \textstyle{\frac{1}{4 } } } } \theta^3 l_\theta \!-\ ! { { \textstyle{\frac{1}{2 } } } } \theta^3 l_\theta \!+\ ! 8 \theta l^3_\theta \!-\ ! 6 \theta l^2_\theta l_\theta \!+\ ! 12 \theta l_\theta l^2_\theta \!-\ ! 8 \theta l^3_\theta \!-\ ! 7 \theta l^3_\theta \biggr\ } \;. \label{s3b_1}\end{aligned}\ ] ] using the analytic continuation procedure described in section 3 , these results can be rewritten in terms of the conformal variable @xmath224 defined in eq .
( [ y<->u ] ) .
below we list these analytically - continued results for eqs .
( [ j^1])([s2_2 ] ) , as well as for those sums from section 2 whose analytical continuations were not presented in section 3 : @xmath429 , \label{s1_y } \\
\sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } \bar{s}_1 & = & \frac{1-y}{1+y } \biggl [ { { \mbox{li}}_{2}\left(y\right ) } - 2 { { \mbox{li}}_{2}\left(-y\right ) } - 2 \ln y \ln(1+y ) + \ln y \ln ( 1-y ) \nonumber \\ & & + { { \textstyle{\frac{1}{4 } } } } \ln^2 y - 2 \zeta_2 \biggr ] , \label{sb1_y } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_1 ^ 2 & = & \frac{1-y}{1+y } \biggl [ 8 { { \mbox{s}}_{1,2\!}\left(-y\right ) } - 4 { { \mbox{li}}_{3}\left(-y\right ) } + 8 { { \mbox{li}}_{2}\left(-y\right ) } \ln(1+y ) + 4 \ln^2 ( 1+y ) \ln y \nonumber \\ & & - 2 \ln(1+y ) \ln^2 y + { { \textstyle{\frac{1}{6 } } } } \ln^3 y + 4 \zeta_2 \ln(1+y ) - 2 \zeta_2 \ln y - 4 \zeta_3 \biggr ] \ ; , \label{s1s1_y } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_1 \bar{s}_1 & \!\!=\!\ ! & \frac{1-y}{1+y } \biggl [ 10 { { \mbox{s}}_{1,2\!}\left(-y\right ) } - { { \mbox{s}}_{1,2\!}\left(y^2\right ) } + { { \mbox{li}}_{3}\left(y\right ) } + 2 { { \mbox{s}}_{1,2\!}\left(y\right ) } - 3 { { \mbox{li}}_{3}\left(-y\right ) } \nonumber \\ & & - 2 \ln ( 1-y ) { { \mbox{li}}_{2}\left(-y\right ) } + 8 \ln(1+y ) { { \mbox{li}}_{2}\left(-y\right ) } - 2 \ln(1+y ) { { \mbox{li}}_{2}\left(y\right ) } \nonumber \\ & & - 2 \ln y \ln ( 1-y ) \ln ( 1+y ) - \ln y \ln^2 ( 1-y ) + 4 \ln y \ln^2 ( 1+y ) \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \ln^2 y \ln(1-y ) - { { \textstyle{\frac{3}{2 } } } } \ln^2 y \ln(1+y ) + { { \textstyle{\frac{1}{12 } } } } \ln^3 y + 6 \zeta_2 \ln(1+y ) \nonumber \\ & & - \zeta_2 \ln(1-y ) - { { \textstyle{\frac{5}{2 } } } } \zeta_2 \ln y - { { \textstyle{\frac{11}{2 } } } } \zeta_3 \biggr ] \ ; , \label{s1sb1_y } \\
\sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_2 & = & - \frac{1-y}{6(1+y)}\ ; \ln^3 y \ ; , \label{s2_1_y } \\ \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j^2 } s_2 & = & { { \textstyle{\frac{1}{24 } } } } \ln^4 y \ ; , \label{s2_2_y }
\\ \label{an_s3_1 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_3 & = & \frac{1-y}{1+y } \bigl [ { { \textstyle{\frac{1}{24 } } } } \ln^4 y + 6 { { \mbox{li}}_{4}\left(y\right ) } + \ln^2 y { { \mbox{li}}_{2}\left(y\right ) } - 2 \zeta_3 \ln y - 4 \ln y { { \mbox{li}}_{3}\left(y\right ) } - 6 \zeta_4 \bigr ] \ ; , \nonumber \\ & & \hspace{10 mm } \\ \label{an_s1s2_1 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_1 s_2 & \!\!=\!\ ! & \frac{1\!-\!y}{1\!+\!y } \bigl [ { { \textstyle{\frac{1}{3 } } } } \ln^3 y \ln(1\!+\!y ) - \ ! { { \textstyle{\frac{1}{24 } } } } \ln^4 y + \ ! { { \textstyle{\frac{1}{2 } } } } \zeta_2 \ln^2 y + \ln^2 y { { \mbox{li}}_{2}\left(-y\right ) } + \ln^2 y { { \mbox{li}}_{2}\left(y\right ) } \nonumber \\ & & + \zeta_3 \ln y \!-\ ! 4 \ln y { { \mbox{li}}_{3}\left(-y\right ) } \!-\ ! 4 \ln y { { \mbox{li}}_{3}\left(y\right ) } \!+\ ! \zeta_4 + \ ! 8 { { \mbox{li}}_{4}\left(-y\right ) } + \ ! 6 { { \mbox{li}}_{4}\left(y\right ) } \bigr ] , \\ \label{an_s2sb1_1 } \sum_{j=1}^\infty \frac{1}{\left ( 2j \atop j \right ) } \frac{u^j}{j } s_2 \bar{s}_1 & \!\!=\!\ ! & \frac{1-y}{1+y } \bigl [ { { \textstyle{\frac{1}{3 } } } } \ln^3 y \ln ( 1+y ) - { { \textstyle{\frac{1}{6 } } } } \ln^3 y \ln ( 1-y ) - { { \textstyle{\frac{1}{48 } } } } \ln^4 y + \zeta_2 \ln^2 y + 4 \zeta_3 \ln y \nonumber \\ & & + \ln^2 y { { \mbox{li}}_{2}\left(-y\right ) } - 4 \ln y { { \mbox{li}}_{3}\left(-y\right ) } + 8 \zeta_4 + 8 { { \mbox{li}}_{4}\left(-y\right ) } - { { \mbox{li}}_{4}\left(y\right ) } \bigr ] \;.
\end{aligned}\ ] ] for the combination ( [ s2s1bar ] ) involving @xmath204 we get @xmath430 ^ 2 + \left ( { { \textstyle{\frac{1}{2 } } } } \ln^2 y - 4 \zeta_2 \right ) \left [ \ln ( 1-y ) - 2 \ln ( 1+y ) \right ] \biggr\ } \ ; , \label{sb2comb_y}\end{aligned}\ ] ] for a separate term of this sum involving @xmath101 , we get @xmath431 \ ; , \ ] ] where @xmath233 is defined in eq .
( [ def_omega_s ] ) .
the results for the _ multiple inverse binomial sums _ presented in this paper are summarized in table 1 .
when two equation numbers are present , the first one refers to the result in terms of the angular variable @xmath122 , whereas the second one corresponds to its analytical continuation .
an asterisk means that the corresponding equation holds for general @xmath106 .
the symbol @xmath432 means that the results for the sums involving the combinations @xmath433 and @xmath434 can be extracted from the expressions ( [ an_c1 ] ) and ( [ an_c2 ] ) given for the sums involving @xmath435 and @xmath436 , respectively , using the definitions ( [ combinations ] ) .
@xmath437 .equation index for the inverse binomial sums [ cols="^,^,^,^,^",options="header " , ] @xmath437
using the procedure described in section 2.2 , one can also construct the @xmath0-expansion of hypergeometric functions of the following types : @xmath438 @xmath439 where @xmath440 and @xmath48 . by analogy with eq .
( [ pfq2 ] ) , the @xmath0-expansion of these functions can be written in the following form : @xmath441 where the function @xmath442 is defined in the same way as in eq .
( [ expansion ] ) .
one should only remember that the upper summation limits for the coefficients @xmath443 and @xmath444 are changed , since the numbers of the parameters @xmath445 and @xmath110 in eqs .
( [ binomial ] ) and ( [ harmonic ] ) are different .
the sums appearing in eq .
( [ binomial ] ) are expressible in terms of the _ multiple binomial sums _ @xcite , whereas the sums of eq .
( [ harmonic ] ) are reduced to the _ multiple harmonic sums_. using relations between hypergeometric function of different arguments , it is possible to express one type of sums in terms of the another one , plus some trivial part .
let us illustrate this on the example of @xmath44 function ( [ 2f1_def ] ) . using the standard formula of analytic continuation to the argument @xmath446
, we obtain a combination of two @xmath44 functions . to bring them to the form of ( [ binomial ] ) , we need to shift some of the parameters using @xmath447 finally , using the duplication formula for the argument of the @xmath290-function , we arrive at the following relation : @xmath448 here , the @xmath0-expansion of the hypergeometric function on the l.h.s . can be expressed in terms of the _ multiple inverse binomial sums _ ( see eq .
( [ pfq2 ] ) ) , whereas the functions on the r.h.s .
yield the _ multiple binomial sums _
( see eq .
( [ binomial ] ) ) . furthermore , the following three quadratic relations for @xmath44 functions ( see , e.g. , in @xcite ) allow us to connect some _ multiple binomial sums _ ( appearing in eq .
( [ binomial ] ) ) with _ multiple harmonic sums _ ( see eq .
( [ harmonic ] ) ) : @xmath449 = ( 1\!+\!\chi)\ ; ~{}_{2}f_1 \left(\begin{array}{c| } 1 , { \varepsilon}\\ 2-{\varepsilon}\end{array } ~\chi\right ) , \nonumber \\ & & \hspace*{10 mm } \label{all - order}\end{aligned}\ ] ] where @xmath450 and all orders of the @xmath0-expansion of the last @xmath44 function on the r.h.s .
( [ all - order ] ) are known through eq .
( 2.14 ) of ref .
@xcite . combining all these relations together with the results for the _ multiple inverse binomial sums _ presented in this paper , we have reproduced several known results for _ multiple binomial sums _ , including those obtained in ref .
@xcite . for completeness , we list a number of such results , including trivial ones , which could be extracted from ref .
@xcite ( although were not explicitly listed there ) , @xmath451 , \nonumber \\ & & \sum\limits_{j=1}^\infty \left ( 2j \atop j\right ) \frac{z^j}{j } s_1 = { { \mbox{li}}_{2}\left(\chi^2\right ) } + 2\ln^2 ( 1+\chi ) , \nonumber \\ & & \sum\limits_{j=1}^\infty \left ( 2j \atop j\right ) z^j \bar{s}_1 = \frac{2}{1-\chi } \left [ \chi \ln(1+\chi ) - ( 1+\chi ) \ln(1-\chi ) \right ] , \nonumber \\ & & \sum\limits_{j=1}^\infty \left ( 2j \atop j\right ) \frac{z^j}{j } \bar{s}_1 = 2 { { \mbox{li}}_{2}\left(\chi\right ) } + \ln^2 ( 1+\chi ) , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_2 = -\frac{4 \chi}{1-\chi } { { \mbox{li}}_{2}\left(-\chi\right ) } + 2 \ln^2 ( 1+\chi ) , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 ^ 2 = \frac{2}{1\!-\!\chi } \biggl [ 4 \chi { { \mbox{li}}_{2}\left(\chi\right ) } + 2 \chi { { \mbox{li}}_{2}\left(-\chi\right ) } - ( 1\!-\ ! \chi ) \ln^2 ( 1\!+\!\chi ) + 2 ( 1\!+\!\chi ) \ln^2 ( 1\!-\!\chi ) \biggr ] , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 \bar{s}_1 = \frac{1}{1-\chi } \biggl [ 6 \chi { { \mbox{li}}_{2}\left(\chi\right ) } - 2(1+\chi ) \ln ( 1+\chi ) \ln(1-\chi ) \nonumber \\ & & \hspace{5 cm } - ( 1-\chi ) \ln^2 ( 1+\chi ) + 4 ( 1+\chi ) \ln^2 ( 1-\chi ) \biggr ] , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j \left ( \bar{s}_2 - \bar{s}_1 ^ 2 \right ) = -\frac{2}{1-\chi } \biggl [ 2 \chi { { \mbox{li}}_{2}\left(\chi\right ) } - 2(1+\chi ) \ln(1+\chi ) \ln(1-\chi ) \nonumber \\ & & \hspace{6 cm } + \chi \ln^2 ( 1+\chi ) + 2 ( 1+\chi ) \ln^2 ( 1-\chi ) \biggr ] . \label{mbs - list}\end{aligned}\ ] ] furthermore , explicit results for the _ multiple binomial sums _ with @xmath452 , @xmath453 , @xmath454 , @xmath455 , @xmath456 , @xmath457 , and @xmath458 are presented in eqs .
( a.4)(a.10 ) of ref .
@xcite ( where the same notation @xmath459 as here was used , with @xmath460 corresponding to our @xmath276 ) .
we confirm all those results .
moreover , using our approach we have obtained a number of new results for the _ multiple binomial sums _ , @xmath461 \ ; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 s_2 = { { \textstyle{\frac{4}{3 } } } } \ln^3 ( 1+\chi ) + 4 \ln(1+\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } + 4 \ln(1+\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \nonumber \\ & & \hspace{20 mm } + \frac{4}{1-\chi } \left [ \chi { { \mbox{li}}_{3}\left(\chi\right ) } + { { \mbox{s}}_{1,2}\left(\chi^2\right ) } - 2 { { \mbox{s}}_{1,2}\left(\chi\right ) } + ( 1+\chi ) \ln(1-\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } \right ] \
; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_2 \bar{s}_1 = { { \textstyle{\frac{2}{3 } } } } \ln^3 ( 1+\chi ) + 4 \ln ( 1+\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \nonumber \\ & & \hspace{20 mm } + \frac{2}{1-\chi } \left [ 2 \chi { { \mbox{li}}_{3}\left(\chi\right ) } - 6 { { \mbox{s}}_{1,2}\left(-\chi\right ) } + 2 { { \mbox{s}}_{1,2}\left(\chi^2\right ) } - 4 { { \mbox{s}}_{1,2}\left(\chi\right ) } - \chi { { \mbox{li}}_{3}\left(-\chi\right ) } \right ] \nonumber \\ & & \hspace{20 mm } + \frac{2(1+\chi)}{1-\chi } \left [ 2 \ln(1-\chi ) - \ln(1+\chi ) \right ] { { \mbox{li}}_{2}\left(-\chi\right ) } \ ; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 \left(\bar{s}_2 - \bar{s}_1 ^ 2 \right ) = \frac{2}{1-\chi } \left [ 16 { { \mbox{s}}_{1,2}\left(\chi\right ) } - 2 \chi { { \mbox{li}}_{3}\left(\chi\right ) } - 6 \chi \ln(1+\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \right ] \nonumber \\ & & \hspace{20 mm } + \frac{2(1+\chi)}{1-\chi } \left [ 4 \ln^3(1-\chi ) - 4 \ln^2(1-\chi ) \ln(1+\chi ) \right . \nonumber \\ & & \hspace{40 mm } \left .
+ \ln(1-\chi ) \ln^2(1+\chi ) + 8 \ln(1-\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \right ] + { { \textstyle{\frac{2}{3 } } } } \ln^3(1+\chi ) \ ; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 ^ 2 \bar{s}_1 = - \frac{2}{1-\chi } \left [ 2 { { \mbox{s}}_{1,2}\left(\chi^2\right ) } - 6 { { \mbox{s}}_{1,2}\left(-\chi\right ) } + 16 { { \mbox{s}}_{1,2}\left(\chi\right ) } - \chi { { \mbox{li}}_{3}\left(-\chi\right ) } \right .
\nonumber \\ & & \hspace{40 mm } \left . - 4 \chi { { \mbox{li}}_{3}\left(\chi\right ) } + 2 ( 1 - 3\chi ) \ln(1+\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \right ] \nonumber \\ & & \hspace{20 mm } - \frac{2(1+\chi)}{1-\chi } \left [ 4 \ln^3(1-\chi ) - 2 \ln^2(1-\chi ) \ln(1+\chi ) + 2 \ln(1-\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } \right .
\nonumber \\ & & \hspace{40 mm } \left .
+ 10 \ln(1-\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } - \ln(1+\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } \right ] - { { \textstyle{\frac{2}{3 } } } } \ln^3(1+\chi ) \ ; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j \left ( 3 \bar{s}_1 \bar{s}_2 - 2 \bar{s}_3 - \bar{s}_1 ^ 3 \right ) = \frac{2}{1\!-\!\chi } \left [ 12 { { \mbox{s}}_{1,2}\left(\chi\right ) } \!-\ ! 6 \chi \ln(1\!+\!\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \!-\ !
\chi \ln^3(1\!+\!\chi ) \right ] \nonumber \\ & & \hspace{20 mm } + \frac{2(1+\chi)}{1-\chi } \left [ 6 \ln(1-\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } - 6 \ln^2(1-\chi ) \ln(1+\chi ) \right .
\nonumber \\ & & \hspace{40 mm } \left .
+ 3 \ln(1-\chi ) \ln^2(1+\chi ) + 4 \ln^3(1-\chi ) \right ] \ ; , \nonumber \\ & & \sum_{j=1}^\infty \left ( 2j \atop j\right ) z^j s_1 ^ 3 = - { { \textstyle{\frac{4}{3 } } } } \ln^3(1+\chi ) - 4 \ln(1+\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } - 12 \ln(1+\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \nonumber \\ & & \hspace{20 mm } - \frac{4}{1-\chi } \left [ 3 { { \mbox{s}}_{1,2}\left(\chi^2\right ) } - 4{{\mbox{s}}_{1,2}\left(-\chi\right ) } + 6 { { \mbox{s}}_{1,2}\left(\chi\right ) } - 2 \chi { { \mbox{li}}_{3}\left(-\chi\right ) } - 3 \chi { { \mbox{li}}_{3}\left(\chi\right ) } \right ] \nonumber \\ & & \hspace{20 mm } - \frac{4(1+\chi)}{1-\chi } \left [ 2 \ln^3(1-\chi ) + 3 \ln(1-\chi ) { { \mbox{li}}_{2}\left(-\chi\right ) } + 6 \ln(1-\chi ) { { \mbox{li}}_{2}\left(\chi\right ) } \right ] \ ; , \ ] ] where @xmath459 is defined in eq .
( [ def_chi ] ) and , as before , @xmath48 .
these results were recently used to construct the @xmath0-expansion of two - loop sunset - type diagrams of special type ( for details , see section 3.1 in @xcite ) .
the results for harmonic sums of the type @xmath462 could be deduced from the following relation : @xmath463 which corresponds to an analytic continuation of the @xmath44 function from the argument @xmath276 to @xmath464 . in particular
, we get @xmath465 \nonumber \\ & & + \frac{1}{4 \sqrt{z } } \left[\ln^2 ( 1-\sqrt{z } ) - \ln^2 ( 1+\sqrt{z } ) \right ] - \frac{\ln 2}{2 \sqrt{z } } \ln \left(\frac{1+\sqrt{z}}{1-\sqrt{z } } \right ) \biggr ] \biggr\}. \hspace*{10 mm } \label{representaion : xi}\end{aligned}\ ] ] introducing a new variable @xmath466 such that @xmath467 we can present the above result in a more compact form , @xmath468 + \frac{1\!-\!\xi^2}{4 \xi } \left [ { { \mbox{li}}_{2}\left(-\xi\right ) } + \ln{\xi } \ln{2 } + { { \textstyle{\frac{1}{2 } } } } \zeta_2 \right ] + \frac{1\!-\!\xi}{8\xi } \ln^2{\xi } \ ; , \label{representaion : xi:2}\ ] ] where we have taken into account that @xmath469 integrating the representation ( [ representaion : xi:2 ] ) , it is easy to get the following result : @xmath470 in appendix e of ref .
@xcite the results ( up to weight * 4 * ) for the harmonic sums of the type @xmath471 have been presented .
it is easy to extend those results to the case @xmath472 , using the following property @xcite : @xmath473\;.\ ] ] as a consequence , we get @xmath474 \ ; .\end{aligned}\ ] ] as a further illustration , we also present results for other sums up to weight * 3 * , @xmath475 - \frac{\sqrt{z}}{2(1\!+\!\sqrt{z } ) } \left [ { { \mbox{li}}_{2}\left ( -\sqrt{z}\right ) } + \ln^2 ( 1+\sqrt{z } ) \right ] \ ; , \nonumber \\ \sum_{j=1}^\infty \bar{s}_2 \frac{z^j}{j } \!\ ! &
= & \!\ !
-\ln ( 1-\sqrt{z } ) { { \mbox{li}}_{2}\left(\sqrt{z}\right ) } -\ln ( 1+\sqrt{z } ) { { \mbox{li}}_{2}\left(-\sqrt{z}\right ) } - 2 { { \mbox{s}}_{1,2}\left(\sqrt{z}\right ) } - 2 { { \mbox{s}}_{1,2}\left(-\sqrt{z}\right ) } \ ; , \nonumber \\ \sum_{j=1}^\infty \bar{s}_1 ^ 3 z^j \!\ ! & = & \!\ ! \frac{\sqrt{z}}{2(1-\sqrt{z } ) } \left [ { { \mbox{li}}_{3}\left(\sqrt{z}\right ) } - 3 \ln(1-\sqrt{z } ) { { \mbox{li}}_{2}\left(\sqrt{z}\right ) } + \ln^3(1-\sqrt{z } ) - 3 { { \mbox{s}}_{1,2}\left(\sqrt{z}\right ) } \right ] \nonumber \\ & & \hspace*{-3 mm } - \frac{\sqrt{z}}{2(1\!+\!\sqrt{z } ) } \left [ { { \mbox{li}}_{3}\left(-\sqrt{z}\right ) } \!-\ ! 3 \ln(1\!+\!\sqrt{z } ) { { \mbox{li}}_{2}\left(-\sqrt{z}\right ) } \!+\ ! \ln^3(1\!+\!\sqrt{z } ) \!-\ !
3 { { \mbox{s}}_{1,2}\left(-\sqrt{z}\right ) } \right ] , \nonumber \\ \sum_{j=1}^\infty \bar{s}_1 ^ 2 \frac{z^j}{j } \!\ ! & = & \!\ ! - \ln(1-\sqrt{z } ) { { \mbox{li}}_{2}\left(\sqrt{z}\right ) } - \ln(1+\sqrt{z } ) { { \mbox{li}}_{2}\left(-\sqrt{z}\right ) } - 2 { { \mbox{s}}_{1,2}\left(\sqrt{z}\right ) } - 2 { { \mbox{s}}_{1,2}\left(-\sqrt{z}\right ) } \nonumber \\ & & + { { \textstyle{\frac{1}{3 } } } } \ln^3(1-\sqrt{z } ) + { { \textstyle{\frac{1}{3 } } } } \ln^3(1+\sqrt{z } ) \ ; , \nonumber \\ \sum_{j=1}^\infty \bar{s}_1 \bar{s}_2 z^j \!\ ! & = & \!\ ! \frac{\sqrt{z}}{2(1-\sqrt{z } ) } \left [ { { \mbox{li}}_{3}\left(\sqrt{z}\right ) } - \ln(1-\sqrt{z } ) { { \mbox{li}}_{2}\left(\sqrt{z}\right ) } - { { \mbox{s}}_{1,2}\left(\sqrt{z}\right ) } \right ] \nonumber \\ & & -\frac{\sqrt{z}}{2(1+\sqrt{z } ) } \left [ { { \mbox{li}}_{3}\left(-\sqrt{z}\right ) } - \ln(1+\sqrt{z } ) { { \mbox{li}}_{2}\left(-\sqrt{z}\right ) } - { { \mbox{s}}_{1,2}\left(-\sqrt{z}\right ) } \right ] \;. \ ] ] to investigate relations between multiple harmonic sums the technique proposed in ref
. @xcite can be useful . as a illustration
, we present the hihger order @xmath0-expansion of the some of the hypergeometric functions : @xmath476 { { \mbox{s}}_{1,2\!}\left(z\right ) } + \left [ ( a_1+a_2)c - c^2 - a_1 a_2 \right ] \ln(1-z ) { { \mbox{li}}_{2}\left(z\right ) } \nonumber \\ & & \hspace{10 mm } + c^2 { { \mbox{li}}_{3}\left(z\right ) } - \frac{(c - a_1-a_2)^2}{6 } \ln^3 ( 1-z ) \biggr\ } \nonumber \\ & & - { \varepsilon}^3 \biggl\ { c \left [ ( a_1+a_2)c - c^2 - 2 a_1 a_2 \right ] { { \mbox{s}}_{2,2\!}\left(z\right ) } + c \left [ ( a_1+a_2)c - c^2 - a_1 a_2 \right ] \ln(1-z ) { { \mbox{li}}_{3}\left(z\right ) } \nonumber \\ & & \hspace{10 mm } + ( c - a_1 ) ( c - a_2 ) ( c - a_1-a_2 ) \left [ \ln(1-z ) { { \mbox{s}}_{1,2\!}\left(z\right ) } + \frac{1}{2 } \ln^2 ( 1-z ) { { \mbox{li}}_{2}\left(z\right ) } \right ] \nonumber \\ & & \hspace{10 mm } + \frac{(c - a_1-a_2)^3}{24 } \ln^4 ( 1-z ) + c ( c - a_1 - a_2)^2 { { \mbox{s}}_{1,3\!}\left(z\right ) } + c^3 { { \mbox{li}}_{4}\left(z\right ) } \biggr\ } + { \cal o } ( { \varepsilon}^4 ) \biggr ) \ ; .\end{aligned}\ ] ]
here we present an example of a physically relevant calculation that can be expressed in terms of the master integrals @xmath258 studied in section 4.3 .
let us consider the two - loop propagator - type diagrams shown in fig .
all of these @xmath477 contributions to the polarization function of the gauge bosons involve a quark loop with a gluon exchange .
it was analytically calculated in @xcite , up to the finite term of the @xmath0-expansion . here
we present the bare two - loop results in @xmath478-dimensional space - time ( see also in ref .
in contrast to the calculations performed in @xcite , here we use tarasov s recurrence relations @xcite for the reduction of the original integrals to the set of master integrals . in this appendix
we use the euclidean notation @xcite , @xmath479 , so that the on - shell limit would read @xmath480 .
let us decompose the polarization tensor into the transverse @xmath481 and longitudinal @xmath482 parts , @xmath483 then , the two - loop corrections can be written as @xmath484 \nonumber \\ & & - j_{011}(1,1,2 ) \frac { 4}{\left ( n-4\right ) \left ( n-3\right ) \left ( n-1\right ) t } \bigl [ 112 { a } { { m}}^{4}n + { a}{t}^{2}{n}^{3 } -7 { a}{t}^{2}{n}^{2 } + 18 { a}{t}^{2}n \nonumber \\ & & + s{t}^{2}{n}^{3 } -7s{t}^{2}{n}^{2 } + 18s{t}^{2}n -16 { a } { { m}}^{4}{n}^{2 } -192 { a } { { m}}^{4 } + 4 { a } { { m}}^{2}t{n}^{3 } -32 { a } { { m}}^{2}t{n}^{2 } -16 { a}{t}^{2 } \nonumber \\ & & + 100 { a } { { m}}^{2}tn + 4s { { m}}^{2}t{n}^{3 } -36s { { m}}^{2}t{n}^{2 } + 96s { { m}}^{2}tn -112 { a } { { m}}^{2}t -16s{t}^{2 } -80s { { m}}^{2}t \bigr ] \nonumber \\ & & + \left [ a_0(m ) \right ] ^2 { \frac { 2 \left ( n-2\right ) } { \left ( 4 { { m}}^{2}+t\right ) { { m}}^{2}t\left ( n-4\right ) \left ( n-1\right ) \left ( n-3\right ) } } \bigl [ -4st{n}^{4 } { { m}}^{2 } + { a}{t}^{2}{n}^{3 } + 4 { a } { { m}}^{2}t{n}^{3 } \nonumber \\ & & + 36s { { m}}^{2}t{n}^{3 } + s{t}^{2}{n}^{3 } -32 { a } { { m}}^{2}t{n}^{2 } -7s{t}^{2}{n}^{2 } -16 { a } { { m}}^{4}{n}^{2 } -120s { { m}}^{2}t{n}^{2 } -7 { a}{t}^{2}{n}^{2 } \nonumber \\ & & + 112 { a } { { m}}^{4}n + 184s { { m}}^{2}tn + 100 { a } { { m}}^{2}tn + 18 { a}{t}^{2}n + 18s{t}^{2}n -192 { a } { { m}}^{4 } -16 { a}{t}^{2 }
\nonumber \\ & & -112s { { m}}^{2}t -112 { a } { { m}}^{2}t -16s{t}^{2 } \bigr ] \nonumber \\ & & - a_0(m ) b_0(m , m , t ) { \frac { 2 \left ( n-2\right ) } { { { m}}^{2}\left ( n-3\right ) \left ( 4 { { m}}^{2}+t\right ) \left ( n-4\right ) \left ( n-1\right ) } } \bigl [ 432 { a } { { m}}^{4}n -2 { a}t{n}^{4 } { { m}}^{2 } \nonumber \\ & & + 64 { { m}}^{4}s + 40 { { m}}^{4}sn + 8 { { m}}^{4}s{n}^{3 } -48 { { m}}^{4}s{n}^{2 } + 80 { a } { { m}}^{4}{n}^{3 } -8 { a}{n}^{4 } { { m}}^{4 } -2st{n}^{4 } { { m}}^{2 } \nonumber \\ & & + { a}{t}^{2}{n}^{3 } -7 { a}{t}^{2}{n}^{2 } + 18 { a}{t}^{2}n + s{t}^{2}{n}^{3 } -7s{t}^{2}{n}^{2 } + 18s{t}^{2}n -280 { a } { { m}}^{4}{n}^{2 } -256 { a } { { m}}^{4 } \nonumber \\ & & + 24 { a } { { m}}^{2}t{n}^{3 } -98 { a } { { m}}^{2}t{n}^{2 } + 180 { a } { { m}}^{2}tn + 24s { { m}}^{2}t{n}^{3 } -102s { { m}}^{2}t{n}^{2 } + 176s { { m}}^{2}tn \nonumber \\ & & -16 { a}{t}^{2 } -128 { a } { { m}}^{2}t -16s{t}^{2 } -96s { { m}}^{2}t \bigr ] \nonumber \\ & & - \left [ b_0(m , m , t ) \right]^2 { \frac { 2}{\left ( 4 { { m}}^{2}+t\right ) \left ( n-4\right ) \left ( n-1\right ) } } \bigl [ 448 { a } { { m}}^{4}n -32 { { m}}^{4}s + 16 { a } { { m}}^{4}{n}^{3 } \nonumber \\ & & + { a}{t}^{2}{n}^{3 } -9 { a}{t}^{2}{n}^{2 } + 30 { a}{t}^{2}n + s{t}^{2}{n}^{3 } -9s{t}^{2}{n}^{2 } + 30s{t}^{2}n -144 { a } { { m}}^{4}{n}^{2 } -448 { a } { { m}}^{4 } \nonumber \\ & & + 8 { a } { { m}}^{2}t{n}^{3 } -72 { a } { { m}}^{2}t{n}^{2 } + 232 { a } { { m}}^{2}tn + 4s { { m}}^{2}t{n}^{3 } -44s { { m}}^{2}t{n}^{2 } + 152s { { m}}^{2}tn \nonumber \\ & & -32 { a}{t}^{2 } -240 { a } { { m}}^{2}t -32s{t}^{2 } -160s { { m}}^{2}t \bigr ] \biggr\ } \ ; , \label{pt } \\ & & \hspace*{-10 mm } \pi_{\rm l}^{(2)}(p^2 ) = a \frac{g^2 g_{\rm s}^2}{(4 \pi ) ^{n/2 } } n_c c_f \biggl\ { - j_{011}(1,1,1 ) \frac { 16 m^{2}\left ( n-3\right ) } { t } \nonumber \\ & & - j_{011}(1,1,2 ) \frac { 16 m^{2}}{t\left ( n-4\right ) \left ( n-3\right ) } \bigl [ 4 m^{2}{n}^{2 } -28 m^{2}n + 48 m^{2 } + t{n}^{2}-5tn+8 t \bigr ] \nonumber \\ & & + \left [ a_0(m ) \right ] ^2 \frac{8 \left ( n-2\right ) } { t\left ( 4 m^{2}+t \right ) \left ( n-4\right ) } \bigl [ t{n}^{2}-3tn+4 { { m}}^{2}n-16 { { m}}^{2 } \bigr ] \nonumber \\ & & - \left [ b_0(m , m , t ) \right]^2 { \frac { 8 { { m}}^{2}}{\left ( 4 { { m}}^{2}+t\right ) \left ( n-4\right ) } } \bigl [ -4tn+4t+4 { { m}}^{2}{n}^{2}-24 { { m}}^{2}n+t{n}^{2}+40{{m}}^{2 } \bigr ] \nonumber \\ & & - a_0(m ) b_0(m , m , t ) \frac { 8 \left ( n - 2 \right ) \left ( { n}^{2}-5n+8\right ) } { \left(4 m^{2}+t \right ) ( n-4 ) ( n-3 ) } \bigl[-2 { { m}}^{2}n+10 { { m}}^{2}+t \bigr ] \biggr\ } \ ; , \label{pl}\end{aligned}\ ] ] where the vertices @xmath485 ( @xmath486 ) are defined as and @xmath487 for the standard model can be extracted from @xcite . ]
@xmath488 and we have introduced the following notations : @xmath489 @xmath490 is a color factor ( equal to 3 for quark and 1 for lepton ) , @xmath491 is the casimir operator of the fundamental representation of the lie algebra ( equal to @xmath492 for @xmath493 and 1 for qed ) , and @xmath261 is the mass of the loop fermion . finally , the occurring integrals are defined as @xmath494^{\sigma } \left [ k_1 ^ 2+m^2 \right]^{\nu_1 } \left [ ( k_1-k_2)^2 + m^2 \right]^{\nu_2 } } \ ; , \nonumber \\ b_0(m_a , m_b , t ) & = & \frac{\pi^{-n/2}}{\gamma\left(3-{{\textstyle{\frac{n}{2}}}}\right ) } \int \frac{{\rm d}^n k } { \left[(k - p)^2+m_a^2\right ] \left [ k^2+m_b^2 \right ] } \ ; , \nonumber \\
a_0(m ) & = & \frac{\pi^{-n/2}}{\gamma\left(3-{{\textstyle{\frac{n}{2}}}}\right ) } \int \frac{{\rm d}^n k}{k^2+m^2 } \equiv \frac{4 ( m^2)^{n-2}}{(n-2)(n-4 ) } \;.\end{aligned}\ ] ] we note that the integral @xmath258 is defined in the same way as in section 4.3 , one should only remember to substitute @xmath495 . in particular ,
for the zero momentum transfer , @xmath496 , we get @xmath497 ^ 2 \frac{4(n-2)(n^2 - 5n+7)}{m^2 } \ ; , \nonumber \\
\pi_{\rm t}^{(2)}(0 ) & = & - a \frac{g^2 g_s^2}{(4 \pi)^{n/2 } } n_c c_f \left [ a_0(m ) \right]^2 \frac{4(n-2)(n^3 - 6n^2 + 13n-11)}{(n-1)m^2 } \;. \nonumber \end{aligned}\ ] ] to obtain the finite terms of the @xmath0-expansion of the results given in eqs .
( [ pt ] ) and ( [ pl ] ) , the integral @xmath266 should be expanded up to the @xmath0-part . furthermore , using the approach of ref .
@xcite , we obtain the following relations between the integrals investigated in section 4.3 and the master integrals @xmath265 and @xmath266 : @xmath498 ^ 2 ( n-3)(n-2)^2 \biggl [ \frac{t}{4m^4 } - \frac{(7n-24)}{2m^2 ( n-4 ) } \biggr ] \nonumber \\ & & + j_{011}(1,2,2 ) \biggl [ \frac{8 m^2 ( t+m^2)}{n-4 } - 2 t^2 ( n-3 ) + 2 t m^2 ( 7n-17 ) + 8 m^4 ( 2n-5 ) \biggr ] \nonumber \\ & & + \left [ j_{011}(1,2,2 ) + 2 j_{011}(2,1,2 ) \right ] ( n-3 ) ( t+4m^2 ) ( t-2 m^2 ) \label{j111 } \ ; , \\ & & \hspace*{-10 mm } j_{011}(1,1,2)\frac{(3n-8)(3n-10)(n-3)^2}{n-4 } = \left [ a_0(m ) \right]^2 \frac{(n-3)(2n-7)(3n-8)(n-2)^2}{4m^4 ( n-4 ) } \nonumber \\ & & + j_{011}(1,2,2 ) \biggl [ -\frac { 4 ( t+2m^2)}{n-4 } - n^2
\left ( { { \textstyle{\frac{15}{2}}}}t + 12m^2 \right ) + n \left ( { { \textstyle{\frac{79}{2}}}}t+62m^2 \right ) - 55 t - 86 m^2 \biggr ] \nonumber \\ & & + { { \textstyle{\frac{1}{2 } } } } \left [ j_{011}(1,2,2 ) + 2 j_{011}(2,1,2 ) \right ] ( n-3 ) ( 3n-8)(t+4m^2 ) \;. \end{aligned}\ ] ] from these relations we see that , in order to obtain the @xmath499 terms of @xmath265 and @xmath266 , the integral @xmath267 should be expanded up to @xmath499 , whereas the combination @xmath272 $ ] up to @xmath500 only .
the expansion of the integrals near the threshold can be performed by using the numerical algorithm described in ref .
@xcite .
the higher - order terms of the @xmath0-expansion of the one - loop integral @xmath501 can be extracted from refs .
@xcite . here
we present coefficients up to the order @xmath271 for the particular case @xmath502 : @xmath503 \nonumber \\ & & + { \varepsilon}^2 \bigl [ 4 { { \mbox{s}}_{1,2\!}\left(-y\right ) } - 2 { { \mbox{li}}_{3}\left(-y\right ) } + 4 \ln(1+y ) { { \mbox{li}}_{2}\left(-y\right ) } - \ln^2 y \ln(1+y ) \nonumber \\ & & + 2 \ln y \ln^2 ( 1+y ) + { { \textstyle{\frac{1}{6 } } } } \ln^3 y - \zeta_2 \ln y + 2 \zeta_2 \ln(1+y ) - 2\zeta_3 \bigr ] + { \cal o}({\varepsilon}^3 ) \biggr\ } \biggr ) \ ; , \hspace*{6mm}\end{aligned}\ ] ] where the variable @xmath224 is defined in section 3 .
the bare two - loop amplitudes ( [ pt ] ) and ( [ pl ] ) contain subdivergencies which should be canceled by proper counterterms , @xmath504 where @xmath505 is the one - loop amplitude , while @xmath506 is the one - loop mass counterterm defined in a particular renormalization scheme .
the derivatives of the bare one - loop amplitudes in @xmath478 dimensions read @xmath507b_0(m , m , t ) \!+\ !
\frac{2 s ( n\!-\!2 ) } { ( 4m^2\!+\!t ) } a_0(m ) \biggr \ } \ ; , \nonumber \\ \frac{\partial}{\partial m^2 } \pi_{\rm l}^{(1)}(p^2 ) & \!\!=\!\ ! & a \frac{g^2}{(4 \pi ) ^{n/2 } } n_c 4 \biggl \ { \biggl [ \frac{t ( n-3 ) } { ( 4m^2+t ) } + 1-n \biggr]b_0(m , m , t ) + \frac{2 ( 2-n ) } { ( 4m^2+t ) } a_0(m ) \biggr \ } \;. \nonumber \end{aligned}\ ] ] as an example of application of these formulae , let us consider the transversal part of the @xmath508 propagator @xmath509 in the @xmath510-scheme .
up to the finite in @xmath0 part , the result for the subtracted quantity @xmath511 is @xmath512 \nonumber \\ & & - { { \textstyle{\frac{16}{3 } } } } m^2 \frac{(1 - 4y+y^2)}{y(1-y)^2 } \left [ \ln(1-y ) + 2 \ln(1+y ) \right ] \ln y \left [ ( 1+y^2 ) \ln y - 2 ( 1-y^2 ) \right ] \nonumber \\ & & + { { \textstyle{\frac{8}{3 } } } } m^2 \frac{(2 + 7y-22y^2 + 6y^3)}{(1-y)^2 } \ln^2 y - 4 m^2 \ln y \frac{(1 - 6y-46y^2 - 6y^3+y^4)}{y(1-y^2 ) } \nonumber \\ & & + 4 t \ln \frac{m^2}{\mu^2 } \left [ 1- \frac{12 y}{(1-y)^2 } \right ] - 96 m^2 \frac{y}{1-y^2 } \ln y \ln \frac{m^2}{\mu^2 } \nonumber \\ & & + { { \textstyle{\frac{32}{3 } } } } m^2 \frac{(1 - 4y+y^2)}{y(1-y)^2 } \left [ { { \mbox{li}}_{2}\left(y\right ) } + 2 { { \mbox{li}}_{2}\left(-y\right ) } \right ] \left [ 1-y^2 - 2\left(1+y^2\right ) \ln y \right ] \nonumber \\ & & + 32 m^2 \frac{(1 - 4y+y^2)}{y(1-y)^2 } ( 1+y^2 ) \left [ { { \mbox{li}}_{3}\left(y\right ) } + 2 { { \mbox{li}}_{3}\left(-y\right ) } \right ] + { \cal o}({\varepsilon } ) \biggr\ } \ ; , \end{aligned}\ ] ] where we have taken into account the one - loop massive counterterm @xmath513 the small - momentum expansion of eq . ( [ gammaz ] )
is @xmath514 where @xmath515 .
let us remind that in a theory with the spontaneous symmetry breaking , like the standard model , the inclusion of tadpoles @xcite is also required for the renormalization group invariance of the massive parameters @xcite .
the proper bare two - loop tadpole contribution is given in section 4.3 of ref .
@xcite .
99 g. thooft and m. veltman , nucl .
* b44 * ( 1972 ) 189 ; + c.g .
bollini and j.j .
giambiagi , nuovo cim . * 12b * ( 1972 ) 20 ; + j.f .
ashmore , lett .
nuovo cim .
* 4 * ( 1972 ) 289 ; + g.m .
cicuta and e. montaldi , lett .
nuovo cim .
* 4 * ( 1972 ) 329 .
l. lewin , _ polylogarithms and associated functions _ ( north - holland , amsterdam , 1981 ) . k.s .
klbig , j.a .
mignaco and e. remiddi , b.i.t .
* 10 * ( 1970 ) 38 ; + r. barbieri , j.a .
mignaco and e. remiddi , nuovo cim . * a11 * ( 1972 ) 824 ; + a. devoto and d.w .
duke , riv .
nuovo cim .
* 7 * , no.6 ( 1984 ) 1 ; + k.s .
klbig , siam j. math .
17 * ( 1986 ) 1232 .
e. remiddi and j.a.m .
vermaseren , int .
j. mod .
* a15 * ( 2000 ) 725 . a.b .
goncharov , math .
lett . * 5 * ( 1998 ) 497 ; + j.m .
borwein , d.m .
bradley , d.j . broadhurst and p.
lisonk , trans .
soc . * 353 * ( 2001 ) 907 ; + t. gehrmann and e. remiddi , nucl .
* b601 * ( 2001 ) 248 . s. moch , p. uwer and s. weinzierl , j. math . phys . * 43 * ( 2002 ) 3363 . s. weinzierl , hep - ph/0402131 .
f. jegerlehner , m.yu .
kalmykov and o. veretin , nucl .
* b658 * ( 2003 ) 49 .
kotikov , phys .
* b254 * ( 1991 ) 158 ; * b259 * ( 1991 ) 314 .
boos and a.i .
davydychev , teor . mat .
* 89 * ( 1991 ) 56 .
isaev , nucl .
* b662 * ( 2003 ) 461 .
j. fleischer , a.v .
kotikov and o.l .
veretin , phys .
* b417 * ( 1998 ) 163 .
davydychev and r. delbourgo , j. math .
* 39 * ( 1998 ) 4299 .
j. fleischer , a.v .
kotikov and o.l .
veretin , nucl .
* b547 * ( 1999 ) 343 .
davydychev and m.yu .
kalmykov , nucl .
b ( proc .
* 89 * ( 2000 ) 283 .
davydychev and m.yu .
kalmykov , nucl .
* b605 * ( 2001 ) 266 .
j. fleischer , f. jegerlehner and o.v .
tarasov , nucl .
* b672 * ( 2003 ) 303 .
j. fleischer , m.yu .
kalmykov and a.v .
kotikov , phys .
* b462 * ( 1999 ) 169 ; * b467 * ( 1999 ) 310(e ) .
kazakov and a.v .
kotikov , theor .
phys . * 73 * ( 1988 ) 1264 ; nucl .
* b307 * ( 1988 ) 721 ; + d. borwein , j.m .
borwein and r. girgensohn , proc .
* 38 * ( 1995 ) 277 ; + j.m .
borwein and r. girgensohn , electron .
j. combin .
* 3 * ( 1996 ) r23 ; + p. flajolet and b. salvy , experimental math .
* 7 * ( 1998 ) 15 ; + o.m .
ogreid and p. osland , j. comput .
* 98 * ( 1998 ) 245 ; * 136 * ( 2001 ) 389 ; + v.a .
smirnov , phys .
* b547 * ( 2002 ) 239 .
kalmykov and o. veretin , phys .
* b483 * ( 2000 ) 315 ; + j.m .
borwein and r. girgensohn , _
`` evaluation of binomial series '' _ , cecm preprint cecm-02 - 188 ; available at http://www.cecm.sfu.ca/preprints/2002pp.html . j. fleischer and m.yu .
kalmykov , phys .
* b470 * ( 1999 ) 168 .
broadhurst , open university preprint out-4102 - 62 ( hep - th/9604128 ) .
broadhurst , eur .
j. * c8 * ( 1999 ) 311 .
davydychev and m.yu .
kalmykov , proc .
workshop `` cpp2001 '' , tokyo , japan , november 2001 , kek proceedings 2002 - 11 , p. 169
( hep - th/0203212 ) .
n. batir , appl .
* 147 * ( 2004 ) 645 .
z. nan - yue and k.s .
williams , pacific j. math .
* 168 * ( 1995 ) 271 .
borwein , d.j . broadhurst and j. kamnitzer , experimental math .
* 10 * ( 2001 ) 25 .
davydychev and j.b .
tausk , phys .
* d53 * ( 1996 ) 7381 ; + a.i .
davydychev , proc .
workshop `` aihenp-99 '' , heraklion , greece , april 1999 ( parisianou s.a . , athens , 2000 ) , p. 219
( hep - th/9908032 ) ; + a.v .
kotikov and l.n .
lipatov , nucl .
* b582 * ( 2000 ) 19 . a.i .
davydychev , phys .
* d61 * ( 2000 ) 087701 .
prudnikov , yu.a .
brychkov and o.i .
marichev , _ integrals and series , v.3 : more special functions _ , gordon and breach , new york , 1990 .
tarasov , phys .
* d54 * ( 1996 ) 6479 ; nucl .
b ( proc . suppl . )
* 89 * ( 2000 ) 237 ; + s. moch and j.a.m .
vermaseren , nucl .
phys . * b573 * ( 2000 ) 853 ; + s. moch , j.a.m .
vermaseren and a. vogt , nucl .
* b646 * ( 2002 ) 181 ; + s. laporta , int .
j. mod .
* a15 * ( 2000 ) 5087 ; phys .
b504 * ( 2001 ) 188 ; * b549 * ( 2002 ) 115 .
broadhurst , j. fleischer and o.v .
tarasov , z. phys .
* c60 * ( 1993 ) 287 .
broadhurst , z. phys .
* c47 * ( 1990 ) 115 .
davydychev , _ loop calculations in qcd with massive quarks " _ , talk at int .
. relativistic nuclear dynamics " ( vladivostok , russia , september 1991 ) , + http://wwwthep.physik.uni-mainz.de/@xmath516davyd/preprints/vladiv.ps.gz a. bashir , r. delbourgo and m. l. roberts , j. math .
* 42 * ( 2001 ) 5553 .
tarasov , nucl .
* b502 * ( 1997 ) 455 .
davydychev and v.a .
smirnov , nucl .
* b554 * ( 1999 ) 391 .
tarasov , phys .
* d54 * ( 1996 ) 6479 .
j. fleischer , v.a .
smirnov , a. frink , j.g .
krner , d. kreimer , k. schilcher and j.b .
tausk , eur .
j. * c2 * ( 1998 ) 747 .
a. djouadi , m. spira and p.m. zerwas , phys .
* b264 * ( 1991 ) 440 ; + s. dawson and r. kauffman , phys .
* d49 * ( 1994 ) 2298 ; + m. spira , a. djouadi , d. graudenz and p.m. zerwas , nucl .
b453 * ( 1995 ) 17 .
g. passarino and s. uccirati , nucl .
* b629 * ( 2002 ) 97 ; + a. ferroglia , g. passarino , s. uccirati and m. passera , nucl .
instrum .
* a502 * ( 2003 ) 391 ; + a. ferroglia , m. passera , g. passarino and s. uccirati , nucl .
* b650 * ( 2003 ) 162 ; + a. ferroglia , g. passarino , m. passera and s. uccirati , nucl .
* b680 * ( 2004 ) 199 .
r. bonciani , p. mastrolia and e. remiddi , hep - ph/0311145 .
r. bonciani , p. mastrolia and e. remiddi , nucl .
* b661 * ( 2003 ) 289 .
smirnov , phys .
* b524 * ( 2002 ) 129 .
davydychev , p. osland and l. saks , phys .
d * 63 * ( 2001 ) 014022 .
davydychev and v.a .
smirnov , nucl .
* a502 * ( 2003 ) 621 .
avdeev , comput .
* 98 * ( 1996 ) 15 .
m. steinhauser , comput .
* 134 * ( 2001 ) 335 .
broadhurst , z. phys .
* c54 * ( 1992 ) 599 .
kotikov , j. high energy phys . * 09 * ( 1998 ) 001 ; + j.m . chung and b.k .
chung , phys .
* d59 * ( 1999 ) 105014 ; j. korean phys .
* 40 * ( 2002 ) 435 ( hep - ph/0203143 ) .
s. groote , j.g .
krner and a.a .
pivovarov , phys .
* d60 * ( 1999 ) 061701 .
ferguson , d.h .
bailey and s. arno , math .
* 68 * ( 1999 ) 351 .
f. jegerlehner and o.v .
tarasov , nucl .
* b549 * ( 1999 ) 481 .
chang , k.j .
gaemers and w.l .
van neerven , nucl .
* b202 * ( 1982 ) 407 ; + a. djouadi and c. verzegnassi , phys . lett .
* b195 * ( 1987 ) 265 ; + a. djouadi , nuovo cim .
* a100 * ( 1988 ) 357 ; + b.a .
kniehl , j.h .
khn and r.g .
stuart , phys .
* b214 * ( 1988 ) 621 ; + b.a .
kniehl , nucl .
* b347 * ( 1990 ) 86 ; + a. djouadi and p. gambino , phys .
* d49 * ( 1994 ) 3499 ; * d53 * ( 1996 ) 4111(e ) .
m. veltman , _ diagrammatica _ , cambridge university press , 1994 ; + d. bardin and g. passarino , _ the standard model in the making _ , clarendon , oxford , 1999 . j. fleischer and f. jegerlehner , phys .
* d23 * ( 1981 ) 2001 . f. jegerlehner , m.yu .
kalmykov and o. veretin , nucl .
* b641 * ( 2002 ) 285 .
chetyrkin and m. steinhauser , nucl .
* b573 * ( 2000 ) 617 . f. jegerlehner and m.yu .
kalmykov , nucl .
* b676 * ( 2004 ) 365 .
j. blmlein , hep - ph/0311046 . | when calculating higher terms of the @xmath0-expansion of massive feynman diagrams , one needs to evaluate particular cases of multiple inverse binomial sums .
these sums are related to the derivatives of certain hypergeometric functions with respect to their parameters .
we explore this connection and analytically calculate a number of such infinite sums , for an arbitrary value of the argument which corresponds to an arbitrary value of the off - shell external momentum .
in such a way , we find a number of new results for physically important two - loop two- and three - point feynman diagrams .
the results are presented in terms of generalized polylogarithmic functions .
desy/02 - 214 + hep - th/0303162 + * massive feynman diagrams and inverse binomial sums * a. i. davydychev@xmath1 and m. yu .
kalmykov@xmath2 + @xmath3_institute for nuclear physics , moscow state university , + 119992 , moscow , russia
_ + @xmath4 _ desy zeuthen , theory group , platanenallee 6 , + d-15738 zeuthen , germany _ + @xmath5_bltp , jinr , 141980 dubna , russia . _ + |
consider two probability measures on @xmath5 , @xmath6 and @xmath7 , and the set @xmath8 of maps transporting @xmath6 to @xmath7 , that is , the set of all measurable maps @xmath9 such that , if the initial space is endowed with the probability @xmath6 , then the distribution of the random variable @xmath10 is @xmath7 .
monge s optimal transportation problem consists of relocating a certain amount of mass from its original distribution to a different target distribution minimizing the transportation cost . in more abstract terms , the problem consists of finding a transportation map @xmath11 such that @xmath12 here , and throughout the paper , we assume @xmath13 . remarkably , under some smoothness assumptions , monge s problem is intimately related to the @xmath14-wasserstein distance by @xmath15 where the @xmath14-wasserstein distance between @xmath6 , @xmath7 is defined as @xmath16 and @xmath17 is the set of probability measures on @xmath18 with marginals @xmath6 and @xmath7
. this functional is related to very important problems in mathematics , the study of which has led to deep developments in several fields of research and applications , linked to such important names as ampre , kantorovich , rubinstein , zolotarev and dobrushin , among others . to avoid a huge number of references , we refer to the books by rachev and rschendorf @xcite and by villani @xcite , for an updated account of the interest and implications of the problem . however , we emphasize the importance of the topic in the development of the theory of probability metrics and its implications in statistics , particularly in goodness of fit problems . focusing on such kind of problems , the functional of interest is @xmath19 , where @xmath20 is the empirical measure associated to a sample @xmath21 , of independent identically distributed ( i.i.d . ) random vectors with law @xmath6 , and @xmath7 is any target probability measure on @xmath5 , or @xmath22 , where @xmath23 stands for the empirical measure on a second , independent i.i.d .
random sample , @xmath24 .
these empirical versions are connected to a combinatorial optimization problem , namely the _ optimal matching _ problem .
in fact , @xmath25 , where @xmath26 and @xmath27 ranges over the permutations of the set @xmath28 .
a lot of work has been devoted to analyzing the rate and mode of convergence of ( [ transportfunctional ] ) and several variants of it , beginning with the seminal paper by ajtai , komlos and tusnady @xcite in the case in which both samples come from the same underlying probability law @xmath6 .
the problem can be equivalently formulated in terms of @xmath29 , the distance between the empirical and true distributions .
further references will be provided later , but now let us mention the series of papers authored by talagrand @xcite , talagrand and yukich @xcite and dobri and yukich @xcite , which in the case when @xmath6 is the uniform distribution on the @xmath0-dimensional unit cube , @xmath1^d$ ] , essentially shows that @xmath30 this paper deals with the empirical cost of optimal _ incomplete _ or _ partial _ transportation .
this is the case in which the amount of mass required in the target distribution is smaller than that in the original one .
then , we do not have to move all the original mass , but we can dismiss a fraction of it . of course , we would like to complete this task with a minimal cost .
a more general version is possible if we admit that we only have to fulfill a fraction of the target distribution .
the general formulation of this problem , with quadratic cost , has been introduced by caffarelli and mccann @xcite , relating it to a monge ampre double obstacle problem .
they obtain remarkable results on the existence , uniqueness and regularity of the optimal solutions in a well - separated situation .
figalli @xcite improved the results covering the case of nondisjoint supports for the involved probability measures . independently , lvarez - esteban et al .
@xcite introduced the problem in the context of similarity of probabilities , obtaining a more general result of existence and uniqueness of the optimal solution .
moreover @xcite includes almost sure consistency of sample solutions to the true ones . in a subsequent paper , lvarez - esteban et al .
@xcite noticed the faster rate of decay of the cost of empirical incomplete transportation ( in the @xmath31 case ) and introduced a procedure for testing similarity of probabilities based on this fact .
a convenient mathematical formulation of this optimal incomplete transportation problem can be done with the help of the concept of _ trimmings _ of a probability .
[ definition1 ] given @xmath32 and borel probability measures @xmath6 , @xmath33 on @xmath5 , we say that @xmath33 is an @xmath34-trimming of @xmath6 if @xmath33 is absolutely continuous with respect to @xmath6 , and the radon nikodym derivative satisfies @xmath35 .
the set of all @xmath34-trimmings of @xmath6 will be denoted by @xmath36 .
note that in the extreme case @xmath37 , @xmath38 is just @xmath6 , while @xmath39 is the set of all probability measures absolutely continuous with respect to @xmath6 .
see @xcite for useful alternative characterizations of trimmings of a probability , as well as mathematical properties of the set @xmath36 .
turning back to the partial mass transportation problem , we could represent the target distribution by the probability @xmath7 and the initial distribution of mass by @xmath40 , @xmath6 being another probability if the mass required in the target distribution is @xmath41 times the mass in the original locations .
an incomplete transportation plan is then a probability measure @xmath42 on @xmath43 with second marginal equal to @xmath7 and first marginal in @xmath44 , and the cost of optimal incomplete transportation is @xmath45 in the more general case , with slackness in the target distribution , the optimal incomplete transportation cost would be @xmath46 , the minimal @xmath47 distance between trimmings of @xmath6 and @xmath7 .
this paper gives exact rates of convergence for empirical versions of the optimal incomplete transportation cost . as with classical optimal transportation , the results can be considered in terms of a combinatorial optimization problem , that we call _ optimal incomplete matching_. to be precise , assume that we can trim ( eliminate ) a fixed proportion @xmath34 of @xmath48 s points and also of @xmath49 s points , and we should only search for the best matching between the nontrimmed samples . taking for simplicity @xmath50 to be an integer ,
the new functional of interest is @xmath51 where @xmath27 varies in the set of permutations of @xmath52 , @xmath53 ranges in the subsets of size @xmath54 of @xmath55 , and similarly @xmath56 ranges in the subsets of size @xmath54 of @xmath57 .
it is easy to check that @xmath58 in fact , @xmath59 equals the minimum [ in @xmath60 of the linear function @xmath61 subject to the linear constraints @xmath62 , @xmath63 , @xmath64 , @xmath65 [ we are assuming @xmath66 . rescaling we see that @xmath67 equals the minimum [ in @xmath68 of the linear function @xmath69 subject to the linear constraints @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 .
the constraint matrix in this last linear program is totally unimodular ( see , e.g. , theorem 13.3 in @xcite ) and the right - hand side is integer .
hence , the minimum is attained at some integer solution , that is , satisfying @xmath76 , and this implies ( [ partialmatching ] ) .
we will show the somewhat unexpected result that , independently of the value of @xmath77 , the rates in ( [ untrimmed ] ) change to @xmath78 for _ any _ dimension @xmath79 .
in fact , ( [ trimmed ] ) follows from the triangle inequality and @xmath80 which is the formulation we choose for the results we prove .
our approach relies only on elementary or rather classical tools .
in particular , we do not use subadditivity arguments as in @xcite .
subadditivity yields a.s .
convergence to a constant , rather than just a rate of convergence . on the other hand ,
the approach in @xcite relies on showing subadditivity of a certain poissonization of the matching functional ( subadditivity does not hold for the original matching functional ; see remark 1.1 in @xcite ) .
we could also use that approach here , at least for @xmath81 ( otherwise duality for optimal matching , which is essential in the cited approach , becomes harder to deal with ) but the approximation rate for the poissonization of the incomplete matching functional would not allow us to recover the present result in dimension @xmath82 . the study of the rate of convergence of @xmath83 for @xmath77 , for some probability metrics @xmath84 was started in del barrio and matrn @xcite . in the case of the wasserstein metric and dimension @xmath85 , the results in @xcite already show a very different behavior with respect to the untrimmed case , namely @xmath86 we close this mentioning the connection of empirical optimal incomplete tranportation to another important problem in probability , that of _
random quantization_. taking @xmath87 ( full trimming ) we have that @xmath88 is the set of all probability measures concentrated on the sample points , and @xmath89 is the minimal @xmath14-cost of relocating a mass distributed according to some probability measure @xmath6 to a collection of randomly chosen spots @xmath21 . when @xmath21 are @xmath5-valued i.i.d . random vectors and
@xmath6 is absolutely continuous , the problem can be formulated , in monge s way , as the minimization of @xmath90 where @xmath91 varies in the set of all measurable functions with values in @xmath92 .
since , for a fixed @xmath93 in the integrand in ( [ equacion1 ] ) , the distance @xmath94 is minimized for @xmath95 , without any constraint on the capacity to be stored at @xmath96 , we obtain that the optimal @xmath91 is given by this last expression , and hence the optimal cost equals @xmath97 random quantization is a well - studied problem ; see , for example , the graf and luschgy monograph @xcite or the more recent paper by yukich @xcite . in particular , the asymptotic behavior of the @xmath14 quantization error is known , hence the rate at which @xmath98 vanishes .
a trivial consequence of definition [ definition1 ] is that @xmath99 if @xmath100 .
this implies @xmath101 hence rates of convergence for the random quantization error are a lower bound for rates of convergence of @xmath102 for general @xmath34 . in a first look , classical optimal transportation is a _ global _
problem while random quantization is a _ local _ one : a point @xmath93 is mapped through the optimal map @xmath91 to a sample point which in the case of random quantization , is determined just by sample points which are close to @xmath93 ( the nearest neighbor in fact ) while in the case of optimal transportation , two samples with the same sample points in a neighborhood of @xmath93 may result , however , in very different destinations for @xmath93 due to capacity constraints .
it turns out though , that this different character is only apparent , in terms of rates , in dimensions @xmath85 or @xmath103 .
the most relevant fact which we show in this paper is that , again in terms of rates , optimal incomplete transportation shows the same _ local _ nature as random quantization in _ any _ dimension .
the remainder of this paper is organized as follows . in section
[ preliminary ] we give a quick survey on known results about rates of convergence for optimal transportation and random quantization .
section [ dimension1 ] contains new results for optimal incomplete transportation .
we consider first the case @xmath85 , and in this case we construct upper and lower envelopes for the optimal solution to the incomplete transportation problem .
these are not optimal , but attain the correct rate .
finally , we construct a nearly optimal solution in general dimension starting from the one - dimensional construction .
we will use @xmath104 to denote the expected value of a random variable @xmath48 .
by @xmath105 [ resp . , @xmath106 , or even @xmath107 we refer to the conditional probability ( resp .
, conditional expectation ) given the set @xmath108 .
the indicator function of @xmath108 will be denoted by @xmath109 , while the notation @xmath110 will be reserved for dirac s probability measure on the point @xmath93 .
unless otherwise stated , the random vectors will be assumed to be defined on the same probability space @xmath111 .
we write @xmath112 for lebesgue measure on the space @xmath113 . finally , convergence in probability ( resp .
, weak convergence of probabilities ) will be denoted by @xmath114 ( resp .
, by @xmath115 ) , and @xmath116 will denote the law of the random vector @xmath48 .
the results on the asymptotic behavior of @xmath14-wasserstein distances between the empirical and parent distributions in the one - dimensional case have been obtained through a quantile representation . if @xmath117 and @xmath118 are the distribution functions of @xmath6 and @xmath7 and @xmath119 and @xmath120 are the respective quantile functions , then ( see , e.g. , bickel and freedman @xcite ) @xmath121^{1/p}\ ] ] [ where @xmath122 . in particular ,
when @xmath6 is the uniform distribution on @xmath123 , this representation leads to @xmath124^{1/p}\ ] ] with @xmath125 a brownian bridge on @xmath1 $ ] ; see , for example , @xcite . for dimension @xmath126 , there are not explicit expressions for the optimal transportation maps , and limit distribution results as in ( [ bgu ] ) are not available
. rates of convergence to 0 of @xmath29 can be given based on different approaches .
the case @xmath127 is the most interesting from the point of view of the mass transportation problem .
ajtai , komls and tusndi @xcite showed that , with probability @xmath128 , @xmath129 where @xmath20 and @xmath23 are the sample distributions corresponding to two independent samples obtained from the uniform distribution on the unit square , @xmath130 ^ 2)$ ] .
their combinatorial partition scheme method was refined in talagrand and yukich @xcite to show [ theorem 1 and remark ( ii ) there ] that for some constant , @xmath131 ,
@xmath132 ^ 2\bigr)\bigr)\bigr)\leq c(p ) \biggl ( { \frac{\log n}n } \biggr)^{1/2}.\ ] ] the case @xmath4 ( and uniform distribution on the @xmath0-dimensional unit cube ) is covered in talagrand @xcite .
that paper uses a different approach , based on duality for the optimal transportation problem to give a result ( theorem 1.1 ) , formulated for a very general class of costs functions which includes exponential costs and , as a consequence , implies @xmath133^d\bigr)\bigr)\bigr)\leq c(k , p ) \frac1 { n^{1/d}}.\ ] ] further results , dealing with distributions other than the uniform , possibly with unbounded support , are given in barthe and bordenave @xcite . as we already noted in the
, the so - called random quantizers provide an easy way of giving a lower bound for the rates of convergence of our interest .
we give a simple version of the mean asymptotics for the random quantizers , rewritten in terms of the @xmath14-wasserstein distance between the set of `` fully trimmed '' sample probabilities and the theoretical distribution , that suffices for our purposes ; this is a particular case of theorem 9.1 in @xcite .
[ tasaconrecorte1 ] if @xmath21 are i.i.d .
random vectors uniformly distributed on @xmath1^d$ ] , then @xmath134^d\bigr)\bigr)\bigr)\to\gamma { \biggl(1 + \frac p k \biggr){\omega_d^{-p / d } } } \qquad\mbox{as } n\to\infty,\ ] ] where @xmath135 . for optimal incomplete transportation ,
a first result on rates of convergence is theorem 5 in the appendix of @xcite , for dimension 1 , but it has been largely improved in @xcite , in the terms expressed in ( [ rate ] ) . for dimension 2
our approach was not successful in going beyond the characteristic `` @xmath136 '' term in the ajtai kmlos
tusndy result ( [ untrimmed ] ) .
this task and the fundamental improvement in dimension 1 are the main goals in this paper .
moreover we notice del barrio and matrn @xcite also treat the improvement of the `` in probability bounds '' involved in ( [ theproblem ] ) to almost surely bounds .
this follows talagrand s approach @xcite , continued by dobri and yukich @xcite , but , in our case , using a powerful concentration inequality of boucheron et al .
we focus first on the one - dimensional case .
let us consider @xmath3 distinct points @xmath137 and set @xmath138 , where @xmath110 denotes dirac s measure on @xmath93 .
an @xmath34-trimming of @xmath20 can be written , in terms of a vector @xmath139 , as @xmath140 with @xmath141 ( we set , for convenience , @xmath142 ) .
we therefore write @xmath143\\[-8pt ] & & \hspace*{212pt}\qquad i=1,\ldots , n \biggr\}.\nonumber\end{aligned}\ ] ] our first result is an elementary , but useful , representation of @xmath144 .
[ expresionutil ] if @xmath137 , @xmath145 , @xmath6 is the uniform distribution on @xmath1 $ ] , @xmath146 is defined by ( [ setca ] ) and @xmath13 , then @xmath147 where @xmath148 and @xmath149 denotes the odd extension to @xmath150 of the function @xmath151 on @xmath152 . we note first that the quantile function associated to @xmath153 takes the value @xmath154 in the interval @xmath155 $ ] . hence , using ( [ prop1 ] ) , we see that @xmath156 , where @xmath157 .
since @xmath158 is a primitive of @xmath159 , we can write @xmath160.\end{aligned}\ ] ] from this , recalling that @xmath161 , @xmath162 , we get @xmath163\ ] ] with @xmath164 .
now it is easy to see that @xmath165 , which completes the proof .
the function @xmath166 in lemma [ expresionutil ] is a piecewise polynomial for integer @xmath167 .
for instance @xmath168 , @xmath169 , @xmath170 , @xmath171 ; @xmath172 , @xmath173 . for general @xmath13
, @xmath166 is a nonnegative , even and convex function , strictly increasing on @xmath152 , which attains its minimum at @xmath174 , with @xmath175 .
this suggests that a good trimming vector @xmath176 should be as close as possible to the midranks , @xmath177 . with this observation in mind
, we denote @xmath178 and define @xmath179 @xmath180 , @xmath181 , and @xmath182 .
finally , we set @xmath183 note that , for any @xmath176 , @xmath184 is a sequence that decreases from 0 to @xmath185 , while @xmath186 is the lowest decreasing sequence from 0 to @xmath187 which lies above the sequence @xmath188 . in the next result
we see that @xmath189 is a feasible trimming and that feasible solutions that exceed this one can not be optimal .
@xmath190 and @xmath191 are obvious . to prove @xmath193 we check , equivalently , that @xmath195 , @xmath196 .
clearly , @xmath197 and @xmath198 and , consequently , @xmath199 . to see that @xmath200 observe that @xmath201 unless @xmath202 , but then @xmath203 and the claim follows . we show now that @xmath204 , where @xmath205 .
since @xmath206 , we see that @xmath207 , provided @xmath208 .
now , if @xmath209 , then @xmath210 .
let us assume that @xmath211 ( hence @xmath212 ) and @xmath213 , @xmath214 , but @xmath215 .
let us write @xmath216 for the smallest integer @xmath217 such that @xmath218 ( observe that @xmath219 since @xmath220 ) .
we define @xmath221 if @xmath214 or @xmath222 and @xmath223 if @xmath224 . also , write @xmath225 . clearly , @xmath226 .
but for integer @xmath227 we have @xmath228 , which implies @xmath229 . consequently , @xmath230 against optimality of @xmath231 .
hence , @xmath232 for all @xmath233 , and the upper bound for @xmath234 follows . a similar lower bound for @xmath231
can be obtained taking @xmath235 to be the greatest decreasing sequence from 0 to @xmath187 which lies below the sequence @xmath236 and setting @xmath237 .
we note also that lemma [ expresionutil ] , combined with lemma [ splitting ] , gives the following useful lower and upper bounds for the incomplete transportation cost . to be precise , @xmath238\\[-8pt ] & \leq&v_n(p)+\frac1 { p+1 } \sum _ { i=1}^{n-1 } \biggl(\frac { x_{i+1}-x_i}{2 } \biggr)^{p+1 } f_p \biggl(\frac{\bar{h}_i-({x_{i+1}+x_i})/{2 } } { ( { x_{i+1}-x_{i}})/{2 } } \biggr),\nonumber\end{aligned}\ ] ] where @xmath239 .
we could replace @xmath189 with @xmath240 or @xmath241 , but in terms of rates , the upper bound above can not be improved , as we will see later .
next , we consider the case of a uniform random sample on the unit interval , namely @xmath21 are i.i.d .
@xmath242 r.v.s , @xmath243 is the order statistic and @xmath20 the empirical distribution on the sample .
we will use the well - known fact @xmath244 where , @xmath245 and @xmath246 are i.i.d .
exponentials random variables with unit mean .
the following elementary lemma about the concentration of the @xmath247 s around their means will be used repeatedly in the remainder of this section .
[ cotaenesperanza ] if @xmath6 is the uniform distribution on @xmath1 $ ] , @xmath21 are i.i.d .
random variables with common distribution @xmath6 , @xmath20 is the empirical measure on @xmath21 , @xmath77 and @xmath13 , then there exist constants , @xmath253 , depending only on @xmath167 and @xmath34 , @xmath254 depending only on @xmath167 , such that for every @xmath255 , @xmath256 for the lower bound simply observe that @xmath257 , with @xmath258 as in ( [ upperlower ] ) . the spacing @xmath259 follows a beta distribution with parameters @xmath260 and @xmath261 , and from this fact it follows that @xmath262 .
it is easy to check ( using stirling s formula , e.g. ) that @xmath263 as @xmath264 , and hence we can take @xmath265 .
for the upper bound we use the representation ( [ equivalence ] ) , fix @xmath266 and write @xmath267 .
then we split @xmath268 , @xmath269 and proceed to bound @xmath270 , @xmath271 . to deal with @xmath272 we note that @xmath273 , @xmath274 being the distribution function asociated to @xmath20 . a simple computation , similar to the proof of lemma [ expresionutil ] , shows @xmath275 , both terms equaling , in fact , @xmath276.\ ] ] hence , from schwarz s inequality we get @xmath277 using the fact that @xmath278 ( this follows from hoeffding s inequality applied to bernoulli random variables ) , we see that @xmath279 . also , from lemma [ concentracionsi ] we get @xmath280
combining these two estimates we get @xmath281 the last upper bound is a vanishing sequence ( hence , bounded ) since @xmath282 . note that the bound depends on @xmath34 through the choice of @xmath266 .
we consider now @xmath283 and recall ( [ upperlower ] ) .
we have @xmath284 since , as noted above,@xmath285 is a convergent sequence as @xmath264 .
observe now that @xmath286 if @xmath169 , while @xmath287 if @xmath288 . therefore @xmath289 , and it suffices to give an upper bound for @xmath290 where we are using representation ( [ equivalence ] ) and @xmath291 it only remains to find an upper bound for @xmath292 with @xmath293 we split the sum in @xmath294 into three terms , @xmath295 , @xmath296 collecting the summands with @xmath297 , @xmath298 those with @xmath299 and @xmath300 the others .
we bound first @xmath301 .
we write @xmath302 and note that @xmath303 . now
, @xmath304 convexity implies that @xmath305 for @xmath306 . from the schwarz inequality and the moment inequality @xmath307 , @xmath13
, we get that @xmath308 for some absolute constant @xmath309 ( not depending on @xmath233 or @xmath3 ) . on the other hand , using again lemma [ concentracionsi ] with @xmath310 and @xmath311 [ which is positive for @xmath312 we have @xmath313 , where . since @xmath314
, we get , for some constant , @xmath315 , @xmath316 [ @xmath317 suffices for @xmath318 ; with a larger constant , if necessary , the bound is true for all @xmath233 ] . combining the last bounds and schwarz s inequality we obtain , with a new constant @xmath320 , @xmath321 and , again , the right - hand side is a vanishing sequence . to deal with @xmath322
we define @xmath323 , @xmath324 , @xmath325 .
observe that @xmath326 , @xmath327 , and @xmath328 .
we also write @xmath329 .
then @xmath330 with the above notation we see that @xmath331 , while @xmath332 .
these observations imply that @xmath333 the last upper bound and @xmath296 are equally distributed .
hence @xmath334 .
we turn now to the central part , @xmath300 .
obviously @xmath335 where @xmath336 .
once more we use schwarz s inequality to get @xmath337 .
thus , it only remains to show @xmath338 .
chernoff s inequality yields @xmath339 . from this and the fact @xmath340 , @xmath341 , we get , for integer @xmath342 , @xmath343\\[-8pt ] & \leq & ( i+k)^k \frac{e^i i!}{i^i } \sum _ { j=0}^i \frac{e^{-k i } ( ki)^j}{j ! }
\nonumber\\ & = & ( i+k)^k \frac{e^i i!}{i^i}p(n_{ki}\leq i),\nonumber\end{aligned}\ ] ] where @xmath344 denotes a random variable having poisson distribution with mean @xmath345 .
chernoff s inequality ( for the left tail ) gives @xmath346 where @xmath347 , @xmath348 ; see , for example , @xcite , page 19 .
this , ( [ cota23 ] ) and the fact @xmath349 for some constant @xmath350 , imply @xmath351 for some constant @xmath352 ( which depends on @xmath216 ) , where we have used that@xmath353 .
this completes the proof .
finally , we turn to general dimension . in our last result
we combine the upper bound in theorem [ cotaenesperanza ] with a combinatorial approach to give the exact rate of convergence of the empirical cost of optimal incomplete transportation to the uniform distribution on the @xmath0-dimensional unit cube .
the result is not given in terms of expectations as in theorem [ cotaenesperanza ] .
our approach allows also to get that type of result , but we refrain from adding more technicalities . for the sake of simplicity
we consider the case @xmath127 .
the idea carries over smoothly to higher dimension . on the other hand ,
the case @xmath4 follows from known results for the usual transport ; recall ( [ talagrandbound ] ) .
we write @xmath355 $ ] and @xmath356^t$ ] , @xmath357 .
we denote also @xmath358 \}$ ] , @xmath359 .
the random vector @xmath360 follows a multinomial distribution with parameters @xmath3 and @xmath361 .
given @xmath362 , we denote by @xmath363 the indices @xmath216 such that @xmath364 $ ]
. then @xmath365 are an i.i.d .
@xmath242 sample .
we write @xmath366 for the @xmath367-trimming of the empirical distribution on @xmath368 considered in the proof of theorem [ cotaenesperanza ] .
then we have @xmath369 .
we write also @xmath370 for the optimal transportation map from @xmath242 to @xmath366 .
then @xmath371 .
we recall that @xmath91 takes values on the set @xmath372 and , with @xmath373 denoting @xmath0-dimensional lebesgue measure , @xmath374 .
next we define the map @xmath91 on @xmath375\times(0,1]$ ] as follows .
if @xmath376^t$ ] is such that @xmath377 $ ] and @xmath378 , then @xmath379 .
in other words , points on the stripe @xmath380\times(0,1]$ ] are mapped to one of the observations on that stripe , the precise one being determined by the @xmath381 trimming function on the second coordinate .
clearly , for @xmath382 $ ] , @xmath383 from this we get @xmath384\times(0,1 ] } \bigl\| x-\varphi(x)\bigr\|^p \,dx & = & \sum _
{ i=1}^n \int_{(({i-1})/n , i / n]\times(0,1 ] } \bigl\| x-\varphi(x)\bigr\|^p \,dx \\ & \leq & \frac{2^{p-1 } } { n^p}+\frac{2^{p-1 } } n \sum_{i=1}^n \int_0 ^ 1 \bigl|x_2- \varphi_i(x_2)\bigr|^p \,dx_2.\end{aligned}\ ] ] furthermore , @xmath385 if @xmath386\times ( 0,1]$ ] .
thus , @xmath91 maps @xmath6 into an @xmath34-trimming of @xmath20 if @xmath387 or , equivalently , if @xmath388 , with @xmath389 . as a consequence ,
on the set @xmath390 , @xmath391 now , we note that @xmath392 but the last two displays imply that @xmath393 this , toghether with the fact that @xmath394 as @xmath264 ( see theorem 7 , page 112 , in @xcite ) implies that @xmath395 and completes the proof . | we consider the problem of optimal incomplete transportation between the empirical measure on an i.i.d .
uniform sample on the @xmath0-dimensional unit cube @xmath1^d$ ] and the true measure .
this is a family of problems lying in between classical optimal transportation and nearest neighbor problems .
we show that the empirical cost of optimal incomplete transportation vanishes at rate @xmath2 , where @xmath3 denotes the sample size . in dimension @xmath4 the rate is the same as in classical optimal transportation , but in low dimension it is ( much ) higher than the classical rate . |
the @xmath3 atomic nucleus is one of the best studied few - body systems , both experimentally and theoretically , as summarized in the recent @xmath4 compilation @xcite . besides the many textbook examples of gross structure , there are subtle points yielding large effects that are only qualitatively understood . except for @xcite
none of the existing calculations aims at a complete understanding of the many features of @xmath3 , which is not surprising in view of the number of different phenomena studied so far @xcite . with the recent compilation @xcite and the comprehensive @xmath1-matrix analysis @xcite of a large amount of scattering data below @xmath5 mev , a new , microscopic calculation for the @xmath3 system in this energy range using modern realistic two- and three - nucleon forces is most desirable .
it is well known that realistic nucleon - nucleon ( @xmath6 ) forces can not reproduce the @xmath7 , @xmath8he , and @xmath9he binding energies .
three - nucleon interactions ( tnis ) are added to give the necessary small corrections but they still fail to reproduce certain properties of the three nucleon system , most notably the @xmath10 analyzing power in @xmath11 scattering @xcite . yet the 30% deviation of @xmath10 can be resolved by tiny changes in the @xmath11 scattering phase shifts ( on the order of @xmath12 degrees @xcite ) . furthermore very many operators can contribute to a tni and the lack of stringent conditions in the three - nucleon system on the structure of the tni makes its application to other systems desirable . in @xcite
it was shown that although a realistic @xmath6 force can generally reproduce the @xmath13he system , there remain differences , most notably in the analyzing powers . since the intensely studied @xmath14 system @xcite is unfortunately very difficult to describe due to the many resonances and the @xmath15 bound state , the much simpler systems @xmath16he and @xmath17h where data exist in the energy range of interest
were investigated in @xcite .
the essential findings of this work are that realistic @xmath6 interactions describe most of the phase shifts quite well but fail to reproduce the @xmath18 and @xmath19 phase shifts .
the calculated splitting between these two channels is much too small , and neither the urbana ix ( uix ) @xcite nor the tla @xcite three - nucleon force is able to improve the splitting significantly .
in fact , there it is more important to include in the calculation negative parity states of the three - nucleon subsystem than one of these two tnis .
these findings suggest that new contributions to the @xmath20 force acting on the @xmath21-waves should be considered , like an @xmath22 type tni , as proposed in @xcite for the @xmath23 analyzing powers , or the @xmath24 operators proposed in @xcite .
based on these findings we choose as @xmath6 force only the av18 @xcite and as tni the urbana ix @xcite and in addition the @xmath24 @xcite .
we organize the paper in the following way .
the next section contains a brief discussion of the resonating group model calculation together with the model spaces used .
then we compare r - matrix and rgm results of a few typical examples of scattering phase shifts for various model spaces and combinations of interactions .
finally we compare with data for examples sensitive to the tnis .
we use the resonating group model @xcite to compute the scattering in the @xmath9he system using the kohn - hulthn variational principle @xcite .
the main technical problem is the evaluation of the many - body matrix elements in coordinate space .
the restriction to a gaussian basis for the radial dependencies of the wave function allows for a fast and efficient calculation of the individual matrix elements @xcite .
however , to use these techniques the potentials must also be given in terms of gaussians . in this work
we use suitably parametrized versions of the argonne av18 @xcite @xmath6 potential and the urbana ix @xcite and the @xmath24 @xcite .
the inclusion of an additional tni requires almost two orders of magnitude more computing power than the realistic @xmath6 forces alone . in the @xmath13he system
we use a model space with six two - fragment channels , namely the @xmath25h , the @xmath26he , the @xmath27h@xmath28h , the singlet deuteron and deuteron d - @xmath27h , the d - d , and the @xmath29 channels .
the last three are an approximation to the three- and four - body breakup channels that can not in practice be treated within the rgm .
the @xmath13he is treated as four clusters in the framework of the rgm to allow for the required internal orbital angular momenta of @xmath8he , @xmath30 or @xmath31h . for the scattering calculation
we include the @xmath32 , @xmath21 and @xmath33 wave contributions to the @xmath34 and @xmath35 channels . from the @xmath1-matrix analysis these channels
are known to give essentially the experimental data .
( we discuss cases where this is not the case . )
the full wave functions for these channels contain over 200 different spin and orbital angular momentum configurations , hence it is too complicated to be given in detail .
the simplest wave functions we use for @xmath36he are those described in @xcite . this small 29-dimensional model space yields -6.37 mev binding energy , an _
rms _ radius of 1.78 fm and a @xmath33 state probability of 7.7% for the @xmath8he using av18 . in order to avoid fake effects the relative thresholds in @xmath13he should be reproduced well , therefore we used also a 35-dimensional modelspace , called large , by allowing additional configurations with two orbital angular momenta on the two jacoby coordinates yielding -6.69 mev binding energy .
this must be compared to @xmath37mev known from faddeev calculations @xcite . for the deuteron we use the structure given in @xcite , yielding a binding energy of -1.921 mev , which could be easily improved .
but then the relative threshold energies deteriorate , see table [ thres ] .
all the gaussian width parameters were obtained by a non - linear optimization using a genetic algorithm @xcite for the combination av18 and uix . once the fragment wave functions are fixed the scattering problem is solved with our rgm code relying on the kohn - hulthn variational principle @xcite : @xmath38 where @xmath39 denotes the reactance matrix .
the model space described above ( consisting of four to ten physical scattering channels for each @xmath40 ) is by no means sufficient to find reasonable results .
so - called distortion or pseudo - inelastic channels @xcite have to be added to improve the description of the wave function within the interaction region .
accordingly , the distortion channels have no asymptotic part . for practical purposes
it is obvious to reuse some of the already calculated matrix elements as additional distortion channels . in that way
we include all the positive parity states of the three - nucleon subsystems with @xmath41 in our calculation . however , it was recently pointed out by a. fonseca @xcite that states having a negative parity @xmath42 in the three - nucleon fragment increase the @xmath43h cross section noteably
. therefore we also added the appropriate distortion channels in a similar complexity as in the @xmath44 case to our calculation , thereby roughly doubling the size of the model space .
the 20 percent increase of the model space for the 3n bound states from 29 to 35 resulted in almost a factor of two in the computing time .
the parameters of the @xmath24 tni were adjusted that the binding energy of triton and @xmath36he did change by less than 10 kev .
therefore we do not give the corresponding energies in table [ thres ] .
0.2 cm c|c|c|c|c potential & & + & @xmath36h & @xmath36he & @xmath45 & @xmath46 + av18 & -7.068 & -6.370 & 0.698 & 3.227 + av18 , large & -7.413 & -6.588 & 0.725 & 3.572 + av18 + uix & -7.586 & -6.875 & 0.710 & 3.745 + av18 + uix , large & -8.241 & -7.493 & 0.748 & 4.400 + exp .
& -8.481 & -7.718 & 0.763 & 4.033 +
since we are mainly interested in the effects of 3n - forces , we mention the bound state results only briefly . in the large model space av18 plus uix yields
-27.81 mev close to the experimental -28.296 mev , to which the parameters of the uix are fitted to . although the parameters of the @xmath24 tni were chosen as to give only minute changes in the three - nucleon system , for @xmath13he it resulted in 650 kev additional binding .
the most detailed comparison between calculation and data is on the level of an energy dependent phase - shift analysis .
this is given by the r - matrix analysis as described in @xcite in detail .
the lowest channel , triton - proton contains the intriguing first exited state of @xmath13he , a @xmath47-resonance , sometimes considered a breathing mode , which is clearly seen in the @xmath47 phase shift , see fig .
[ nullp ] , but does not show - up in the angular distributions , see fig . [ 120_nnn ] . neglecting the coulomb force
this resonance is moved just below the p - triton threshold , i.e. it becomes a bound state , bound by less than 50 kev depending on the force and model space used .
therefore all approaches , which neglect the coulomb force like @xcite can not aim at this energy region . in fig .
[ nullp ] the r - matrix results are compared to the pure nn - calculations for various model spaces .
for the small model space the calculation is slightly above the r - matrix data . adding the negative parity distortion channels
, we find a small increase due to the enlarged attraction .
this effect is much smaller than the one found in triton - neutron scattering @xcite .
increasing the 3n - model space reduces the phase shifts considerably , due to the better @xmath13he binding and the additional thresholds shifted to higher energy , see table [ thres ] . in fig .
[ 0-nnn ] we find strong sensitivity to the additional tni .
this sensitivity in specific partial waves might help to unravel the operator structure of the tni , especially as all the thresholds are unchanged from uix to @xmath24 .
triton - proton phase shifts from the r - matrix analysis ( crosses ) and calculations employing av18 in the small model space ( av18 ) , adding negative parity distortion channels ( av18n ) , for the large model space ( av18-l ) and adding negative parity distortion channels ( av18n - l),height=302 ] , but r - matrix results ( crosses ) are compared to the full nn - calculation ( av18 ) , adding uix ( av18u ) and adding @xmath24 ( av18uv),height=302 ] for 58 and 120 degrees exist measured exitation functions for the p - triton differential cross section . for the forward angle
the r - matrix analysis is on top of the data @xcite , the pure nn - calculation a bit below , with uix almost on top of the data and with @xmath24 a bit above .
since there are only minor differencies , we do not show them .
for the backward angle , however , see fig .
[ 120_nnn ] , even the r - matrix analysis can not fully reproduce the data in the coulomb - nuclear interference region .
the data from @xcite seem to be consistently above those from @xcite and @xcite . in the coulomb - nuclear interference
the sensitivity to tni is very large , whereas around the ( p , n ) threshold data , r - matrix and microscopic calculations agree essentially .
since all the data are very old and not consistent a new measurement is urgently called for . unfortunately due to radiation hazards of the triton this is not very likely .
.,height=302 ] a recent measurement of the real parts of the neutron-@xmath36he spin - dependent scattering lengths @xcite @xmath48 = 7.370(58 ) fm and @xmath49 = 3.278(53 ) fm is therefore very important .
the corresponding r - matrix results are @xmath48 = 7.398 fm and @xmath49 = 3.257 fm . due to the strong coupling via the @xmath47 resonance @xmath48
has a large imaginary part .
hence , the numerical extraction of this scattering length is not easy . therefore we give only preliminary numbers in the following . for the large model space the microscopic calculations yield @xmath50 fm and @xmath51 fm , depending on the combination of forces .
these results are close enough to be used for a further determination of the operator structure of the three - nucleon force . for the reaction @xmath27h(d , p)@xmath36h at 2.0 mev center - of - mass energy .
the data are from @xcite , the lines are as in fig [ 0-nnn ] and additionally the r - matrix results without f - waves ( nof).,height=302 ] out of the many possible data we choose an example , which demonstrates the limitation of our total partial wave model space .
the well studied deuteron - deuteron reactions allow a detailed comparison since together with differential cross sections for many energies also analysing powers are available .
the tensor analysing power @xmath52 displays a pronounced angular distribution , see fig .
[ t20_nnn ] .
the full r - matrix analysis nicely reproduces the data , whereas all microscopic calculations fail at forward angles and show no sign of the double - hump structure .
omitting all f - wave s - matrix elements in the r - matrix analysis corresponding to the model space used for the rgm calculation yields an angular dependence quite similar to the microscopic results , see fig .
[ t20_nnn ] .
this demonstrates the need for @xmath53 partial waves in the rgm calculations .
we presented a complete microscopic calculation in the @xmath13he system employing modern realistic two- and three - nucleon forces .
we demonstrated that in specific examples the inclusion of nnn - forces yields large effects in phase shifts , differential cross sections and analysing powers .
hence , the @xmath13he - system seems well suited for a detailed study of different nnn - forces , especially since a comprehensive r - matrix analysis exists , which reproduces the vast amount of data for various reactions very well , thus allowing for a comparison on the level of individual partial waves .
therefore a determination of the operator structure of the nnn - force is within reach , provided the microscopic calculations are converged . to aim at this goal the internal triton and @xmath36he wave functions have to be improved , such that the binding energy is within say 50 kev of the experimental value .
this can only be achieved by a major increase of the model space . for the deuteron
the corresponding modification is trivial . to describe the deuteron - deuteron reactions
, we have to allow also for f - waves on all relative coordinates .
all these improvements are relative straightforward , but require about a factor 5 in the cpu - time compared to the needs of the work reported here .
acknowledments : the work of h.m.h is supported by the bmbf ( contract 06er926 ) and that of g.m.h . by the department of energy
. the grant of computer time at the hlrb and the rrze is gratefully acknowledged .
we want to thank g. wellein and g. hager at the rrze for their help . | we report on a consistent , microscopic calculation of the bound and scattering states in the @xmath0 system employing modern realistic two - nucleon and three - nucleon potentials in the framework of the resonating group model ( rgm ) .
we present for comparison with these microscopic rgm calculations the results from a charge - independent , coulomb - corrected @xmath1-matrix analysis of all types of data for reactions in the @xmath2 system .
comparisons are made for selected examples of phase shifts and measurements from reactions sensitive to three - nucleon force effects . |
in addition to its primary mission of detecting exoplanets @xcite , the _ kepler _ satellite s exquisite photometry has allowed groundbreaking studies in stellar astrophysics .
most notably , analyses of the seismic modes of stellar lightcurves have enabled otherwise inaccessible measurements of stellar ages , masses , evolutionary states , and internal structure ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
_ kepler _ data are also providing fundamental new insights into mass loss on the red giant branch ( e.g. , * ? ? ? * ) , the variation of stellar rotation and stellar dynamos with age ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the origin of the blazhko effect in rr lyrae variables ( e.g. , * ? ? ?
one of the most confounding _ kepler _ discoveries is the light curve of kic 8462852 . first presented by @xcite ,
kic 8462852 appears to be a typical f3v star @xcite . however , its light curve exhibits 10 significant dips over the timespan of the _ kepler _ mission .
these dips are irregular in shape , aperiodic , and vary in depth from fractions of a percent up to 20% of the total flux of the star .
while the dips as a whole do not obey an obvious periodicity , a subset of them are consistent with a period of @xmath6 d , although some of those are 180 out of phase with the others .
@xcite present several physical models to account for these dipping events , concluding that a large family of exocomets or planetesimal fragments orbiting the star could plausibly explain the data .
kic 8462852 quickly became the focus of considerable attention .
@xcite place tight constraints on the circumstellar dust around kic 8462852 from millimeter and sub - millimeter observations , ruling out the planetesimal fragment hypothesis ; @xcite reach the same conclusion from an analysis of warm _ spitzer _ data .
@xcite model the light curve with a swarm of comets .
they find that the kic 8462852 light curve in quarters 16 and 17 of the _ kepler _ mission can be explained by approximately 100 comets in a very tight cluster .
this model , however , is unable to match the earlier dips in the light curve .
@xcite proposed that the star may make an ideal target for seti programs .
as of yet , seti searches in the optical @xcite and radio @xcite have only resulted in null detections .
other studies have made use of archival observations of kic 8462852 .
@xcite analyzed photographic plates from the `` digital access to a sky century @ harvard '' ( dasch ) archive @xcite , obtained between 1890 and 1989 .
this work found kic 8462852 to be dimming at an average rate of @xmath7 mag century@xmath2 , or @xmath8
this result was quickly called into question .
both @xcite and @xcite suggest that the observed dimming is the result of systematics in the dasch data , particularly the menzel gap of the 1950s , and report the flux of kic 8462852 to be consistent with no change over the baseline of the dasch plates .
however , the dasch team has not found evidence for such systematics @xcite . additionally , the claimed systematics appear to depend on the choice of reference stars ( j. grindlay 2016 , priv . comm . ) . if the star is dimming with time , then a survey with a long time baseline and high photometric precision would be able to detect this variation .
modern ground - based imaging surveys generally operate over timescales of @xmath9 yr , and they lack the photometric accuracy to detect the @xmath10 mmag variation that would be expected from the analysis of the dasch data by @xcite . the best - calibrated existing surveys , sdss and pan - starrs @xcite , would need a time baseline @xmath9 times as long in order to measure such a brightness change at @xmath11 significance ( ignoring that stars as bright as kic 8462852 with @xmath12 are badly saturated in such data sets ) .
data from the _ kepler _ mission , however , provide exactly such an opportunity . the _ kepler _ mission was designed to detect and characterize transit events with timescales of minutes to hours . to that end ,
_ kepler _ processing pipelines aim to remove longer - term trends in the data caused by instrumental effects .
this process is well - described by @xcite and @xcite and is the basis for the creation of the `` pre - search data conditioning '' ( pdc ) light curves .
the pdc light curves , among many other corrections , remove long - term trends in the observed flux of stars that correlate with either trends from nearby stars or with the centroid of a star s motion during an observing quarter .
the main source of target motion is differential velocity aberration , which leads to an approximately linear drift in the centroid of a star at the 0.1 pixel level over the course of a quarter .
any signals that are approximately linear over a quarter will then be removed as an instrumental artifact .
therefore , while the @xcite analysis of kic 8462852 was sensitive to short - timescale variability , they or any other group analyzing long cadence photometry of the target will not be able to detect any long - term trends .
the end result is that signals with characteristic timescales longer than a single observing quarter will not appear in the long cadence data . searches for rotation periods in _ kepler _ data corroborate this claim : @xcite identified more than 34,000 rotation periods for stars in the _ kepler _ field , but their detection efficiency drops for periods above 40 days and they do not detect any rotation periods longer than 70 days .
other sets of light curves designed to detect transiting planets will more aggressively remove stellar variability through a `` de - trending '' process , removing all non - transit signals ( e.g. * ? ? ?
the pixel - level data delivered by the telescope present a better opportunity to preserve long - term trends in the _ kepler _ light curves .
photometry with the pixel - level data has its own difficulties .
aperture photometry is complicated as the apertures around each star recorded by the telescope and downlinked to earth are often smaller than the full point spread function ( psf ) of the star @xcite .
small deviations in the pointing of the telescope can then cause large variations in the total flux recorded inside the photometric aperture as the edges of the star s psf move relative to the edge of the downloaded set of pixels .
instead , one could consider modeling the psf of the star on the detector . while possible in some cases ( e.g. , * ? ? ?
* ) , the lack of a reliable flat field for the _ kepler _ detector array and the lack of background `` sky '' pixels make psf modeling difficult , and nearly impossible for saturated stars ( @xmath13 , * ? ? ?
an alternative approach to search for long - term linear trends is to leverage the full - frame images ( ffis ) collected during the _ kepler _ mission . at the beginning of the mission ,
_ kepler _ recorded a series of eight ffis over two days .
it then continued to record one ffi each month throughout the mission @xcite .
as _ kepler
_ rotated by 90 every three months , there are generally three consecutive ffis with all stars located on the same pixels .
a star then cycles through four different detectors over the course of a year , returning to its original position and repeating the cycle a year later .
because there are @xmath14 observations with any given star landing on the same pixels , accurate relative photometry can be obtained without flat fielding .
the ffis have been used to study stellar variability in rr lyrae stars in the _ kepler _ field @xcite , highlighting their utility for photometry .
the primary _ kepler _ mission lasted four years .
a star dimming at a linear rate of 0.165 mag century@xmath2 ( @xmath8 yr@xmath2 ) , as has been purported for kic 8462852 @xcite , would be expected to decrease in brightness by 0.6% over the _
baseline , well above the photometric precision of the telescope . by analyzing the ffi data
, one could avoid the removal of this trend by the data processing pipeline and the loss of flux from limited apertures in the raw pixel - level data . in this study
, we attempt to characterize the long - term photometric behavior of kic 8462852through an analysis of _ kepler _ ffi data .
this paper is organized as follows . in section [ sec : data ] , we describe our data analysis and photometry . in section [ sec : results ] , we present our results . in section
[ sec : compare ] , we compare our results to a similar analysis of stars close to kic 8462852 on the detector and stars with similar stellar properties . in section [ sec : discuss ] , we rule out instrumental effects that could be responsible for apparent brightness changes and consider whether background contamination , a transiting cloud of material , or a polar spot can explain the observed ffi light curve .
we present our conclusions in section [ sec : conclusions ] .
we downloaded all 53 ffi images from the mikulski archive for space telescopes ( mast ) and sorted by the season in which they were collected . in this work ,
we use the `` cal '' frames , which have been calibrated from the raw data @xcite .
specifically , these frames have had a bias level and dark current subtracted , the smear from ccds being illuminated during readout corrected , the gain divided , and a flat field correction applied .
we removed one frame , 2009170043915 , as it was collected with the spacecraft mispointed by four pixels on the sky , making calibration with the other frames in that spacecraft orientation infeasible .
fluxes from the 2009170043915 frame differ from the other frames in that particular orientation taken during that same quarter by 0.5% , suggesting that the offset in the flat field from pixel to pixel is at the sub - percent level . for the remaining 52 frames ,
we first select a @xmath15 pixel ( 8 arcmin square ) region of the detector centered on the target star ( figure [ fig : ffi ] ) . as a part of the ffi data processing ,
a background has already been subtracted . unlike in the _ k2 _
mission , which observes in the ecliptic and detects a changing background due to increasing zodiacal light during each campaign @xcite , the background level is low and does not significantly change during the _ kepler _ mission .
we begin our processing by masking any saturated stars . while _ kepler
_ was designed with preservation of photometry of saturated stars in mind , recovering the absolute brightness of these stars requires careful placement of photometric apertures , making automated procedures challenging .
we then identify the ten brightest nearby stars in the subframes based on their brightest central pixel .
for all selected stars , including our target , we perform aperture photometry . the _ kepler _ telescope is defocused to produce an image with a full - width at half - maximum of 1.5 pixels in diameter . however , due to the location of kic 8462852 in the corner of the _ kepler _ field , the schmidt optics of the telescope produce a non - gaussian and asymmetric psf for each star @xcite .
as can be seen in figure [ fig : lffi ] , these psfs are several pixels in size and extend well beyond the pixels downloaded at 30-minute cadence during standard _ kepler _ observations . to collect all the flux from the star , we then create @xmath16 pixel apertures centered on the brightest pixel of each star .
not all of these stars may be appropriate as calibration stars for relative photometry ; some may be intrinsically variable , while others may be too faint to reliably obtain accurate photometry in each single exposure .
we discard all stars with a standard deviation of the residuals from a linear fit to the data larger than 0.5% .
such a cut removed one bright star near kic 8462852 ( kic 8462738 ) with periodic variations at the @xmath17% level as well as the two faintest stars in the sample , where photon noise prohibited accurate relative photometry , leaving us with seven reference stars .
we then determine the flux level of kic 8462852 by comparing its aperture photometry to the aperture photometry recorded from the calibrator stars , weighted by the signal to noise ratio of each of the calibration stars .
we use the standard deviation of the flux measured among all observations recorded on that particular pixel as the uncertainty associated with each observation . the scatter between points on
a particular detector can vary by as much as a factor of two from detector to detector , in line with previous analyses of the noise in the primary _ kepler _ mission @xcite .
finally , we combine the data from each separate channel on the telescope . because of the uncertainty in the underlying flat field , we expect an offset between each telescope orientation which is not known _ a priori _ , qualitatively similar to those observed in the _ kepler _ long cadence data between quarters .
we first normalize the flux values as recorded in the previous paragraph by dividing the observations from each particular channel by their mean value .
we then apply a linear offset term between each channel , so that if the sensitivity of one particular channel is lower than the others , our model would not try to model that signal is actual astrophysical variation . in practice , this requires fitting three parameters defining the offset for each channel relative to the first channel .
the results are shown in figure [ fig : photometry ] and table 1 .
for display purposes in figure [ fig : photometry ] we must select a particular set of offsets .
we choose the maximum likelihood model of a linear fit to the first three years of data from the _ kepler _ mission , which minimizes the scatter from quarter to quarter . in our analysis of any long - term dimming , we allow the offset terms to float as free parameters .
ccccccc 112.742 & 277998 & 1384241 & 0.200831 & 1.00896 & 0.00163 & 3 + 112.867 & 278052 & 1384642 & 0.200811 & 1.00886 & 0.00157 & 3 + 113.018 & 278029 & 1384472 & 0.200820 & 1.00890 & 0.00158 & 3 + 113.234 & 277993 & 1384621 & 0.200772 & 1.00867 & 0.00161 & 3 + 113.338 & 278018 & 1384317 & 0.200834 & 1.00898 & 0.00161 & 3 + 113.550 & 278053 & 1384618 & 0.200816 & 1.00888 & 0.00163 & 3 + 113.733 & 278046 & 1384486 & 0.200830 & 1.00895 & 0.00162 & 3 + 114.166 & 277968 & 1384404 & 0.200785 & 1.00873 & 0.00165 & 3 + 229.825 & 283733 & 1408715 & 0.201413 & 1.01067 & 0.00257 & 0 + 258.006 & 283423 & 1407913 & 0.201307 & 1.01014 & 0.00267 & 0 + 290.086 & 268042 & 1332411 & 0.201171 & 1.00574 & 0.00153 & 1 + 320.980 & 268832 & 1332528 & 0.201746 & 1.00861 & 0.00162 & 1 + 349.037 & 268720 & 1331627 & 0.201798 & 1.00887 & 0.00156 & 1 + 382.955 & 275452 & 1377897 & 0.199908 & 1.00908 & 0.00484 & 2 + 383.035 & 275495 & 1378097 & 0.199910 & 1.00909 & 0.00483 & 2 + 412.766 & 275353 & 1381124 & 0.199369 & 1.00636 & 0.00460 & 2 + 441.740 & 273902 & 1381975 & 0.198196 & 1.00044 & 0.00502 & 2 + 474.535 & 276125 & 1379789 & 0.200121 & 1.00541 & 0.00156 & 3 + 503.428 & 275547 & 1378483 & 0.199891 & 1.00427 & 0.00154 & 3 + 537.695 & 275378 & 1375253 & 0.200238 & 1.00600 & 0.00173 & 3 + 566.057 & 279781 & 1402980 & 0.199419 & 1.00100 & 0.00256 & 0 + 597.811 & 280641 & 1400955 & 0.200321 & 1.00520 & 0.00248 & 0 + 628.829 & 280421 & 1400350 & 0.200251 & 1.00485 & 0.00266 & 0 + 659.806 & 265617 & 1324798 & 0.200496 & 1.00238 & 0.00153 & 1 + 689.762 & 266120 & 1324664 & 0.200896 & 1.00437 & 0.00157 & 1 + 719.084 & 265682 & 1324891 & 0.200531 & 1.00255 & 0.00158 & 1 + 752.576 & 272275 & 1368110 & 0.199015 & 1.00458 & 0.00483 & 2 + 781.739 & 271728 & 1372220 & 0.198021 & 0.99956 & 0.00462 & 2 + 844.444 & 273245 & 1370321 & 0.199402 & 1.00183 & 0.00166 & 3 + 873.644 & 272944 & 1368418 & 0.199922 & 1.00211 & 0.00163 & 3 + 905.459 & 272206 & 1365061 & 0.199409 & 1.00186 & 0.00159 & 3 + 936.477 & 275546 & 1391437 & 0.198030 & 0.99371 & 0.00247 & 0 + 968.762 & 276586 & 1389492 & 0.199055 & 0.99885 & 0.00236 & 0 + 999.801 & 277627 & 1387847 & 0.200415 & 1.00380 & 0.00248 & 0 + 1031.800 & 263255 & 1316791 & 0.199922 & 0.99951 & 0.00153 & 1 + 1062.757 & 263785 & 1317059 & 0.200283 & 1.00132 & 0.00151 & 1 + 1097.862 & 263738 & 1316417 & 0.200345 & 1.00162 & 0.00156 & 1 + 1125.427 & 269393 & 1357792 & 0.198405 & 1.00150 & 0.00453 & 2 + 1153.523 & 268327 & 1359507 & 0.197371 & 0.99628 & 0.00474 & 2 + 1181.558 & 266378 & 1361569 & 0.195641 & 0.98754 & 0.00505 & 2 + 1214.517 & 266884 & 1357982 & 0.196530 & 0.98748 & 0.00168 & 3 + 1244.453 & 264824 & 1356515 & 0.195224 & 0.98096 & 0.00166 & 3 + 1272.590 & 264792 & 1353002 & 0.195707 & 0.98337 & 0.00166 & 3 + 1304.527 & 268808 & 1380987 & 0.194649 & 0.97677 & 0.00250 & 0 + 1335.832 & 269135 & 1378297 & 0.195266 & 0.97986 & 0.00237 & 0 + 1370.855 & 268435 & 1377283 & 0.194902 & 0.97803 & 0.00256 & 0 + 1403.835 & 255025 & 1308199 & 0.194944 & 0.97467 & 0.00168 & 1 + 1434.914 & 255839 & 1307564 & 0.195661 & 0.97825 & 0.00150 & 1 + 1470.673 & 255686 & 1307448 & 0.195561 & 0.97775 & 0.00156 & 1 + 1497.564 & 261275 & 1345623 & 0.194167 & 0.98010 & 0.00469 & 2 + 1524.495 & 260101 & 1347812 & 0.192980 & 0.97411 & 0.00456 & 2 + 1557.495 & 261323 & 1349036 & 0.193711 & 0.97780 & 0.00496 & 2 [ tab : results ]
from figure [ fig : photometry ] , three main features are apparent .
the first is a long - term dimming at a linear rate from the beginning of the _ kepler_mission until approximately quarter 13 .
the dimming then increases rapidly such that the observed flux from the star decreases by approximately 2.5% over 200 days . for the final 200 days of the _ kepler _ mission , the light curve flattens out , either returning to the original rate of decay or remaining constant . in the primary _ kepler _ mission ,
data were collected in four orientations as the telescope rolled every 93 days to keep its heat shield pointed at the sun . as a result ,
data for each star are collected on four different pixels , each for three months of every year .
therefore , we can separate many instrumental effects caused by a faulty pixel on the detector from astrophysical phenomena by looking for the same trends when only considering data from each specific orientation . in practice , we allow for a linear offset between the observed fluxes recorded on each individual module , reflecting our uncertainty about the underlying flat field of the detector itself . we note that these results are apparent in data from each individual detector , not just the combined light curve , suggesting that the decline in flux is an astrophysical effect rather than an instrumental one , as discussed more fully in section [ sec : discuss ] .
we fit a line to the first region of data , from the beginning of the mission until barycentric julian date ( bjd ) 2456003 ( or bjd-@xmath18 ) , encompassing 39 data points .
we allow the relative offset between each detector to vary , three in total to account for possible differences in the flat field between different detectors .
it is possible that the recorded measurement uncertainties are not representative of the actual uncertainties . to account for a possible systematic underestimation of the uncertainties , we allow for an extra uncertainty term applied to data from each detector to be added in quadrature to our recorded uncertainties estimated from the data themselves , so that the @xmath19th data point recorded in the @xmath20th telescope orientation has uncertainty @xmath21 such that @xmath22 here , @xmath23 is the uncertainty for the @xmath19th observation ( as described in [ sec : data ] ) and @xmath24 is the level of underestimation of the photometric uncertainty in quarter @xmath20 , similar to the concept of jitter in radial velocity ( rv ) observations @xcite .
we assume all observations on a given detector have the same @xmath24 value , so that there are four @xmath24 values fit in our analysis .
including a slope and zeropoint for the linear fit , there are a total of nine free parameters : the slope , zeropoint , three offset terms describing the linear offset for each quarter relative to the first , and four `` jitter '' terms describing the excess photometric noise in each channel above the listed uncertainties .
the resultant fit estimates the excess noise required in order to explain the data for the first three years of the mission as a straight line , and estimates the offset in flux between the different spacecraft orientations under the assumption that the rate of change of flux is constant over the first three years of the mission .
we note that the data shown in figure 3 and table 1 are the uncertainties from the data themselves ( @xmath23 ) rather than @xmath21 . by marginalizing over all parameters except the slope
, we can measure the decay rate of the observed flux .
we determine that the star is fading from our perspective at a rate of @xmath1 yr@xmath2 , or @xmath25 parts per thousand ( ppt ) yr@xmath2 .
this is equivalent to fading at a rate of @xmath26 magnitudes per century , exceeding the purported 0.165 mag century@xmath2 ( @xmath8 yr@xmath2 ) rate of dimming over the duration of the dasch plates @xcite by more than a factor of two .
the total change in brightness over this portion of the _ kepler _ light curve is almost 1% .
the four maximum likelihood `` jitter '' values are 2.8 , 0.52 , 1.1 , and 0.28 parts per thousand , corresponding to orientations 0 , 1 , 2 , and 3 respectively in the parlance of table 1 , suggesting that the recorded uncertainties listed there are not significantly underestimated . beginning around or bjd@xmath27 , lasting for approximately 200 days , the rate of dimming increases dramatically . repeating the above exercise in the region between dates 1100 and 1250
, we measure a decay rate of @xmath28 yr@xmath2 . here
, we allow the break points at which the slope changes to float as free parameters , and the uncertainty in the actual dimming rate during this event is dominated by the uncertainty in the exact timing of the change in slope .
regardless of the exact rate at which the star appears to dim from our line of sight , it is clear that from the _ kepler _ data , the star becomes approximately 2.5% dimmer over a period of 200 days .
this rapid dimming is apparent in data from all four orientations of the _ kepler _ spacecraft .
we note that we would expect the _ kepler _ pre - search data conditioning ( pdc ) pipeline to remove signals such as this one in the pre - processing of the data , explaining why this signal does not appear in the originally published long - cadence light curve .
after the conclusion of this event , there are only nine epochs of ffi data . over this range
, the rate of change of flux from the star is poorly measured : it is consistent with the original rate of dimming , but also consistent with a flat light curve until the final ffi image is collected at the end of quarter 16 of the _ kepler _ mission .
we offer no definitive explanation that could explain the observed light curve in this work
. the effect could be stellar in nature , although there are no known mechanisms that would cause a main - sequence f star to dim in brightness by 2.5% over a few months .
the effect could also be caused by a passing dust cloud in orbit around kic 8462852 .
indeed , the light curve at times larger than 1000 days has a morphology broadly similar to a transit event , although on markedly different timescales .
we discuss this morphology more fully in section [ sec : transit ] .
the long - cadence light curve as observed by _ kepler
_ contains features in which the observed flux from the star decreases by as much as 20% for a few days at a time .
our results could be substantially affected if any of these events were to overlap with the collection of a full frame image .
we can test this possibility by comparing the times recorded for the ffi data with the largest dipping events in the _ kepler _ light curve .
the result is shown in figure [ fig : lc ] . .
none of the ffis are taken during a dip that could significantly affect our photometry .
we produce our own long cadence light curve as explained in section 5.3 , finding no evidence for the rise before the dip at the end of quarter 8 observed in the pdc light curve .
the light curve in this figure is available as the data behind the figure associated with figure 3 and available in the online version of this journal .
, scaledwidth=45.0% ] we find that the dips do not overlap with any of the ffi observations .
the only ffi image that is sufficiently close to a dipping event to possibly be affected is the final ffi of quarter 8 .
if we repeat our analysis removing this data point , we recover the same long - term dimming ; none of the results in this work are significantly affected by the inclusion or exclusion of this single data point .
ffi images are typically collected as the last observation of a month of continuous data collection immediately before data downlink to earth .
after the ffi is taken , there is typically a @xmath291 day gap before data collection is resumed .
it is unlikely that a large dip could fall in one of these gaps while evading detection in the long cadence data , as the ingress duration for a dip would need to be smaller than the 30 minutes between the last long cadence image of a single month and the ffi collected immediately afterward .
as can be seen in figure 1b of @xcite , some of the dipping events are surrounded by apparent flux increases of @xmath30 .
if these increases are real , they could be the signature of forward scattering from dust grains ( e.g. , * ? ? ?
most notably , the pdc light curve produced by the _ kepler _
mission appears to rise just before a 15% dip .
this particular ffi frame records a lower flux value than its predecessor by approximately 1 mmag , although the two are consistent with no change in flux . to test whether the small rise in the pdc light curve is induced by the initial data processing in the creation of the pdc light curves or
if it is real , we create our own long cadence light curve from the pixel - level data .
we perform aperture photometry using the entire postage stamp downloaded by the _
kepler_mission , not just the smaller aperture selected by the _
kepler _ team for the creation of sap and pdc light curves .
we do the same for kic 8462934 , the single star within 2 arcmin with @xmath31 within one magnitude of kic 8462852 .
we then divide the flux recorded for kic 8462852 by the flux recorded for its apparent neighbor to account for instrumental trends . in the full light curves ,
we detect a significant variation in the total flux received from quarter to quarter , a sign of changes in the underlying flat field from channel to channel . for each quarter
, we fit an offset term to minimize the variation between quarters ( see [ sec : longtermdimming ] ) .
all light curves produced from _
data contain long - term variability that can result from a combination of astrophysical and instrumental effects . with the ffi data , we are able to use the nearby stars on the detector to separate the astrophysical effects from any instrumental effects shared by nearby stars . however , as the vast majority of stars in the kepler input catalog
were not targeted for regular photometry during the _ kepler _ mission , there is a paucity of stars with long cadence data to compare against to robustly detect long - term astrophysical trends in the long - cadence data . in this case , for the purposes of plotting figure 4 we remove long - term trends .
we divide the light curve into distinct regions separated by gaps in the data of at least 0.5 days .
we then mask the regions of the light curve identified as short - term dips by @xcite and fit a spline to the remainder of the data , dividing the measured light curve by the spline to flatten the light curve .
the result is a normalized light curve that removes trends but preserves the shape and magnitude of the rapid dips .
these data are shown in figure [ fig : lc ] and we make them publicly available as `` data behind the figure '' associated with figure 3 .
once we have re - processed the light curve , we can compare the rising events to those observed in the pdc light curve .
the rise in the pdc light curve in q8 does not exist in our re - processed light curve , suggesting that it is an artifact of the processing in the development of the pdc light curve .
similarly , the rise in the light curve at day 1148 , observed in the pdc light curve and this light curve , is almost certainly an artifact of data processing .
there are also small rises observed before the dips at dates 140 and 260 .
we note that these also occur immediately after data downlinks , as the telescope is reestablishing thermal balance after rotating to point to earth and changing the position of its heat shields .
it is well established that this maneuver leads to a ramp in the observed flux lasting @xmath32 days after observations resume @xcite .
as both rises occur during these windows , it is likely that they are induced by these thermal effects .
there is a small rise ( @xmath33 mmag ) observed at the end of quarter 13 that may be real ; it appears at low significance in the ffi data as well . while our re - processed light curve is useful for studying the short - term dimming and potential rising events in the light curve of @xcite , it and other versions of the _ kepler _
long cadence data for this star are not well - suited for the detection of long - term variations in the observed flux ( see [ sec : intro ] ) . the total aperture recorded by _ kepler _ at 30-minute cadence is smaller than the psf of the detector , so small changes in the position of the star cause flux variations that easily exceed the magnitude of the long - term trend . moreover ,
many of the nearby reference stars used in this work that should share systematics with kic 8462852 were not observed at 30-minute cadence , meaning the opportunity to co - trend systematics as robustly as done here is lost . the best opportunity to create a light curve that is accurate
both on traditional _ kepler_timescales and over years is to combine the ffi data with long cadence data , as we discuss in section 5.3 .
this light curve can also be used to test the robustness of some of the signals in the pdc light curve .
the long - term dimming of kic 8462852 observed during the first 12 quarters of the _ kepler _ mission differs from zero by @xmath34 . in modeling the light curve , we assumed that the sources of uncertainty are entirely statistical in nature . if there are other , long - term systematics that cause a slow drift in the recorded _ kepler _ flux from a star , those could evade our statistical analysis and cause us to erroneously record a long - term trend that does not exist .
if such systematics exist , their effect should be detectable for many stars , so we can compare the observed light curve in our analysis with other stars in the _ kepler _ field to see how often a star exhibits a dimming of @xmath1 yr@xmath2 .
we repeat the analysis in section [ sec : results ] , performing photometry on other stars that fall onto the @xmath35 arcminute postcard around kic 8462852 .
if long - term trends of a similar magnitude are recovered in any of these , it is an indication that the _ kepler _ photometry is dominated by systematics from one or a few active stars in particular . and
are located within @xmath36 of kic 8462852 on the sky . in no cases
do we detect a similar dimming to that observed for kic 8462852 , suggesting that the dimming is not caused by systematics due to variability of nearby reference stars .
, scaledwidth=45.0% ] to test for such systematics , we repeat our process to develop light curves for eight other stars , all with @xmath37 and all located within @xmath36 ( 60 pixels ) of kic 8462852 on the sky . in these cases , kic 8462852 was not selected by our pipeline as a potential reference star due to its photometric variability relative to the other nearby stars .
the photometry for these stars is shown in figure [ fig : refs ] .
none of these stars exhibit a large offset in either direction around 1100 days into the mission , at the time of the rapid change in flux of kic 8462852 .
we can fit a long - term trend to each of these stars to measure the photometric long - term variability as recorded by _
kepler_. in practice , we fit five parameters : a slope , a zeropoint , and three relative offsets between observing seasons as the star falls on different detectors , maximizing the likelihood of this model to the data .
we also inspect each light curve by eye to look for significant deviations from non - linearity similar to that detected for kic 8462852 . for none of these stars
do we recover a slope as extreme as @xmath25 ppt yr@xmath2 .
similarly , we do not detect any 2.5% decreases in flux between days 1100 and 1250 ( or at any other time ) , suggesting that this effect observed in the light curve of kic 8462852 is not due to a spurious reference star , nor is there a systematic effect in all nearby stars .
we do note that we recover the same changes in photometric precision as observed for kic 8462852 as the stars move from one detector to the next while the spacecraft rolls during its orbit .
we note that for some of the targets included in figure [ fig : refs ] , the scatter between data points is larger than the photometric error bars .
this is particularly true for kic 8462696 and kic 8462836 . in these cases ,
the scatter is likely due to astrophysical variability rather than an underestimation of our uncertainties . for the former , long cadence photometry exists from the primary mission .
this star exhibits a clear photometric modulation due to starspots . here
, spots modulate the flux of the star by 3% with a rotation period of 16 days .
because the rotation period is close to half the period of ffi observations , the phase of the star s rotation when the telescope collects an ffi observation varies slowly , leading to features such as the apparent trend in the ffi flux between day 400 and 800 .
we note that this time period matches well with the peak of starspot activity for this star . for kic 8462852 we know from the _ kepler _ data that the amplitude of photometric modulation is considerably smaller than the photometric uncertainties in the calibrated ffi data ,
so such a false positive trend can not occur . for kic 8462836 , we do not have _ kepler _
long cadence data .
photometry of the target is consistent with that expected of a mid - m dwarf . for mid - m dwarfs , it is not uncommon to observe photometric variability caused by starspots at the 3 - 5% level @xcite , so again
, the excess scatter is likely the result of starspot induced variability that should not be present for kic 8462852 .
we can extend our analysis to search for variations in not only the most nearby stars used as reference stars , but to all stars on the same detector .
we select all stars with @xmath31 within 0.5 mag of kic 8462852 and located in the same `` skygroup , '' meaning they all fall on the same channel at the same time .
we remove stars that would not be expected to provide reliable tests in a search for systematic trends in the _ kepler _ data , including known variable stars , eclipsing binaries , and those with starspot - induced variability at the level of 1% or more through a visual inspection of the long cadence light curves .
these cuts leave 193 viable stars against which to compare .
we find , of these 193 stars , for only one does our automated pipeline measure a long - term trend : kic 8395126 appears to decrease in brightness over the _ kepler
_ mission , with a maximum likelihood measurement of the slope of 3.9 ppt yr@xmath2 .
however , on closer inspection of the data for this star , we find that its center of light is approximately one pixel from the position listed in the kepler target database available on mast . reanalyzing this target after accounting for this one pixel offset causes the decay rate to decrease to 1.7 ppt yr@xmath2 , half that of kic 8462852 .
we note that by repeating our analysis for kic 8462852 while moving the centroid in any direction does not change the light curve beyond the level of the quoted photometric uncertainties .
the measurements of the maximum likelihood slopes for the full sample of 193 nearby stars , again fitting for a slope and four offset terms , are distributed with a mean of @xmath38 ppt yr@xmath2 and a standard deviation of 1.23 ppt yr@xmath2 .
we plot a kernel density estimator ( kde ) of the distribution of fitted slopes in figure [ fig : kde ] . for all cases ,
we apply a gaussian kernel with a bandwidth of 0.41 ppt yr@xmath2 equal to the uncertainty in the measured q0-q12 slope for kic 8462852and a reasonable estimate of the typical measurement uncertainty .
we find that none of the comparison stars and only 0.3% of the mass of the kde have a measured variation equal to or larger than the maximum likelihood value of the slope of @xmath39 ppt yr@xmath2 recorded over the first three years of the _ kepler _ mission for kic 8462852 .
even more significantly , there is no mass at 7.8 ppt yr@xmath2 , the value recovered if one were to blindly fit a linear model to the distinctly non - linear light curve for kic 8462852 . in no cases do we observe a 2.5% dip over 6 months as observed for kic 8462852 , suggesting that the observed behavior of our target star is unique compared to all the comparison stars on the same detector .
we can also compare the flux variations observed for kic 8462852 to ffi light curves for stars of similar spectral types . here
, we use the sample of kic stars developed by @xcite and considered in their analysis of the dasch plates . in that work ,
the authors selected 559 stars listed in the updated kepler input catalog @xcite with inferred stellar effective temperatures within 100 k , radii within 5% , and @xmath40 within 10% of kic 8462852 . from this sample
, we remove all stars for which we do not expect to be able to acquire reliable photometry or observe long - term photometric trends over four - year baselines .
we remove stars with brighter neighbors located within ten pixels of the target star , so that their flux is likely to leak into our own aperture .
we also remove stars near the edge of the detector and stars that do not fall onto the _ kepler _ detectors in at least three of the four observing seasons .
we remove stars that saturate the detector , and also those that are intrinsically variable , including contact binaries , stars that vary due to spots at larger than the 2% level , and pulsating variable stars .
we note that the two targets that @xcite noted as long - term variable , kic 3868420 and kic 11802860 , both are removed by this last requirement .
the former exhibits 15% variability with a five hour period in the _ kepler _ light curve ; @xcite suggest it may be a high - amplitude delta scuti . the latter exhibits 40% variability with a 16.5 hour period in its _ kepler _ light curve ; the star is also known as aw dra and has long been known as an rr lyrae @xcite .
the long - term variability observed by @xcite may be the result of unfortunate timing of the observations across different phases of the star s light curve , causing an apparent long - term trend .
this is plausible as the claimed photometric trends have different signs in the pre- and post-1970 subsets of the dasch data .
the long - term variability could also be a case of rapid stellar evolution : @xcite suggest rr lyrae stars may be undergoing rapid mass - loss , which may lead to a dimming over a timescale of decades .
we note that the overall distribution of f star variability is well - approximated by a gaussian with mean @xmath41 ppt yr@xmath2 and standard deviation 0.094 ppt yr@xmath2 .
stellar evolution does not explain a long - term trend in the typical f star ; detection of a nonzero mean dimming is likely indicative of a systematic in the _ kepler _ data that is not understood .
we repeat our analysis on these stars . again , none of the target stars exhibit a 3% change in flux over the duration of the _ kepler _ mission , nor do any stars change in flux by 2.5% over six months . of all stars analyzed in this work
, the rapid dimming between quarters 12 and 15 appears to be unique to kic 8462852 .
in fact , only a single other star , kic 5868753 , has an observed change in brightness larger than the maximum likelihood slope for kic 8462852 from the first three years of the _ kepler _ mission .
the _ kepler _ ffi light curve for kic 8462852 features a long - term slow dimming over the first three years of the mission . formally , the observed slope is statistically significant , but from our analysis of nearby stars it is likely that systematic uncertainties dominate over the statistical uncertainties .
there are no stars that we detect that exhibit a 7.8 ppt yr@xmath2 dimming , the level that a simple linear model would produce when fit to the full , nonlinear data set . additionally , there are no stars that exhibit a 2.5% decrement in flux over approximately six months .
even if we consider only the relatively modest dimming over the first three years of the _ kepler _ mission , there is only a single star that exhibits a slope that is at least as large in magnitude as the one observed here .
the observed behavior of the star is thus very likely astrophysical in nature , suggesting this star is indeed undergoing some process leading to a decrease in its observed brightness over the span of the _ kepler _ mission . in the remainder of this section we consider possible interpretations of this result . as mentioned in section [ sec : intro ] , @xcite used 99 years of photometry from the dasch project to analyze the behavior of kic 8462852 , finding a decrease in brightness of 14% from 1890 to 1989 .
@xcite performed an independent analysis of the dasch photometry , confirming that the photographic data yield a fainter magnitude for kic 8462852 in the late 20th century compared to the end of the 19th century .
however , @xcite argue that the dasch measurements from 1890 to 1952 are best described by a constant brightness and measurements from 1967 to 1989 are best described by a different ( fainter ) constant brightness , with systematic errors accounting for the offset .
we note that such systematic offsets have not been detected in dasch light curves by the dasch team @xcite .
the unfortunate gap of 15 years with very few observations in the middle of the century makes it difficult to distinguish between a real astrophysical variation and a systematic offset .
which explanation to prefer then depends on one s assessment of whether a star steadily fading for a century or a change in the photometric calibration of the harvard plates around 1962 is more likely ( or less unlikely ) .
the fading that we detect from 2009 to 2013 with the _ kepler _ ffi images does not necessarily represent a confirmation that kic 8462852 also dimmed over the preceding 129 yr , but could make that interpretation of the dasch data more plausible . both previous analyses considered only linear models for the flux of the star as a function of time , as the dasch data are insufficient to constrain more complex models .
the _ kepler _ ffi data are plainly nonlinear , suggesting that the true light curve during the 20th century is likely more complex than either simple model . additionally , a decrease of 3.5% in the flux of kic 8462852 is four times what would be expected over the four - year _ kepler _ mission from @xcite s century - long light curve .
conversely , if the brightness change over the duration of the dasch survey were as extreme as that observed during the _ kepler _ mission , we would expect the star to have decreased in brightness by 60% over the 20th century , which does not appear to be consistent with comparison of archival and modern ( e.g. , * ? ? ?
* ) measurements . as there is a gap of two decades between the end of the dasch data and the start of the _ kepler _ mission ,
a direct comparison between the two is difficult .
still , it is clear that neither a linear decrease in flux nor a constant flux model provide a good description of our observations , so there is no reason to expect the behavior at earlier times to have followed a simpler form .
the observed dimming does not seem to be associated with any instrumental effect .
some known effects could cause a 2.5% decrement in the observed flux from a star .
the most common such effect is a sudden pixel sensitivity dropout ( spsd ) , in which a pixel abruptly decreases in sensitivity , leading to a decrease in the observed flux on that pixel .
the pixel generally recovers in sensitivity over a few hours , but does not return to the same sensitivity level as before @xcite .
an spsd could not explain our observed light curve , as the observed flux decrement is visible in all four detector orientations , so the same spsd would have had to occur on all four apertures , at the same level , at approximately the same time .
similarly , we do not expect this effect to be the result of any calibration issues during the processing of the ffis .
the calibration pipeline accounts for the bias and dark current of the detector , cosmic rays , smearing effects from reading out without a shutter , and distortions induced from the readout electronics @xcite .
most of these effects are dependent on the location of the star , rather than the physical properties of the star itself , so we would not expect them to induce a dimming observed over all four modules .
the exception is the smear correction , but the smear is well - understood and can be robustly estimated from the data .
significant errors in the smear correction would be easily observable in the ffi data itself as `` trails '' in the ffi image near the star , which are not detected in these observations .
therefore , we can rule out instrumental effects or calibration errors as a plausible source of the observed variability .
the long cadence light curves produced by the _ kepler _ mission are accurate over timescales of hours to days . over longer timescales , systematics overwhelm long - term astrophysical photometric variations .
this is largely beause of two effects .
first , there is a lack of reference stars , as only @xmath42 of all stars in the _ kepler _ field are targeted for photometry at 30-minute cadence .
therefore , nearby stars that should share systematics are often not observed at the cadence that would be necessary for co - trending to identify shared systematics .
second , the apertures downloaded by _ kepler _ are typically smaller than the psf of each individual star ( figure 2 ) , so that small changes in the pointing of the telescope can cause significant flux variations , overwhelming any signal from the star itself . with the ffi data in hand , we create our own long - cadence light curve from the pixel - level data .
as the ffi data provide absolute calibration of the _ kepler _ data once per month , the ffi data can be combined with the flattened _ kepler _ long cadence light curve to show one plausible iteration of the true astrophysical long - cadence light curve of kic 8462852 , with instrumental effects removed . to create this light curve
, we fit a cubic spline to the ffi data using the spline tool in ` scipy.optimize ` .
we use the measured photometric precision for each ffi as the input weights , applying a smoothing factor of 50 . the resultant smooth light curve
is then multiplied by the flattened light curve of section 3.3 to create a realistic long - cadence light curve that captures both the short - term variability and the long - term dimming observed during the _ kepler _ mission .
these data are plotted in figure [ fig : photometry ] and included in the same `` data behind the figure '' as both the intermediate , flattened light curve described in the previous paragraph and the smooth spline fit to the ffi data . as the data are not calibrated between the ffis , and each ffi observation has an associated photometric uncertainty , this long - cadence light curve represents one possible realization of the photometric variability of kic 8462852 and
should only be considered in that context .
approximately 25 arcsec from kic 8462852 is kic 8462860 , a star 3.65 magnitudes fainter on the edge of our photometric aperture . as approximately half of the star falls in our simple aperture
, we would expect it to contribute approximately 1.8% of the total flux in our aperture .
this star can not explain the long - term trend over the first three years of the mission unless it were to decrease in flux by 50% ; it can not explain the rapid dimming event unless it disappeared entirely ( which is obviously not the case ) .
kic 8462860 could explain the observed rapid dimming if the star had a large proper motion and moved out of the aperture during quarters 13 and 14 .
however , in the ppmxl catalog @xcite its proper motion is @xmath43 mas yr@xmath2 , @xmath44 mas yr@xmath2 . the first _ gaia _
data release does not include a proper motion measurement of kic 8462860 . by combining the _ gaia _ measurement of the position of the star @xcite with the position recorded in the 2mass point source catalog @xcite , we find the proper motion must be less than @xmath45 mas yr@xmath2 , in line with the ppmxl result . the star is listed in the kic as having a temperature of 5464 k , and its magnitude and colors ( @xmath46 = 15.6 , @xmath47 = 1.4 , @xmath48 = 0.5 ) are broadly consistent with a late g or early k dwarf at several hundred parsec or an evolved g / k star at larger distances .
the photometry is not consistent with what would be expected of a nearby star .
if the star did have a large proper motion , we would see it move by several pixels over the course of the mission , which visual inspection of the ffi data does not show .
moreover , to move far enough so that the entire core of the psf moved across the edge of our aperture over a span of approximately six months , the proper motion would need to be approaching 10 arcsec yr@xmath2 . in this case
, we would expect the star to traverse the entire aperture over the four years of ffi data , which we do not observe .
perhaps most significantly , the same light curve as shown in figure [ fig : photometry ] is recovered when we mask the pixels corresponding to kic 8462860 , or if we modify the size of our aperture to either move this other star fully inside or outside of the aperture rather than on the edge .
therefore , we can exclude the possibility that the observed variations in the light curve are caused by the presence of kic 8462860 .
moreover , we do not detect the presence of any centroid shifts correlated with the measured flux in the light curve of this star , suggesting this result is not the effect of the neighboring star or changes in the underlying flat field .
we can rule out background contamination as explaining our light curve with high confidence .
we note that the shape of the light curve after quarter 12 appears broadly similar to that of a transit event . over six months
, the rapid dimming could represent the ingress of material blocking the stellar disk , leading to a decrease in observed flux .
the primary mission then ends before third contact is observed .
clouds of transiting material due to disintegrating bodies have been observed in _ kepler _ data , but only with very short orbital periods @xcite . in the case of kic 8462852
, there are problems with this model .
first , it does not explain the observed long - term dimming over the first three years of the _ kepler _ mission , nor does it explain the long - term trend observed over the past 100 years through the dasch data .
a single transit model also does not explain the phenomena observed by @xcite , especially those dips before the time of ingress .
nevertheless , we investigate whether it is a possible explanation for the rapid dimming .
a transit event with a timescale similar to that observed here is not unprecedented .
@xmath49 aurigae is transited every 27 years , with a transit lasting approximately 1.5 years @xcite .
recently , @xcite observed transits of tyc 2505 - 672 - 1 with a period of 69.1 years and a duration of 3.45 years . in both cases ,
the primary is orbited by a hot source that hosts an extended disk of circumstellar material .
@xcite confirmed this model with closure - phase interferometric imaging , directly imaging the disk occulting @xmath49 aurigae . for both @xmath49 aurigae and tyc 2505 - 672 - 1 ,
the disk is optically thick and circumstellar around a binary companion .
the sub - millimeter observations of @xcite place a limit of 7.7 m@xmath50 of material within 200 au of kic 8462852 , ruling out a direct analogue of these other systems .
first , we consider that this signal represents the transit of a solid body across the stellar disk of kic 8462852 . in this scenario
the observed timescales of the event constrain the size and distance of the eclipsing object . for an optically thick transiting object ,
the 2.5% transit depth indicates a minimum radius of 0.15 r@xmath51 ( @xcite estimate a radius of 1.58 r@xmath52 for kic 8462852 ) .
if the transiting body is in a keplerian orbit , the extremely slow ingress time and long transit duration place it at the implausibly large distance of @xmath53 pc , with a transit probability of @xmath54 .
note that while a six - month transit ingress and transit duration of more than six months are similar to @xmath49 aurigae and tyc 2505 - 672 - 1 , this case is distinct from those because of the much smaller radius of kic 8462852 .
the transit of a main sequence star is necessarily much shorter than that of a supergiant with @xmath55 r@xmath52 . even if every star had a companion at 10 pc , then given @xmath56 _ kepler _
missions we would expect to observe one such transit event ; we can confidently disfavor this hypothesis .
if the transiting material is instead an optically thin cloud , then its size could be comparable to or larger than that of kic 8462852 .
the optical depth of such a cloud would be @xmath57 . because the orbiting body considered in the previous paragraph was already a significant fraction of the stellar radius , though , a much bigger cloud does not qualitatively change the result : in order for the transit ingress to last @xmath58 d , a cloud in a keplerian orbit would need to be located @xmath59 pc away from the star .
we thus conclude that a simple transit of any kind of orbiting object is not a reasonable explanation for the rapid fading that begins in quarter 12 . however , this conclusion relies on the assumption that the duration of the transit ingress is set by the orbital velocity of the transiting body .
more complex scenarios in which the ingress timescale reflects the spreading of debris along its orbit after a recent collision or the precession of an occulting disk into our line of sight could perhaps explain the appearance of a transit .
for example , a cloud that slowly increases in density would manifest itself in the light curve as inducing a change in flux similar to that observed in quarter 12 .
however , to produce the apparent flat bottom of the supposed transit event , such a cloud would then need to be extended over a fraction of its keplerian orbit and would need to maintain an approximately constant density through its entire length as it passes in front of this star . as illustrated above
, it is very difficult to come up with a physical model that can even qualitatively explain all of the major features of kic 8462852 s photometric behavior simultaneously . of the ideas that have been proposed so far , we suggest that the most promising explanation involves a recent collision between large bodies ( planetesimals or comets ) in the kic 8462852 system , leading to a spreading of debris as in the previous paragraph . in this picture ,
a recent collision could create a cloud of circumstellar material and push a family of objects into a highly eccentric orbit , analogous to the period of late heavy bombardment observed in our own solar system .
however , this idea does not naturally account for the steady decline in the flux of kic 8462852 in the years preceding the more rapid dimming and the concentrated sequence of dips .
the data presented in this paper can not fully exclude any of these models , but we note that the circumstellar dust and debris produced by such an event is unlikely to maintain this arrangement for long timescales .
the sub - mm limits of @xcite and continued photometric monitoring will significantly constrain future models that attempt to invoke circumstellar dust or transiting models to explain this light curve . transiting
material remains as a plausible explanation for the kic 8462852 light curve , but requires particular , _ a priori _ unlikely density profiles for the circumstellar material in order to match the data . under certain conditions ,
a long - lasting spot growing at polar latitudes on the surface of kic 8462852 could possibly reproduce the long - term light curve observed here .
while polar spots have not previously been detected on an f3v dwarf star , they have been observed on the surface of an f9v star , albeit one in a tight ( 1.15 day ) binary with another stellar companion @xcite . through interferometric aperture synthesis imaging , polar spots on more evolved stars
have been seen to evolve on similar timescales to the flux variability observed here @xcite . for a polar spot to create the observed decrease in flux in the light curve , the projection of the polar region onto the observer s line of sight would need to be large enough to allow for a large starspot to be observed . at high ( edge - on ) inclinations ,
a polar spot would be foreshortened , diminishing its effect on the light curve .
moreover , limb darkening would decrease the overall contribution of the polar regions on the light curve . at lower ( more face - on ) inclinations
, the pole would be always visible and a polar spot would have a larger effect on the overall light curve .
@xcite measure a rotation period and @xmath60 of the star and combine these with an estimate of the radius to infer an inclination of @xmath61 degrees at 68% confidence , leaving open the possibility that the pole can indeed be observed at all times well away from the edge of the stellar disk .
if a spot were growing near the polar latitudes of the surface of the star between quarters 13 and 15 , we might expect a corresponding increase in overall magnetic activity , leading to an increase in starspots at other latitudes over this time period .
this indeed appears to be the case .
@xcite measure a periodic signal with a period of 0.88 days , as well as additional signals at 0.90 and 0.96 days , all of which change in intensity during the _ kepler _ mission .
the authors suggest that this signal may be induced by starspots .
spots at different latitudes evolving and rotating differentially would produce a signal like this one .
the strength of the signal they observe grows between days 1100 and 1300 of the mission , suggesting either an increase in the number of spots or an increase in their coherence .
interestingly , the growth of the signal corresponds to the time of rapid decrease in total flux recorded in the ffi images .
the inclination of kic 8462852 and the coincidence between the spot activity and the rapid dimming do not present a confirmation of a polar spot . a long - lasting polar spot
itself would be remarkable given the f3v spectral type of the star , but with the current data the hypothesis can not be excluded .
broadly , the ffi observations are consistent with the growth of a polar spot that grows at a similar time to the growth of spots at lower latitudes .
spots , however , can not explain the short - term dips originally observed by @xcite .
more observations are needed to separate the spot hypothesis from other possible explanations of the observed light curve . as the polar spot hypothesis can not account for the short - term dips of @xcite
, it does not seem to be a particularly likely scenario , but can not be ruled out given the available data . in this work ,
we do not present a model that can explain the entire suite of observations of kic 8462852 .
there are now three distinct photometric variations observed : rapid , irregular decreases of @xmath62 or more in flux lasting for a few days , a @xmath63 decrease in flux lasting for at least one year , and a likely long - term dimming perhaps spanning more than a century .
additional observations would be helpful in order to better understand physical phenomena that could cause any or all of these events .
multi - color photometry is essential to help characterize this star .
if any of these events are caused by solid bodies , we would expect the photometric variations to be largely achromatic .
however , if they are caused by a cloud of dust and gas , we would expect the cloud to redden the star .
similarly , spots with a lower effective temperature than the rest of the stellar surface would cause an apparent change in the color of the star as they cover more of the stellar disc . a transiting cloud would be expected to induce dips that are largely periodic , while changes in the spot patterns would not necessarily be periodic .
high precision observations of the color of kic 8462852 over time could explain the nature of each of these events , supporting or ruling out the transiting cloud or spot hypotheses .
similarly , additional ir and sub - mm observations of the system could be used to place tighter upper limits on the amount of circumstellar material surrounding the star .
additional radial velocity ( rv ) monitoring to search for companions in few - au orbits around the star , especially those that could hold together a disk of circumstellar material into a gravitationally bound system , would also be useful .
@xcite obtained four high resolution spectra over the course of approximately 500 days .
the measured rv of kic 8462852 in these observations has a scatter of @xmath64 km s@xmath2 and each observation has a precision of @xmath65 km s@xmath2 . while the data are consistent with no rv variations , continued monitoring could detect the presence of massive companions on wider orbits .
high resolution spectra could also be used to probe any evolution in the magnetic activity of the star correlated with the growth or decay of spots .
recently , multiple analyses of dasch photometry have produced conflicting results about the possible detection of a long - term dimming of kic 8462852 by @xmath7 magnitudes over the 20th century , or @xmath8
. the dimming of a star at that rate should be detectable in _ kepler _ data . here
, we analyze monthly _ kepler _ ffi images to search for similar dimming in the ffi light curve of kic 8462852 .
we perform aperture photometry on kic 8462852 and seven nearby comparison stars .
we observe that during the first three years of the _ kepler _ mission , the star dimmed at a rate of @xmath66% yr@xmath2 . over the following 6 months , the star decreased in brightness by 2.5% , then remained at that level for the duration of the primary _ kepler _ mission .
this result is not sensitive to the size of the chosen aperture or the particular choice of reference stars .
we then compare this result to a similar analysis of other stars of similar brightness on the same detector , as well as stars with similar stellar properties , as listed in the kic , in the _ kepler _ field .
we find that 0.3% of stars on the same detector and 0.6% of stars with similar stellar properties exhibit a long - term trend consistent with that observed for kic 8462852 during the first three years of the _ kepler _ mission .
however , in no cases do we observe a flux decrement as extreme as the 2.5% dip observed in quarters 12 - 14 of the mission .
the total brightness change of kic 8462852 is also larger than that of any other star we have identified in the _ kepler _ images .
broadly speaking , the morphology of the light curve is generally consistent with the transit of a cloud of optically thick material orbiting the star .
such a dust cloud could be small enough to evade submillimeter detection in the analysis of @xcite , who place a limit of 7.7 m@xmath50 of material orbiting kic 8462852 .
the breakup of a small body or a recent collision that could produce a cloud of material could also plausibly produce a family of comets that transit the host star together as one group @xcite , explaining the light curve observed by @xcite .
however , in order to match the observed time of ingress and transit duration , some fine - tuning is required .
to explain the transit ingress timescale , the cloud would need to be at impossibly large distances from the star or be slowly increasing in surface density .
the flat bottom of the transit would then suggest a rapid transition into a region of uniform density in the cloud , which then continues to transit the star for at least the next year of the _ kepler _ mission .
moreover , such a model does not naturally account for the long - term dimming in the light curve observed in both dasch and the _ kepler _ ffi data , suggesting that this idea is , at best , incomplete .
there is no known or proposed stellar phenomenon that can fully explain all aspects of the observed light curve .
non - stellar explanations , such as circumstellar material , offer the best opportunity to provide an explanation for the observed light curve , but simple models are unable to match the light curve as observed over the last century .
we strongly encourage further refinements , alternative hypotheses , and new data in order to explain the full suite of observations of this very mysterious object .
we thank jason dittmann and jieun choi ( harvard ) for conversations about data analysis and figure design that improved the quality of this paper .
we thank john brewer ( yale / columbia ) for finding a typo in equation 1 in an earlier version of this paper .
we also thank rachael roettenbacher ( stockholm university ) and john johnson ( harvard ) for discussions of starspots , ryan foley ( uc santa cruz ) for initial conversations about photometry and the dasch results , george preston ( carnegie ) for insights into variable stars , and eugene chiang ( uc berkeley ) and jason wright ( penn state ) for ideas about other potentially similar systems .
funding for _ kepler _ , the tenth discovery mission , was provided by nasa s science mission directorate .
we are grateful to the entire _ kepler _ team , past and present .
their tireless efforts were essential to the tremendous success of the mission and the successes of _ k2 _ , present and future .
all of the data presented in this paper were obtained from the mikulski archive for space telescopes ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555
. support for mast for non - hst data is provided by the nasa office of space science via grant nnx09af08 g and by other grants and contracts . , j. , tang , s. , simcoe , r. , et al .
2009 , in astronomical society of the pacific conference series , vol .
410 , preserving astronomy s photographic legacy : current state and the future of north american astronomical plates , ed .
w. osborn & l. robbins , 101 | kic 8462852 is a superficially ordinary main sequence f star for which _ kepler _ detected an unusual series of brief dimming events .
we obtain accurate relative photometry of kic 8462852 from the _ kepler _ full frame images , finding that the brightness of kic 8462852 monotonically decreased over the four years it was observed by _
kepler_. over the first @xmath0 days , kic 8462852 faded approximately linearly at a rate of @xmath1 yr@xmath2 , for a total decline of 0.9% .
kic 8462852 then dimmed much more rapidly in the next @xmath3 days , with its flux dropping by more than 2% .
for the final @xmath3 days of _ kepler _ photometry the magnitude remained approximately constant , although the data are also consistent with the decline rate measured for the first 2.7 yr . of a sample of 193 nearby comparison stars and 355 stars with similar stellar parameters , none exhibit the rapid decline by @xmath4% or the cumulative fading by 3% of kic 8462852 .
moreover , of these comparison stars , only one changes brightness as quickly as the @xmath5 yr@xmath2 measured for kic 8462852 during the first three years of the _ kepler _ mission .
we examine whether the rapid decline could be caused by a cloud of transiting circumstellar material , finding while such a cloud could evade detection in sub - mm observations , the transit ingress and duration can not be explained by a simple cloud model .
moreover , this model can not account for the observed longer - term dimming .
no known or proposed stellar phenomena can fully explain all aspects of the observed light curve . |
over the past decade , hundreds of brown dwarfs ( bds ) have been identified in the solar neighborhood and in star - forming regions , yet there is no consensus on their origins .
one possibility is that they form like stars , as a result of the turbulent fragmentation and collapse of molecular cloud cores ( e.g. , padoan & nordlund 2004 ) .
another scenario , which has gained popularity in recent years , is that bds are stellar embryos ejected from multiple proto - stellar systems ( reipurth & clarke 2001 ; bate et al .
recent observations have attempted to distinguish among these scenarios , by comparing physical properties of brown dwarfs and very low mass stars ( which we refer to together as very low mass objects , or vlmos , here ) with those of solar - mass stars .
numerous studies have shown that many _
young _ vlmos exhibit infrared excesses ( e.g. , muench et al .
2001 ; natta et al . 2002 ; jayawardhana et al .
2003a ; scholz et al .
2007 ) and spectroscopic signatures of accretion ( e.g. , jayawardhana et al .
2002 , 2003b ; natta et al . 2004
; mohanty et al .
2005 ) , indicative of disks . but
both these diagnostics are only sensitive to the innermost portions of the disk and therefore can not test whether bd disks are truncated , as expected in the ejection scenario .
recently , by modeling mid - infrared and millimeter emission , scholz et al .
( 2006 ) found that @xmath525% of taurus bds in their sample have disks with radii @xmath510 au , contrary to predictions of certain ejection simulations , but some truncated disks may be hidden among their non - detections at 1.3 mm .
the spatial distribution and kinematics of bds , in comparison with stars , do not provide definitive tests either ; ejection models predict velocities comparable to the velocity dispersions of star - forming associations and clusters , i.e. , 1 - 2 km / s ( moraux & clarke 2005 ) . in this context
, the multiplicity properties of vlmos such as frequency , separations , and mass ratios could be among the most important diagnostics of their origin .
in particular , ejection models predict a very low rate of binaries among bds ( @xmath65 - 8% ) and does not favor the formation of many wide ( @xmath520 au ) binary systems ( e.g. , bate et al .
direct imaging surveys of ( old ) field vlmos have yielded a binary frequency of @xmath115% for ( apparent ) separations @xmath71 au , with maximum separations of @xmath820 au ( e.g. , bouy et al .
2003 ; burgasser et al .
a search for spectroscopic binaries among the field vlmos add another @xmath111% ( basri & reiners 2006 ) .
open clusters older than @xmath1100 myr also appear to lack wide binary brown dwarfs ( e.g. , martn et al .
2003 ; bouy et al .
2006a ) . at face value
, these findings seem to support the ejection scenario though typical binary separation may decrease at lower masses even in the turbulent fragmentation model .
however , perhaps somewhat surprisingly , several wide binary vlm systems have been discovered within the past few years . among them
are : 2mass j1101 - 7732 ( @xmath9240 au ) in the @xmath12-myr - old chamaeleon i star - forming region ( luhman 2004 ) , 2mass j1207 - 3932 ( @xmath940 au ) in the @xmath18-myr - old tw hydrae association ( chauvin et al .
2004 ; mohanty et al .
2007 ) , denis - p j161833 - 251750 and usco-160028.4 ( @xmath9140 and 120 au , respectively ) in the @xmath15-myr - old upper scorpius association ( luhman 2005 ; bouy et al .
2006b ) , oph 162225 - 240515 ( @xmath9240 au ) in the @xmath11-myr - old ophiuchus region ( jayawardhana & ivanov 2006 ) , and denis - j055146.0 - 443412.2 ( @xmath7200 au ) in the field ( billres et al . 2005 ) .
2mass j1207 - 3932 and oph 1622 - 2405 are especially intriguing because the secondaries have masses approaching the planetary regime .
if wide binaries , such as these six , are common among vlmos , that would present a critical challenge to ejection models .
recent high angular resolution surveys in taurus and upper scorpious by kraus et al .
( 2005 , 2006 ) and konopacky et al .
( 2007 ) have turned up 4 vlm binary systems ( all in the konopacky et al .
2007 sample ) that would be considered wide by the ejection model standards . here
we report on an additional effort to constrain the frequency of wide vlm systems in other star forming regions .
in particular , we describe a near - infrared adaptive optics imaging survey of 28 vlmos in chamaeleon i designed to investigate whether multiplicity in the vlm regime depends on age .
given its proximity ( 160pc ; whittet et al .
1997 ) and youth ( @xmath12myr ; luhman 2004 ) , the chamaeleon i star - forming region is well suited for investigating multiplicity of young vlmos .
we observed 28 targets with spectral types of m5.25m8 from the recent compilation by luhman ( 2004 ) ; their membership in cha i is supported by several diagnostics including proper motion , radial velocity and position on the h - r diagram .
six additional ( random ) vlm members reported in the census ( 3 m5.25 , 1 m5.5 , 1 m5.75 , 1 m6 ) were not observed due to worsening atmospheric conditions . according to baraffe et al .
( 1998 , 2003 ) models , our targets span the mass range of @xmath10 ( table 1 ) .
we obtained near - infrared @xmath11- and @xmath12-band images of 26 of these targets with the naos - conica ( naco , for short ) instrument on the 8.2 m yepun unit of the european southern observatory s very large telescope on paranal , chile over three consecutive nights starting on march 24 , 2006 .
we used the high - resolution lens of naco , with a pixel scale of 13.26 mas / pix and a field - of - view of @xmath13 .
for all observations , we utilized the infrared wavefront sensor ( wfs ) mode n90c10 , which directs 90% of the light to the wfs and the rest to conica .
targets themselves served as natural guide stars ( ngss ) for the wfs . for each target , six 20 second exposures were combined to produce the final image .
two other cha i targets , chah@xmath14 and eso559 , were observed in service mode between 2005 may
september using the same instrument set up , along with c-41 in chamaeleon ii ( see barrado y navascus & jayawardhana 2004 ) , gy 5 in @xmath15 oph ( greene & young 1992 ) and 2massw j1207334 - 393254 in the tw hydrae association ( gizis 2002 ) .
the latter is known to harbor a planetary mass companion at a separation of @xmath140 au ( chauvin et al .
2004 ; mohanty et al .
2007 ) , and thus served as a cross - check on our sensitivity .
these five targets were observed only in the @xmath16-band , each with a total integration time of 18 minutes .
we inspected the final images , reduced in a standard manner using the naco pipeline software , visually to look for possible companions . in order to determine our ability to detect companions , we also derived contrast sensitivity curves through statistical analysis of the noise in annuli surrounding the central object
. then we calculated the 5-@xmath17 detection limit as a function of separation from the primary for each of the targets .
more details on the algorithm for deriving these sensitivities are given in brandeker et al .
we checked the accuracy of the procedure by artificially inserting faint companions into the images at various separations and trying to recover them ; the results were consistent .
we converted the contrast sensitivity limits into mass detection limits given the spectral type of the primaries , luhman et al .
( 2003 ) temperature conversion scale and baraffe et al .
( 1998 , 2003 ) models for the appropriate age : 2 myr for cha i , 1 myr for @xmath15 oph and cha ii , and 8 myr for tw hydrae . to search for companions at smaller separations
, we carried out point spread function ( psf ) subtraction for the cha i targets from the primary naco run . to construct a template psf
, we averaged the psfs of our brightest targets . by varying the combinations of objects used for the template psf and comparing the results for consistency
, we made sure not to include a binary in the template psf .
none of our objects shows a clear signature of a companion after psf subtraction . for a more quantitative check
, we used an approach similar to the procedure suggested by bouy et al .
( 2006a ) : after psf subtraction , we measured the residual scatter and looked for objects which show excessive noise in the psf area .
again , none of our objects shows significantly increased noise , defined as 3@xmath17 excess above the average . to probe the adequacy of this test
, we inserted artificial companions for selected objects at separations of 10au and 6au with roughly equal brightness .
companions at 10au were reliably detected , by visually inspecting the image before and after psf subtraction .
they also cause clearly enhanced scatter after psf subtraction . thus , we are confident that our companion search is complete down to 10au for roughly equal mass systems . at 6au , we can still recover artificial companions around the brightest of our targets , but for the fainter ones both the visual inspection and the noise analysis fail .
we conclude that our companion search is probably not complete for separations @xmath18au . for the brightest @xmath19 objects , however , we can safely exclude the existence of roughly equal mass companions down to separations of 6au .
we find three resolved binaries in the cha i sample : chah@xmath3 2 , chxr 15 and hn 13 ( fig . 1 and table 2 ) .
the first was suspected to be a binary by neuhuser et al .
( 2002 ) based on its elongated psf in _ hubble space telescope _ images .
we are also able to recover the previously known companion to 2m1207 in our images .
we do not find companions to any other targets .
interestingly , chah@xmath3 2 , chxr 15 and hn 13 are all at the high mass end of our target sample : all have masses @xmath20 , clearly above the sub - stellar boundary .
two of the binaries , chah@xmath3 2 and hn 13 , have mass ratios close to unity ( @xmath210.9 ) and are slighthly wider ( @xmath22 ) than most binaries found in the field .
chxr 15 stands out more : it has a lower mass ratio ( @xmath23 ) and a wider separation ( @xmath24 ) .
2 shows our detection limits as a function of companion mass ( according to baraffe et al .
models ) and separation ( assuming a distance of 160 pc ) .
the survey is able to detect binaries with mass ratios ( @xmath25 ) as low as 0.2 for all but one target at 50 au , and planetary mass companions at larger separations .
our results suggest that wide vlm systems with ultra low mass companions , such as oph 1622 - 2405 ( jayawardhana & ivanov 2006 ) , are rare .
we return to this point in the discussion . for comparison , we also plot the binaries found by fischer & marcy ( 1992 ) among field stars with primary masses of 0.1 - 0.6@xmath26 and by close et al .
( 2003 ) , burgasser et al .
( 2003 ) and siegler et al .
( 2005 ) among field vlm objects with @xmath27@xmath28 .
our survey would have detected nearly all binaries with @xmath520 au and @xmath29 0.6 in these comparison samples .
for our cha i sample , with primary masses between @xmath10.02 - 0.2@xmath26 , we find a binary frequency of @xmath4 in the regime where our survey is sensitive to companions ( @xmath30 ) .
our findings in cha i are similar to what kraus et al .
( 2005 , 2006 ) have reported for vlmos in taurus and upper scorpius star - forming regions . a recent survey of taurus by konopacky et al .
( 2007 ) targeting mainly vlm stars has revealed a larger multiplicity fraction of wider systems with lower mass ratios , although the binary frequency is still in statistical agreement with our results .
in fact , the four surveys target primaries in the same mass range and have comparable detection limits for companions . ] .
thus , we can combine the results of all four surveys ( 12 vlm targets in upper scorpius , 32 in taurus and 28 in chamaeleon i ) to arrive at a more statistically significant sample of 72 young vlmos with an overall resolved binary frequency of @xmath31 .
if we just consider t - associations ( taurus and chamaeleon i ) , then the multiplicity fraction is @xmath32 . in both cases we use @xmath33 and @xmath34 as our sensitivity limit .
we can now compare the binary properties of young vlmos with other populations of interest , such as field vlmos and young stars , to explore possible variations with mass and/or age .
we used a fisher `` exact '' statistical test for comparing samples of binomial distributions , as detailed in appendix b of brandeker et al .
the results are given in table 3 , where `` likelihood '' refers to the probability that the multiplicity frequencies of the two samples are the same . in order to explore the dependence of the binary frequency on mass
we statistically compare several samples spanning different mass bins . as shown in table 3 and fig .
4 , we constrain the @xmath25 and separation range as appropriate to ensure that the two samples being compared have the same sensitivity limits for companions . we ensure that samples that are compared are of similar age by only comparing binaries from young associations ( cha i , taurus and upper scorpius ) .
in addition we also address the potential concern about the environmental effects by performing every comparison with and without the 12 targets in upper scorpius as it is the only higher density association that contributes binaries to our combined sample .
we find that our conclusions are not affected by inclusion of these additional binaries .
we start by dividing the young vlmos themselves into two groups , following kraus et al .
( 2006 ) : one with primary masses less than 0.07 @xmath26 and the other with primary masses greater than 0.07 @xmath26 .
all eight resolved vlmo binaries within the sensitivity limits ( fig .
4a ) are in the higher mass bin .
the likelihood of the two mass bins having the same binary frequnecy if only t - associations are compared is low ( @xmath35 ) , but not insignificant .
if we increase the sample size by adding the upper scorpius members , 6 to the lower and 6 to the higher mass bin , the likelihood that the two mass bins have the same binary distribution becomes even lower ( @xmath36 ) .
in addition , we also note that 4 systems that are outside of the sensitivity range fall in the higher mass bin and thus further reinforce our point
. we can also compare the young vlmo sample with young stars ( @xmath10.57 - 1.40 @xmath26 ) in taurus and cha i surveyed by ghez et al .
( 1993 , 1997 ) in the sensitivity regime outlined in fig .
the result ( likelihood of @xmath37 and @xmath36 if we include or exlude the upper scorpius sample respectively ; table 3 ) bolsters the case for a decreasing wide binary frequency with mass . what could account for this observed trend ?
one possibility is that less massive , and therefore fainter , primaries will have even fainter companions that might not be detectable by surveys .
however , all of the samples presented here are sensitive to mass ratios of @xmath38 at separations greater than @xmath39 and even lower mass ratios at wider separations ( fig 2 . and fig 4 . ) for the young vlmo sample , allowing us to rule out such a bias .
another possibility is that binary separations ( i.e. , system size ) decreases with mass below our detection limits . a radial velocity survey of cha i vlmos finds only one candidate companion ( joergens 2006 ) at @xmath400.1 au among 10 targets while basri & reiners ( 2006 ) find a spectroscopic binary fraction of @xmath111% in the 0 - 6 au separation range among field vlmos .
imaging surveys of field vlmos ( eg .
, close et al .
2003 , siegler et al .
2005 ) , which have typical sensitivities of @xmath41 at @xmath42 , also do not reveal a significant number of companions closer to the primaries .
thus , it appears unlikely that a decrease in separations is the primary reason for the observed drop in multiplicity in the vlm regime .
instead , we conclude that the binary frequency most likely decreases with mass .
we investigate whether binarity in the vlm regime depends on age by comparing the young vlmo sample with the late m and early l field dwarfs surveyed by close et al .
they found three nearly equal mass @xmath43 pairs wider than @xmath17.5 au among 85 targets in the @xmath10.05 - 0.1 @xmath26 mass range , resulting in a binary fraction of @xmath44 , which is in good agreement with that for young vlmos suggesting that binarity in the vlm regime does not depend strongly on age between a few myr and several gyr .
one concern is that environments from which these two samples are drawn may be different .
the young vlmo sample comes from low density t - associations ( and ob - associations if we choose to include upper scorpius ) , while most of the field objects probably formed in higher density clusters .
thus one can argue that the multiplicity fraction we see today in the surveyed sfrs might decrease over time below that seen in the field .
it is unlikely , however , that this will happen because the field vlmos have survived more dynamically active environments .
we also examine the stability of vlm binaries by calculating their binding energies in comparison to those of other binary populations .
3 shows that even though there is significant overlap of young and field vlm binaries on the binding energy vs. separation plot , young unbound associations also have a significant population of wider binaries .
as this is the population that might be disrupted by dynamical encounters , exploration of stability of such systems is important . regarding this problem
, we note that a number of field star systems ( those in the boxed area of the figure ) are more fragile than the young vlm binaries , even though the former have evolved dynamically already .
this is all the more surprising if one considers the fact that these field systems were most likely born in higher density regions where dynamical evolution would have had a more profound effect .
therefore , it is unlikely that the `` harder '' vlm binaries , would be disrupted , again supporting the conclusion that age is not a major factor affecting binarity in the vlm regime , at least when one considers ob and t associationsau ) field vlm systems have been discovered ( billres et al . 2005 ; caballero et al . 2007 ;
artigau et al .
these are not included in our analysis as they are not part of a defined sample .
nevertheless , their existence shows that at least some very wide vlm binaries could survive their birth environment . ]
weakly bound binaries can be disrupted by gravitational perturbations from both stellar encounters and giant molecular clouds , as discussed by weinberg et al .
burgasser et al . ( 2003 ) applied the formalism of weinberg et al .
( 1987 ) to the bd regime , and found that a pair of 0.05@xmath45 bds separated by 10au would be completely impervious to disruption by stellar encounters in the field .
scaling the result of burgasser et al .
( 2003 ) to the most weakly bound system in our combined young vlmo sample , with a total mass of @xmath46 and a separation of @xmath47au , we get expected disruption timescales from field stars that are on the order of 600gyr , more than 10@xmath48 times older than the age of the systems .
these estimates are based on field star densities of @xmath49pc@xmath50 ( bahcall & soneira 1980 ) , which is a critical parameter , as the disruption timescale @xmath51 . but even assuming a stellar density as high as @xmath52pc@xmath50 , corresponding to the outer regions of the orion nebula cluster ( khler et al .
2006 ) , the disruption timescale would be 150myr , two orders of magnitude longer than the present age of the system . this estimate suggests that encounters with other stars do not disrupt even the most weakly bound systems in our sample , either in their birth environments ( i.e. , in ob or t associations ) or later as the stellar densities drop to match that in the field .
since perturbations from giant molecular clouds are insignificant in comparison to stellar encounters ( burgasser et al . 2003 ) , the vlm binaries in our sample are likely to survive almost indefinitely . in recent years
, several vlm systems that appear to diverge from established trends have been reported .
in particular very wide systems such as oph162225 - 240515 with ultra low mass companions stand out from field vlm systems whose separation tends to decrease with mass . because the presence of such systems imposes additional constraints on formation mechanisms , it is important to determine their frequency .
5 shows that our survey is sensitive to nearly planetary mass companions ( down to 9m@xmath53 ) around 23 targets in the cha i sample at separations greater than @xmath54 .
in addition , such systems could be detected around all 34 targets from the kraus et al .
( 2005 , 2006 ) surveys of upper scorpius and taurus . yet , no ultra low mass companions at wide separations are detected in these three surveys , giving a frequency of no more than 2% and thus confirming their rarity .
since the typical primary mass of the combined vlmo sample is 110m@xmath53 , and the imaging surveys are sensitive to @xmath55 companions ( i.e. , secondaries with 915m@xmath53 ) at @xmath56 , we can also compare our results to surveys for sub - stellar companions to solar analogs .
for example , metchev & hillenbrand ( 2005 ) targeted 100 f5-k5 primaries to look for brown dwarf companions .
based on a single detection , they derive a sub - stellar companion freqency of at least @xmath57 , and possibly on the order of a few percent .
this result is in statistical agreement with the frequency of @xmath55 systems around vlm primaries in our combined young sample .
therefore , even though vlm systems with planetary mass companions are rare , the frequency of @xmath55 binaries in the vlm regime is no different from that for solar - type stars .
our survey of vlmos in chamaeleon i lends additional support to the established trend for a lower binary frequency with decreasing primary mass . by comparing the multiplicity properties of young vlmos with those of field vlmos and stars , we conclude that there is likely no evolution of multiplicity after the first fewmyr for objects formed in low density regions . while our survey of cha i and the surveys of kraus et al .
( 2005 , 2006 ) and konopacky et al .
( 2007 ) show that wide vlm systems with ultra - low - mass companions thought to be a critical challange to the ejection formation mechanism are infrequent , they also reveal a significant fraction of systems considered wide by the ejection scenario ( @xmath58 ) .
seven out of thirteen reported systems have separations wider than @xmath59 , with three systems separated by more than @xmath24 .
bate et al .
( 2005 ) have recently reported that their simulations can produce binaries wider than the initial @xmath59 limit , albeit via a different mechanism than the one responsible for the production of tighter pairs ( i.e. , simultaneous ejection and capture of previously unassociated objects vs. ejection of coupled tight binaries ) . however , since their hydrodynamic simulations are not followed up with n - body calculations , it is difficult to make a direct comparison ; many of the reported simulated binaries are still accreting and likely unstable and the statistics are still somewhat limited ; there may also be a spread in the ages of objects in a given sfr .
in addition , most simulations model higher density regions , more akin to the orion nebula cluster rather than the diffuse t associations that we and other observers have surveyed mostly .
nevertheless , a successful model for star formation has to account for observed binary properties in the vlm domain , including the existence of some wide pairs .
this paper is based on observations collected at the european southern observatory , chile under programs 075.c-0042 and 076.c-0579 .
we thank the eso staff for carrying out some observations in service mode and for assistance during a visitor run .
we are grateful to the anonymous referee for a careful review .
this research was supported by an nserc grant and an ontario early researcher award to rj .
ma was funded in part by an ontario graduate scholarship and an nserc canadian graduate scholarship .
df received support from the cosmo grid project , funded by the program for research in third level intitutions under the national development plan and with assistance from the european regional development fund .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation . , chxr 15 and hn 13 with separations of @xmath60 , @xmath61 and @xmath62 respectively .
all images are taken with @xmath63 filter .
n is up and e is to the left in all three images.,title="fig : " ] , chxr 15 and hn 13 with separations of @xmath60 , @xmath61 and @xmath62 respectively .
all images are taken with @xmath63 filter .
n is up and e is to the left in all three images.,title="fig : " ] , chxr 15 and hn 13 with separations of @xmath60 , @xmath61 and @xmath62 respectively .
all images are taken with @xmath63 filter .
n is up and e is to the left in all three images.,title="fig : " ] .
four young vlmos that are wider than the given limit come from the recent speckle survey of taurus - auriga by konopacky et al .
( 2007 ) ; if these four are unresolved triples i.e .
, one component is actually an unresolved equal - brightness binary then two of them would still be considered wide by the burgasser et al .
( 2003 ) threshold , while if they are unresolved quadruples , all four would satisfy the derived limit . ]
lccccccc chah@xmath311 & 11 08 29.27 & -77 39 19.8 & 13.54 & m7.25 & 60 & 2838 + chah@xmath310 & 11 08 24.04 & -77 39 30.0 & 13.24 & m6.25 & 92 & 2962 + iso138 & 11 08 18.50 & -77 30 40.8 & 13.04 & m6.5 & 83 & 2935 + chsm17173 & 11 10 22.27 & -76 25 13.8 & 12.45 & m8 & 40 & 2710 + chah@xmath37 & 11 07 37.76 & -77 35 30.8 & 12.42 & m7.75 & 45 & 2752 + chah@xmath31 & 11 07 16.69 & -77 35 53.3 & 12.17 & m7.75 & 45 & 2752 + chsm10862 & 11 07 46.56 & -76 15 17.5 & 12.33 &
m5.75 & 117 & 3024 + iso147 & 11 08 26.51 & -77 15 55.1 & 12.35 & m5.75 & 117 & 3024 + chsm1982 & 11 04 10.60 & -76 12 49.0 & 12.12 & m6 & 102 & 2990 + iso252 & 11 10 41.41 & -77 20 48.0 & 12.27 & m6 & 102 & 2990 + chah@xmath312 & 11 06 38.00 & -77 43 09.1 & 11.81 &
m6.5 & 83 & 2935 + iso220 & 11 09 53.37 & -77 28 36.6 & 12.23 &
m5.75 & 117 & 3024 + iso28 & 11 03 41.87 & -77 26 52.0 & 11.69 & m5.5 & 134 & 3058 + eso559 & 11 06 26.3 & -76 33 42 & 11.49 & m6 & 102 & 2990 + chah@xmath39 & 11 07 18.61 & -77 32 51.7 & 11.80 & m5.5 & 134 & 3058 + iso165 & 11 08 54.97 & -76 32 41.1 & 11.44 & m5.5 & 134 & 3058 + chah@xmath32 & 11 07 42.45 & -77 33 59.4 & 10.68 & m5.25 & 153 & 3091 + chah@xmath34 & 11 08 18.96 & -77 39 17.0 & 11.02 & m5.5 & 134 & 3058 + chah@xmath33 & 11 07 52.26 & -77 36 57.0 & 11.10 & m5.5 & 134 & 3058 + chah@xmath36 & 11 08 39.52 & -77 34 16.7 & 11.04 & m5.75 & 117 & 3024 + hn12w & 11 10 28.52 & -77 16 59.6 & 10.78 & m5.5 & 134 & 3058 + chxr84 & 11 12 03.27 & -76 37 03.4 & 10.78 & m5.5 & 134 & 3058 + hn13 & 11 10 55.97 & -76 45 32.6 & 9.91 &
m5.75 & 117 & 3024 + chah@xmath313 & 11 08 17.03 & -77 44 11.8 & 10.67 & m5.5 & 134 & 3058 + chah@xmath35 & 11 08 24.11 & -77 41 47.4 & 10.71 & m5.5 & 134 & 3058 + iso235 & 11 10 07.85 & -77 27 48.1 & 11.34 & m5.5 & 134 & 3058 + chxr78c & 11 08 54.22 & -77 32 11.6 & 11.22 & m5.25 & 153 & 3091 + chxr15 & 11 05 43.00 & -77 26 51.8 & 10.23 & m5.25 & 153 & 3091 + chaiic41 & 12 59 09.8 & -76 51 04 & 11.4 & m5.5 & 109 & 3058 + gy 5 & 16 26 21.4 & -24 25 59 & 10.92 & m6 & 81 & 2990 + 2mass1207 - 3932 & 12 07 33.4 & -39 32 54 & 11.95 & m8 & 42 & 2710 + ccccccc chah@xmath32 & @xmath64 & @xmath65 & ks & @xmath66 & @xmath67 & 0.94 + hn13 & @xmath68 & @xmath69 & ks & @xmath70 & @xmath71 & 0.95 + hn13 & @xmath68 & @xmath72 & h & @xmath73 & @xmath74 & + chxr15 & @xmath75 & @xmath76 & ks & @xmath77 & @xmath78 & 0.64 + chxr15 & @xmath79 & @xmath76 & h & @xmath80 & @xmath81 & + | characterizing multiplicity in the very low mass ( vlm ) domain is a topic of much current interest and fundamental importance .
here we report on a near - infrared adaptive optics imaging survey of 31 young brown dwarfs and vlm stars , 28 of which are in the chamaeleon i star - forming region , using the eso very large telescope .
our survey is sensitive enough to detect equal mass binaries down to separations of 0.040.07@xmath0 ( @xmath1610 au at 160 pc ) and , typically , companions with mass ratios ( @xmath2 ) as low as 0.2 outside of 0.2@xmath0 ( @xmath130 au ) .
we resolve the suspected 0.16@xmath0 ( @xmath126 au ) binary chah@xmath3 2 and present two new binaries , hn 13 and chxr 15 , with separations of 0.13@xmath0 ( @xmath1 20 au ) and 0.30@xmath0 ( @xmath1 50 au ) respectively ; the latter system is one of the widest vlm systems discovered to date .
we do not find companions around the majority of our targets giving an overall binary frequency of @xmath4 , thus confirming the trend for a lower binary frequency with decreasing mass . by combining our work with previous surveys of vlm objects ( vlmos ) in other star forming regions ,
we arrive at the largest sample of young vlmos ( 72 ) with high angular resolution imaging to date .
its multiplicity fraction is in statistical agreement with that for vlmos in the field .
in addition we note that many field stellar binaries with lower binding energies and/or wider cross sections have survived dynamical evolution and that statistical models suggest tidal disruption by passing stars is unlikely to affect the binary properties of our systems .
thus , we argue that there is no significant evolution of multiplicity with age among brown dwarfs and vlm stars in ob and t associations between a few myr to several gyr . instead
, the observations to date suggest that vlm objects are either less likely to be born in fragile multiple systems than solar mass stars or such systems are disrupted very early ( within the first couple of myr ) . |
today it is usually accepted that millisecond pulsars ( msps ) are old neutron stars ( nss ) recycled in close binary systems ( cbss ) @xcite . in this process , the companion ( a low - mass star ) transfers mass and angular momentum to the ns during a roche lobe overflow ( rlof ) episode , increasing the mass and rotation rate of the ns . during the mass transfer process
the system is a copious source of x - rays , generally ( but not only ) detectable as a low - mass x - ray binary ( lmxb ) . for this type of systems ,
the standard models of cbss predict a long and stable episode of mass transfer , as a consequence of the evolution of the donor star , with a small number of rlofs due to thermonuclear flashes ( see , e.g. , @xcite ; ; @xcite ; @xcite ) . as a final result
, the formation of a msp ( with a spin period @xmath4 ms ) is expected .
the most common companion is a low - mass helium white dwarf ( hewd ) , with characteristic mass values of @xmath5 and @xmath6 , respectively . ] .
these objects are compact remnants of cbss with @xmath7 day ( where @xmath8 is the orbital period of the system ) .
other physical phenomena that play a relevant role in the formation of msps are magnetic braking and gravitational radiation ; see , e.g. , for a review .
standard models support the idea that msps descend from lmxbs , however the existence of many isolated msps represent a serious difficulty to the model .
nevertheless , @xcite discovered a msp in an eclipsing binary system ( psr 1957 + 20 ) .
it has a spin period of 1.6 ms , in orbit around a low - mass companion of only @xmath9 .
due to the strong eclipses and the small mass of the companion , it was proposed that pulsar radiation is ablating the companion @xcite .
this is the representing object of a class of eclipsing binary systems , composed by a msp and a very low - mass companion ( @xmath10 ) , in tight orbits ( @xmath11 d ) , generally called `` black widows '' .
the evaporation of the companion by the pulsar emission in cbss could be a process that allow recycled nss to unbind their companions and finally appear as isolated msps .
it has been recently shown ( @xcite ; hereafter bdvh ) that evolutionary trajectories leading to the evaporating ( black widows ) _ locus _ cross a region in the @xmath12 plane in which accretion - driven irradiation causes an evolution markedly different from the standard one leading to lmxbs .
in fact , there is a wide zone populated by pulsar companions that standard models of close binary evolution can not reproduce .
it is precisely this stage that will be studied in this paper .
sometime ago it has been proposed that the pressure due to radio pulsar irradiation may preclude accretion onto the ns .
this phenomenon is usually referred to as `` radio ejection '' ( @xcite ; @xcite ) .
if radio ejection were fully effective in inhibiting accretion , the results described in bdvh would be no longer valid .
also , evaporation may be relevant at stages earlier than those corresponding to black widows @xcite .
perhaps the most important non - standard effect in the early stages of cbss is irradiation feedback .
the accretion onto the ns releases x - rays , which are partially absorbed by the companion and produce a partial blockage of the energy that emerges from the donor star interior . as a non - trivial result ,
the mass transfer episode becomes _ cyclic _ ( see ; @xcite ) .
calculations ( see bhdv ) predict a back - and - forth switch of the system between lmxb and binary radio pulsar states .
irradiation swells the companion and the latter is filling , or quasi filling its roche lobe .
the inclusion of this new ingredient allowed us to perform more realistic calculations , and to reconcile theory with observations .
direct proof of the theoretical scenario for recycling old nss in msps was found for first time in sax 1808.4 - 3658 .
@xcite reported coherent millisecond x - ray pulsations in the persistent flux of this cbs .
the source will probably become a msp when accretion turns off .
the parameters ( @xmath13 h , eccentricity @xmath14 and pulsar mass function @xmath15 ; @xcite ) imply a companion mass of @xmath16 , if the ns mass is in the range @xmath17 . the orbital period ( @xmath11 d ) and the mass of the companion ( @xmath18 ) place this system in the redback region @xcite in the plane @xmath12 .
@xcite have proposed that this object is undergoing a switch between accretion and msp stages dominated by radio ejection .
a handful of systems placed in the redback region have shown clear signs of disc states and extended radius of the donor ( although not always simultaneously ) .
for example , the first j102347.67 + 003841.2 source with @xmath19 h @xcite is coincident with the msp psr j1023 + 0038 @xcite .
the 2001 observations showed evidence of mass outflow , but since 2002 a quiescent state followed .
later it has been found that for reasonable ns masses , the companion mass is limited to @xmath20 and likely fills its roche lobe because of the presence of ionized material .
the recent ( mid-2013 ) enhancement of uv and x - ray emission ( @xcite ; @xcite ) , together with the interruption of the radio pulsations , has been interpreted as evidence that the lmxb phase returned @xcite . since the system is far from any globular cluster and the orbit is almost circular , it is reasonable to assume that the pulsar has been recycled by the companion and alternates between two states with the companion filling the roche lobe .
this system may be seen as redback in the msp phase and turn to lmxb phenomenology when accretion ( re)starts .
redback seemed to be rare , but since 2009 many of them were discovered in the galactic plane @xcite , as well as in globular clusters ( see pulsars in globular clusters , freire s homepage ) . currently , there are @xmath21 redbacks known in the galactic field , and @xmath22 in globular clusters .
perhaps the best example of this class of objects , changing its state between accreting x - ray msp and radio pulsars state is precisely a system located in the globular cluster m28 , psr j1824 - 2452i ( with @xmath23 ms and @xmath24 h ) .
the system was detected switching from an accretion phase to a radio pulsar phase in the last decade @xcite .
the change between rotation and accretion powered states in a few - days to few - months mark the existence of a phenomenon that occurs on a timescale much shorter that the usual long scale of the order of @xmath0 gigayears in these type of cbss , or the pulsed mass transfer scale recently found in bdvh .
further systems belonging to the redback family and displaying evidence for an extended radius of the companion , even in quiescence , include psr j1723 - 2837 .
this is an eclipsing , @xmath25 ms binary radio pulsar in an almost circular orbit with @xmath26 h discovered by @xcite .
the eclipses suggested that the pulsar companion is a non - degenerate , extended star .
this has been confirmed by @xcite , who identify the companion in the ir , optical , and uv bands .
the spectral analysis put the companion between g5 and k0 spectral types with an effective temperature of 5000 - 6000 k and a surface gravity reasonably consistent with a main sequence star .
these authors derived a mass ratio of @xmath27 , giving a companion mass range of @xmath28 , and an orbital inclination angle between @xmath29 , assuming a pulsar mass in the range of @xmath30 .
the derived radius of the companion is larger than expected , likely close to filling its roche lobe , i.e. , a _
quasi_-rlof object .
an example of a quasi - rlof object is psr j1740 - 5340 .
it is a binary msp ( @xmath31 3.65 ms ) that has @xmath32 32.5 h , which falls outside the usually considered range of values corresponding to redbacks . for this object
@xcite proposed a possible binary progenitor that accounts for the observational data of the companion and considers that it is on a radio ejection stage .
xss j12270 - 4859 , on the other hand , experienced a transition from lmxb state to radio msp ( @xcite ; @xcite ) .
this report confirm the idea that this pair belongs to the group formed by a low - mass quasi - rlof object and a rotation - powered msp , a redback system in the current state of xss j12270 - 4859 .
finally , we quote the cases of psr j1023 + 0038 and xss j12270 - 4859 , which experienced transitions in opposite directions . in the transition of psr j1023 + 0038 from msp to lmxb , it was detected an increase in the flux of @xmath33-rays @xcite .
xss j12270 - 4859 showed a decrease of the @xmath33-ray brightness ( tam et al 2013 ) , although less intense than the growth in the case of psr j1023 + 0038 .
the transition of psr j1023 + 0038 occurs in @xmath34 weeks , while the transition of xss j12270 - 4859 was in @xmath35 weeks .
the process of back - and - forth transitions seem to be fast .
besides the eclipses and a non - degenerate low - mass companion , redbacks seems to have another feature that identified them : the high filling factor of the roche lobe by the donor star component ( @xmath36 in many cases , see @xcite and referenced therein ) .
all the phenomenology described above can not be solely related to the evolution of the donor star , because the latter has a typical timescale far greater .
it seems more natural to interpret the changes between lmxb and binary radio pulsar as due to instabilities of the accretion disc surrounding the ns .
but for these instabilities to occur the companion star must act as a donor .
if we observe a binary msp showing redback features , it would be hard to distinguish if it is visible because of disc instabilities or if it is detached due to donor evolution .
a population synthesis should be helpful to discern between these possible situations , because the time fraction of the cycle in which mass transfer occurs is small ( see bdvh ) and this is the only situation in which the pulsar should be detectable due to accretion disc instabilities ( otherwise there is no accretion ) .
if the scenario we are suggesting here is at least qualitatively correct , we should expect that redbacks should be more abundant than lmxbs with orbital periods and companion masses in the same range .
unfortunately , it seems that the number of already detected systems of these kinds prevent us to perform a solid statistical inference . in any case
, the very existence of quasi - rlof stars should be related to the irradiation feedback phenomenon .
it is the aim of this paper to discuss the quasi - rlof state and also to show that the evolutionary models presented in bdvh provide a natural account of the characteristics of the redback system psr j1723 - 2837 .
the remainder of this paper is organized as follows : in section [ sec : code ] we briefly describe the ingredients of our stellar code . in section [ sec : qrlof ] we present the quasi - rlof state in the evolution of cbss containing radio pulsars . in section
[ sec : aplicando ] we apply this novel evolutionary state to the case of the available observations of the psr j1723 - 2837 redback system . in section [ sec : disc ] we discuss the meaning of the results presented in this paper and finally , in section [ sec : concl ] we give some concluding remarks .
the results to be described below have been found employing our cbs evolutionary code @xcite updated by the inclusion of irradiation feedback and evaporation . during a rlof episode
the code solves for the donor star structure , the instantaneous mass transfer rate @xmath37 , both masses and orbital semi axis _
implicitly_. this method allows for the computation of the mass transfer cycles quoted in section [ sec : introd ] to be presented below in detail . in detached conditions , the code employs the standard henyey technique .
we assumed that the ns accretes a fraction @xmath38 of the transferred matter up to the eddington critical rate @xmath39 , while very low accretion rates ( @xmath40 ) are inhibited by the propeller mechanism @xcite .
angular momentum sinks due to gravitational radiation , magnetic braking and mass loss from the system have been treated as in @xcite .
irradiation feedback has been included following who replace the usual relation between luminosity , radius and effective temperature @xmath41 by @xmath42 where @xmath43 is the stefan - boltzmann constant , @xmath44 is the effective temperature of the non - illuminated part of the star , @xmath45 and @xmath46 , where @xmath47 is the incident irradiating flux .
@xmath48 is computed assuming that at the deep enough layers , perturbations due to irradiation must vanish .
we assume that the ns acts as an isotropic _ point source _ , releasing an accretion luminosity @xmath49 ( @xmath50 , and @xmath51 are the radius and accretion rate of the ns , respectively ) .
the irradiation flux incident onto the donor star , effectively participating in the feedback phenomenon is @xmath52 , where @xmath53 is the fraction of the flux irradiated towards the donor that drives the phenomenon , and @xmath54 is the orbital radius . because of the assumption of isotropy and the way it has been defined , we have @xmath55 . because in the models to be presented below we assume that evaporation do not play a role we shall not detail the way we handled this phenomenon in our code .
for further details we refer the reader to bdvh an references therein .
the occurrence of the quasi - rlof state is easy to understand by performing a close examination of the evolution of the donor mass when irradiation feedback is considered . before the onset of a rlof , we assume there is no accretion onto the ns ; so , there is no irradiation incident onto the donor star . just after the beginning of the rlof ,
irradiation feedback forces the donor star to have an instantaneous mass transfer rate @xmath37 that overflows the value it would have had if irradiation were neglected .
let us define @xmath56 the maximum mass loss rate of irradiated models minus the @xmath57 value it would have had without irradiation feedback .
the general trend is that the stronger the irradiation regime , the larger @xmath58 . in order to understand the quasi - rlof state , for the moment
let us assume that the donor star fulfills the conditions in which undergoes cyclic mass transfer for all values of @xmath59 .
initially , irradiation feedback leads to @xmath60 . after a time of the order of the kelvin - helmholtz timescale @xmath61 ( where @xmath62
, @xmath63 is the gravitational constant , and @xmath64 , and @xmath65 are the radius and luminosity of the donor star , respectively ) the star relaxes , damping the perturbation due to irradiation .
then , the donor star contracts because it has a mass _ lower _ than it would need to stay in a semi - detached state .
thus , the rlof ends .
while the stellar envelope reacts this way , the deep interior has a chemical profile evolving due to nuclear reactions and mixing .
very often , this makes the star to swell enough , leading to the occurrence of another rlof on a nuclear timescale .
this is the very reason why stars may undergo a cyclic behavior .
the time the star needs to spend to reach another rlof conditions ( if any ) depends on the value of @xmath58 . if irradiation feedback is weak ( e.g. , the case of @xmath53= 0.01 ) @xmath58 is small and the star needs to evolve its internal chemical profile very little to rlof again .
therefore , the average duration of the mass transfer cycles is rather short , leading to the occurrence of a large number of cycles . on the contrary ,
if irradiation feedback is stronger ( e.g. , up to @xmath53= 1.00 ) , @xmath58 will be larger , the star will need more time to evolve to swell enough to rlof again , mass transfer cycles will be longer and the star will undergo a much lower number of cycles . the qualitative description given above can be analyzed with the help of fig .
[ fig : qrlof ] .
there we compare the evolution of the donor star for systems that initially had a normal star of 1.50 @xmath66 , a `` canonical '' ns of 1.4 @xmath66 , and 1 day orbital period .
we kept fixed @xmath67 and , in order to explore the whole parameter space of irradiation , we considered values of @xmath68 , and @xmath69 . here
we have modeled the long timescale evolution of the systems . in the upper and middle panels of fig .
[ fig : qrlof ] we display the mass transfer rate and the mass for these systems , complying with the behavior described above . while the model without irradiation shows a smooth evolution ,
irradiated models follow a step - like behavior in the mass .
these steps are smaller the weaker the irradiation strength . in the lower panel of fig .
[ fig : qrlof ] we show the evolution of the radius of the equivalent roche lobe against the stellar radius for the case of @xmath70 .
it is apparent that these radii are almost the same during rlofs ( the difference of these radii is very small , being of the order of the pressure scale height @xmath71 at the photosphere ) , while after detachment the star becomes slightly smaller than its associated roche lobe .
the star detaches because it has lost more mass than the amount it would if irradiation feedback were absent . in the latter conditions
, the star would be on a rlof continuously .
this represents a rather small perturbation to the stellar structure .
thus , there is no possibility for the donor star other than to have a radius slightly smaller than its roche lobe .
the donor star must settle on a _ quasi - rlof state_. these arguments are quite general in the sense that if an irradiated donor star undergoes cyclic mass transfer , it must be in the rlof or in the quasi - rlof state .
this provides a natural explanation to the occurrence of redback pulsars .
redbacks are a rapidly growing family of eclipsing binary systems containing a msp together with a donor star with a mass @xmath72 and orbital periods from 0.1 to 1.0 days . as described in bdvh , some redbacks evolve to the black widow state while others give rise to the formation of low - mass , hewd - pulsar system .
our models support the idea that all redbacks _ must _ have a non - degenerate companion in the quasi - rlof state
. this may also be the case for more massive objects usually interpreted as carbon - oxygen wd companions to pulsars ( see fig . 3 of bdvh ) .
the very existence of the quasi - rlof state indicates that pulsars may be observed accompanied not only by wds but also by non - degenerate companions ( in fact , the deep interior of the donor companion should be at densities high enough for the electrons to become semi - degenerate , but this is not the case of the extended envelope ) .
it also proves the relevance of considering irradiation feedback in cbs evolution .
as stated above , the system containing psr j1723 - 2837 is one of the best observed cases to be explained as a prototype . in order to interpret the available observations of psr j1723 - 2837 in the framework of an evolutionary model we have searched among the set of calculations presented in bdvh .
we found that the cbss formed by a normal , solar composition 1.25 @xmath66 star together with a ns of 1.40 @xmath66 on a 0.75 day circular orbit are suitable for such purpose .
we shall consider models with and without irradiation feedback . with the purpose of exploring the whole parameter space of irradiation we considered four values of the coupling constant @xmath73 0.00 , 0.01 , 0.10 , and 1.00 .
the comparison of the available data on psr j1723 - 2837 and the evolutionary tracks will be performed with the help of figures [ fig : hrd]-[fig : llenado ] in figure [ fig : hrd ] we present the evolutionary tracks in the hr diagram for each value of @xmath53 , where we indicate the range of effective temperatures compatible with the spectroscopic observations corresponding to the donor star presented by @xcite .
there are two epochs compatible with observations , one before bending bluewards and another after it . in both cases , the star has surface conditions similar to those of main sequence stars , but its internal structure is quite different . figure [ fig : perimasa ] depicts the donor mass vs. orbital period relationship for the considered systems .
the non - irradiated model reaches the observed orbital period of psr j1723 - 2837 in three epochs while irradiated models fulfill this condition several times . in all cases ,
the mass range at which the correct period is reached corresponds to the donor mass range given by @xcite .
the evolution of the surface gravity is shown in figure [ fig : grave ] .
again , these values are similar to those of main sequence stars , as it can be expected by the results included in figure [ fig : hrd ] .
the mass transfer rate for these systems is given in figure [ fig : mdot ] .
models considering irradiation feedback are consistent with the observed phenomenology , since we do not observe a lmxb at all times .
figure [ fig : qmasas ] denotes the mass ratio of the components of the systems considered above .
the observed value of @xmath27 is attained at an age of @xmath74 gyr .
finally , in figure [ fig : llenado ] we present the unfilled fraction of the roche lobe as a function of time .
irradiated , detached models have radii smaller than the corresponding roche lobe in at most a few percent . clearly , the donor star is in the quasi - rlof state as indicated by observations . in the frame of our models ,
a cbs capable to account for the properties of the psr j1723 - 2837 system has to undergo cyclic mass transfer episodes .
if , on the contrary , mass transfer were continuous , we would observe the system as a lmxb .
our models provide the observed mass ratio of psr j1723 - 2837 system for an age of @xmath75 gyr , which correspond to the portion of the evolutionary track presented in figure [ fig : hrd ] before bending bluewards . at this age ,
the models with @xmath73 0.00 , 0.10 , and 1.00 do not undergo cyclic mass transfer , meanwhile that with @xmath73 0.01 does .
thus , low - irradiation regimes provide a satisfactory account of the properties observed of the donor star in psr j1723 - 2837 system .
in the previous section we have discussed the relevance of evolutionary tracks including irradiation feedback to the case of the redback system psr j1723 - 2837 .
we found that the observations reported by @xcite are consistent with the theoretical picture with irradiation .
moreover , the analysis presented in section [ sec : aplicando ] indicates that the quasi - rlof state is reached by actual binary systems . in view of the generality of the arguments presented above ,
we suggest that essentially all redbacks should have quasi - rlof companions . in other words , that redbacks are the observational manifestation of the quasi - rlof state .
this is supported by the results presented in @xcite , particularly the data related to the roche lobe filling factors . in this study
we have considered a single value for the parameter @xmath38 ( = 0.5 ) .
this quantity is the fraction of the matter the ns is assumed to accrete ( @xmath76 $ ] ) .
while it is well - known that the evolution of donor stars is largely independent of @xmath38 , the mass of the ns is sensitively affected by this choice .
there have been some studies indicating that the value of @xmath38 may be lower in some cases ( see , e.g. , @xcite for the case of psr j1713 + 0747 ) .
if the ns mass growth were slower , this would modify the results presented in figure [ fig : qmasas ] ; and a good fit to psr j1723 - 2837 system would be provided by different initial conditions . in any case
we should remark that the initial ns mass value is , to a large extent , also uncertain .
evidently there is some degeneracy for the initial parameters of these binary systems and thus , we are _ not _ claiming that the evolutionary tracks we considered here in section [ sec : aplicando ] are the only possible ones . however , in finding the results presented in this paper we have not performed any fine tuning of the parameters of the evolutionary model , and yet found very suitable calculations on a previously computed set .
thus , for different values of @xmath38 other consistent solutions should exist as well .
some redback systems ( psr j1023 + 0038 and xss j12270 - 4859 ) show evidence of surrounding material @xcite while others do not . in the framework of our models
, they should correspond to the rlof and quasi - rlof states , respectively .
consequently , redbacks with surrounding material reveal as pulsars not due to the long term evolution resulting from our calculations but due to some physical processes ( radio ejection and/or accretion disk instabilities ) still to be included in the models .
the very existence of redbacks with a clean environment shows that irradiation feedback plays a key role in accounting for the evolution of these objects .
if there were no clean redback we could interpret that standard binary evolution coupled to radio ejection and/or accretion disk instabilities may be enough to account for observations .
however , this seems not to be the case .
the evolutionary status of redbacks has been recently addressed by @xcite employing models including evaporation .
they state that the evolution leading to redbacks or black widows is determined by the efciency of the irradiation process leading to mass loss via evaporation .
they claim that redback systems do not evolve into black widow ones .
this scenario is deeply different from the one described in bdvh , especially because @xcite do not find some stage similar to what we called quasi - rlof state while in the frame of our calculations we interpret that all redbacks correspond to this state .
moreover , we found that some redbacks evolve to black widows ( see bdvh for further details ) .
we have discussed the occurrence of the quasi - rlof state originally discussed in bdvh and applied it to the well - studied redback system psr j1723 - 2837 @xcite .
we found that low - irradiation models describe very well the properties of this binary system , suggesting that the quasi - rlof actually takes place along binary evolution .
this also supports the relevance of irradiation feedback for this kind of binary systems .
it is likely that the phenomenology observed in psr j1723 - 2837 system and in those quoted in the introduction of this paper ( section [ sec : introd ] ) is representative of the whole class of redbacks
. taken at face value , on human timescales , our models are able to account for the switch of a given system from lmxb to binary radio pulsar , or in the opposite direction , only once .
detecting such a transition in the frame of our evolutionary models has an extremely low probability and would be very unlikely .
however , in some systems ( see section [ sec : introd ] ) this switch has been observed from one state to the other and then back to the original .
therefore , a short ( @xmath0 1 yr ) timescale is clearly present in these systems and does not stem from our theoretical framework .
it seems natural to consider that the main ingredients to be included in the model to find such a short timescale for the switching behavior should be associated to the accretion disc surrounding the ns .
it is known that accretion discs may undergo instabilities on short timescales ( the disc instability model ; see , e.g. , @xcite ; @xcite ) that may fall in the required range .
recently , it has been possible to measure the orbital period change of sax 1808.4 - 3658 ( @xcite ; ) .
the reported value of @xmath77 is far larger the value expected for conservative ( @xmath78 ) binary evolution .
these authors interpret it as evidence for a strong mass loss from the system driven by radio ejection .
while this may correspond to the case of sax 1808.4 - 3658 , it is worth to remark that the radiation pressure of a radio pulsar does not necessarily destroys a surrounding accretion disk @xcite .
finding whether radio ejection and/or disc instability model are able to account for the observations described in section [ sec : introd ] remains an open problem and deserves further investigations .
despite this , we consider that the framework for the occurrence of these complex phenomena is given by the long term binary stellar evolution described in bhdv as well as in the present paper .
burderi , l. , & di salvo , t. 2013 , , 84 , 117 burderi , l. , di salvo , t. , riggio , a. , et al .
2010 , , 515 , a44 burderi , l. , riggio , a. , di salvo , t. , et al .
2009 , , 496 , l17 burderi , l. , dantona , f. , & burgay , m. 2002 , , 574 , 325 burderi , l. , possenti , a. , dantona , f. , et al .
2001 , , 560 , l71 roberts , m. s. e. 2013b , talk given in the workshop `` physical applications of millisecond pulsars '' , aspen center for physics , aspen co , january 19 - 24 , 2013 .
available at http://aspen13.phys.wvu.edu/aspen_talks/roberts_spiders.pdf , together with a ns of 1.4 @xmath66 , ( @xmath38=0.5 ) on a 1 day orbital period , neglecting irradiation and also considering it with @xmath79 , and @xmath69 .
upper panel shows the mass transfer cycles while middle panel shows the corresponding evolution of the donor mass .
lower panel shows the evolution of the radius of the donor star and the equivalent roche lobe .
there , for the sake of clarity , we only considered the case of @xmath70 .
it is noticeable that after detachments , the star is in the quasi - rlof state .
[ fig : qrlof ] ] evolving on a cbs together with a 1.4 @xmath66 mass ns on an 0.75 day orbit .
each panel correspond to different irradiation feedback regimes indicated with the corresponding value of @xmath53 .
the horizontal axis corresponds to the effective temperature of the _ non-_irradiated portion of the stellar surface .
vertical dashed lines indicate the range of effective temperatures for the donor star in psr j1723 - 2837 system @xcite .
[ fig : hrd ] ] .
the orbital period corresponding to psr j1723 - 2837 system is denoted with a horizontal dashed line .
notice that irradiated models reach the observed period value for several times while the non - irradiated one provides three values for the donor mass .
[ fig : perimasa ] ] .
the mass ratio observed in psr j1723 - 2837 system is denoted with a dotted line whereas its uncertainty is depicted with horizontal dashed lines .
the models considered provide the correct mass ratio at an age of @xmath74 gyr .
[ fig : qmasas ] ] .
non - irradiated model has a filled roche lobe for most of the time period depicted here ( for ages @xmath80 gyr ) whereas irradiated models at detached conditions have a radius that differ from that of the roche lobe at most in few percents .
the donor star is in the quasi - rlof state.[fig : llenado ] ] | we study the evolution of close binary systems formed by a normal ( solar composition ) , intermediate mass donor star together with a neutron star . we consider models including irradiation feedback and evaporation .
these non - standard ingredients deeply modify the mass transfer stages of these binaries .
while models that neglect irradiation feedback undergo continuous , long standing mass transfer episodes , models including these effect suffer a number cycles of mass transfer and detachment . during mass transfer
the systems should reveal themselves as low - mass x - ray binaries ( lmxbs ) , whereas when detached they behave as a binary radio pulsars .
we show that at these stages irradiated models are in a roche lobe overflow ( rlof ) state or in a _ quasi_-rlof state .
quasi - rlof stars have a radius slightly smaller than its roche lobe .
remarkably , these conditions are attained for orbital period and donor mass values in the range corresponding to a family of binary radio pulsars known as `` redbacks '' .
thus , redback companions should be quasi - rlof stars .
we show that the characteristics of the redback system psr j1723 - 2837 are accounted for by these models . in each mass transfer cycle
these systems should switch from lmxb to binary radio pulsar states with a timescale of @xmath0million years .
however , there is recent and fast growing evidence of systems switching on far shorter , human timescales .
this should be related to instabilities in the accretion disc surrounding the neutron star and/or radio ejection , still to be included in the model having the quasi - rlof state as a general condition .
@xmath1 member of the carrera del investigador cientfico of the comisin de investigaciones cientficas ( cic ) de la provincia de buenos aires , argentina .
+ @xmath2 member of the carrera del investigador cientfico , conicet , argentina .
+ @xmath3 |
dusty infrared ( ir ) galaxies are known to have a correlation between their infrared luminosities and dust temperatures ( `` @xmath5 relation '' ) . generally speaking ,
ultra - luminous infrared galaxies ( ulirgs ; @xmath7 ) and hyper - luminous infrared galaxies ( hylirgs ; @xmath8 ) have typical dust temperatures of @xmath9 k , while luminous infrared galaxies ( lirgs ; @xmath10 ) and others at lower luminosities have lower dust temperatures of @xmath11 k. this trend was clearly revealed when sub - millimeter galaxies ( smgs ) , which are high - z ulirgs , were compared to ulirgs and other ir galaxies in the local universe ( e.g. , * ? ? ?
* ; * ? ? ?
there have been a number of studies to understand the nature of this relation and its dispersions , such as its dependence on redshifts , different galaxy populations , etc .
( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in ( * ? ?
* hereafter my15 ) , we have studied the quasars from the sloan digital sky survey ( sdss ) that have far - ir ( fir ) counterparts in the wide - field survey data from the _ herschel _ space observatory @xcite .
we have shown that the majority of them are ulirgs , and that their fir emission originated from the cold - dust component is predominantly powered by starbursts rather than active galactic nuclei ( agn ) .
one of our conclusions is that they follow the same @xmath5 relation .
we have further shown that this relation is simply due to the limited maximum size of the combined dusty starbursting regions ( hereafter dsbrs ) in the host galaxies .
the motivation of our current work is that this maximum size is only around @xmath12 kpc , which seems to be significantly smaller than the claimed physical sizes of high - z ulirgs based on a number of direct measurements through sub - millimeter ( sub - mm ) or radio interferometry .
these observations have sub - arcsec resolutions , and typically result in rather extended sizes of @xmath13 kpc or even larger ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . here
we further investigate this problem , using an enlarged sample that incorporates different high - z ir galaxy populations . throughout this _ letter _ , we adopt the following cosmological parameters : @xmath14 , @xmath15 and @xmath16 km / s / mpc .
we followed my15 in this current analysis . only the objects that have spectroscopic redshifts
were used .
the fir counterpart identification was mainly based on the latest source catalogs from three major wide field surveys by _ herschel _ , namely , the herschel multi - tiered extragalactic survey ( hermes , @xmath17 ; * ? ? ?
* ; * ? ? ?
* ) , the herschel stripe 82 survey ( hers , @xmath17 ; * ? ? ?
* ) , and the scientific demonstration phase data of the herschel astrophysical terahertz large area survey ( h - atlas sdp , @xmath17 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the fir spectral energy distributions ( seds ) were constructed using the three - band photometry from the spectral and photometric imaging receiver ( spire ; * ? ? ?
* ) at 250 , 350 and .
we only considered the objects that are detected in all these three bands .
the largest addition to the my15 sample is through the use of the @xmath18 edition vron catalogue of qusars and active nuclei @xcite , which contains and agn and quasars , respectively ( hereafter `` v - agn '' and `` v - qso '' , respectively ) . the next is from ( * ? ? ?
* c12 ) , which contains spectroscopically confirmed spire sources in the hermes fields .
we also included the smgs from ( * ? ? ? * c05 ) and ( * ? ? ?
* i05 ) , and a few sources of mixed selections from ( * ? ? ? * m11 ) and ( * ? ? ? * y14 ) .
any duplicates among these sources and/or the my15 sample were removed .
as in my15 , we obtained the spire photometry by matching the sources to the aforementioned _ herschel _ catalogs , adopting the matching radius of to minimize the contaminated sources due to blending .
we also incorporated a number of objects that are outside of the aforementioned _ herschel _ survey fields but have reported spire photometry .
these include the high - z quasars from ( * ? ? ?
* l13 ) , the radio galaxies from ( * ? ? ?
* d14 ) , and the 3c radio sources from ( * ? ? ?
* p15 ) .
while these objects certainly contain powerful agn , we are convinced ( as are these authors also inclined to believe ) that their fir emissions should be dominated by starbursts as in the sdss quasars presented in my15 .
finally , we included a few high - z ulirgs that are known to be amplified by strong gravitational lensing , which all have spectroscopic redshifts , reported spire photometry and adopted amplification factors ( `` @xmath19 '' ) .
these include the most highly lensed smg from @xcite and ( * ? ? ?
* s&i10 ) , the strong planck source from ( * ? ? ?
* f12 ) , the south pole telescope s lensed galaxies from @xcite and , and the lensed galaxies in the h - atlas sdp from ( * ? ? ?
* b13 ; only the two sources with grade a lensing models and @xmath20 were used ) . as in my15
, we only studied the coldest dust component , which dominates the fir emission that is sampled by the spire bands .
while this will certainly underestimate ( by @xmath21 dex ) the total ir luminosity that consists of the contributions from other components of higher temperatures , the simplification has the advantages that the associated luminosity can be safely attributed to starbursts and that the dust temperature is uniquely defined .
following my15 , we analyzed their fir seds by fitting a single - temperature , modified blackbody ( mbb ) spectrum .
we briefly describe the procedure below .
we used the ` cmcirsed ` code of @xcite to perform the mbb fitting .
the mbb spectrum can be written as @xmath22 where @xmath3 is the characteristic temperature of the mbb , @xmath23 is the scaling factor that is related to the intrinsic luminosity , @xmath24 is the emissivity ( when @xmath25 equation [ eq : mbb ] reduces to the form of a black body ) , and @xmath26 is the reference wavelength where the opacity is unity .
we adopted @xmath27 and @xmath28 .
the total ir luminosity of the cold - dust component , @xmath2 , can be obtained as @xmath29 by integrating the best - fit mbb model @xmath30 from 8 to .
our further analysis is based on @xmath2 and @xmath3 thus obtained .
we only retained the objects that have reasonable sed fitting quality ( @xmath31 ) and reliable @xmath3 measurements ( @xmath32 ) , which sum up to objects in total .
as examples , figure [ fig : sedfitting ] shows the sed fitting results for a few objects from our final sample .
figure [ fig : ltobj ] shows the @xmath5 relation from the entire sample , where the number of contributing objects from each initial sample is also labeled .
the distributions of redshifts and @xmath2 are shown in the first two histograms in figure [ fig : hist ] .
as addressed in my15 , the @xmath5 trend revealed in figure [ fig : ltobj ] is not due to selection effects but is governed by the equivalent of stefan - boltzmann law in case of mbb .
here we fully develop this idea . the location of an object on the ( @xmath2 , @xmath3 ) plane is determined by the intensity of the mbb as expressed in equation . to obtain the total power radiated from the cold - dust component of the object , i.e. , its total luminosity , one should integrate the mbb function @xmath33 over wavelength and over the solid angle ( `` @xmath34 '' ) that its surface area ( `` a '' ) subtends as seen from the source , and multiply by this surface area : @xmath35 where we make the substitution of @xmath36 , set @xmath37 , and also write @xmath3 as @xmath1 for convenience .
if there were no the modified term @xmath38 to the planck s black body function , the integral above would be @xmath39 and equation would reduce to the stefan - boltzmann law , i.e. , @xmath40 , where @xmath41 is the stefan - boltzmann constant . for simplicity
, we introduce @xmath42 which is the equivalent of the stefan - boltzmann constant in the mbb case .
note that @xmath43 is dependent of @xmath1 because @xmath37 is involved . in the context of this work
, we can take the approximation that the integral in equation is only over the conventional total ir regime of @xmath30 from 8 to as in equation such that @xmath44 , i.e. , we have @xmath45 , which we shall refer to as the `` mbb s - b equivalent '' . assuming spherical symmetry , we can define an `` effective radius '' , @xmath4 , such that @xmath46 .
this is to imagine that all the star - forming regions within the galaxy are combined together and that the sum can be approximated by an effective sphere with the radius of @xmath4 .
we thus have @xmath47 for @xmath3 within the range of interest ( @xmath48100 k ) , the deviation of @xmath43 from @xmath49 is within a factor of two and can be well approximated as @xmath50 for our choice of @xmath51 @xmath19 m and @xmath27 ( see the right panel of figure [ fig : ltobj ] ) .
thus the data points on the ( @xmath2,@xmath3 ) plane can be explained by a family of mbb s - b equivalent curves of different @xmath4 , which are shown in figure [ fig : ltobj ] . a pair of ( @xmath2 , @xmath3 ) values allow us determine @xmath4 of the combined dusty star forming regions in the galaxy under question .
my15 uses the same argument , and shows that ( 1 ) in the low luminosity regime the increasing of @xmath2 is due to the increasing of @xmath4 , while in the high luminosity regime the increase of @xmath2 has to be attributed to the increased heating intensity , and ( 2 ) @xmath4 can not be increased arbitrarily and has a limit of @xmath6 kpc .
my15 takes a less rigorous approach and approximates the temperature - dependent @xmath43 by @xmath52 and obtains @xmath53 .
the derivation of equation here is more appropriate , which results in more accurate determination of @xmath4 .
the histograms of @xmath3 and @xmath4 are shown in figure [ fig : hist ] for all the objects in our sample . the vast majority of them ( 98.3% )
have @xmath54 kpc , and this limit is the reason for the observed @xmath5 relation .
following @xcite to convert the infrared luminosity to the sfr and assuming the initial mass function of @xcite , we get @xmath55 @xmath56 . from the @xmath4 values
above , we can derive the sfr surface densities ( hereafter `` sfr2d '' , or `` @xmath57 '' ) that are often used in the literature , where the `` surface area '' going into the calculation is the area of the dsbrs projected on the sky .
assuming spherical symmetry , the total surface area of the @xmath58-th dsbr is @xmath59 , where @xmath60 is its radius .
we further make a highly simplified assumption that all the dsbrs in a given galaxy have the same dust temperature .
the definition of @xmath4 thus implies that @xmath61 , where the summation over @xmath58 goes through all the dsbrs .
the projected surface area of the @xmath58-th dsbr on the sky is @xmath62 , and hence @xmath63 .
considering equation and that @xmath64 , we have @xmath65 .
this means that @xmath57 is constant for a fixed @xmath3 .
similarly , we can calculate the sfr volume density ( hereafter `` sfr3d '' , or `` @xmath66 '' ) . under the same assumptions as above
, we define a different effective sphere with the radius of @xmath67 , whose volume is equal to the sum of the volumes of the individual dsbrs in a galaxy , i.e. , @xmath68 .
it is obvious that @xmath69 , and generally there is no easy way to infer @xmath67 from @xmath4 .
however , in the limiting case that there is only one dsbr in the galaxy under question , we should have @xmath70 . therefore
, we can calculate the _ minimum _ sfr volume density ( hereafter `` minsfr3d '' , or `` @xmath71 '' ) as @xmath72 . the histograms of sfr2d and minsfr3d are shown in figure [ fig : hist ] .
the @xmath0-@xmath1 relation now has a new meaning .
as a pair of ( @xmath2 , @xmath3 ) correspond to one @xmath4 value , and hence one minsfr3d value , the @xmath2-@xmath1 plane can be converted into a minsfr3d `` surface '' , which is shown in figure [ fig : tlir_sfr3d ] , where the average trend based on the data points in figure [ fig : ltobj ] is also displayed .
it is immediately clear that our objects form the observed @xmath0-@xmath1 relation because they outline a region that has a narrow spread in minsfr3d .
we emphasize again that , as shown in the simulation of my15 ( see also 2 above ) , the lack of objects in the area below this relation can not be due to selection bias . from figure [
fig : tlir_sfr3d ] , it is clear that this high-@xmath2 , low-@xmath3 area corresponds to low minsfr3d .
we therefore suggest that this area is a `` zone of avoidance '' in terms of minsfr3d . in other words , very cold ulirgs or hylirgs
should be very rare , because they would require very large @xmath4 ( and hence very low minsfrd ) in order to achieve a high ir luminosity at a low dust temperature . on the other hand , the lack of objects in the area above the current @xmath0-@xmath1 relation outlined by our objects
_ could be _ due to the possible selection bias in our current sample , where we limit to @xmath73 mjy in order to include the most reliable detections in the _ herschel _ spire bands .
we refer the readers to the discussion in my15 for details ( see 4.1 and figure 12 therein ) .
while the limited @xmath4 ( @xmath6 kpc ) seem to contradict the claimed extended sizes of high - z smgs ( see 1 ) , there is a reconciliation when considering that even the sub - arcsec sub - mm interferometry might still can not resolve individual dsbrs at high - z ( @xmath74 kpc at @xmath75 ) if they are too close to each other . in other words
, the very extended sub - mm morphologies could simply be due to the blending of a number of discrete dsbrs that are widely separated but still not resolved by the interferometry available today .
currently , the only way to resolve the individual dsbrs at high - z is through strong gravitational lensing , in which case the highly amplified background source could show its morphology in great details , albeit in the distorted image plane . by properly modeling the lens , its image in the source plane
could be reconstructed to reveal the intrinsic morphology .
however , the accuracy of the reconstruction depends highly on how strong the amplification is and how well the lens is modeled . unfortunately , there are still not many such measurements publicly available for us to compare to , especially when we require spectroscopic redshifts and the spire photometry .
the best example to date is the smg smmj2135 - 0102 from s&i10 , which is amplified by a factor of @xmath76 .
its fir emission is dominated by four individual dsbrs , each being mirrored into two images . by averaging the results from the reconstructions based on the two sets of mirrored images , the intrinsic radii of these four dsbrs are 390.0 , 290.5 , 192.5 , and 94 pc , respectively .
as these radii are the fwhm sizes , and hence only enclose @xmath77% of the total light from each dsbr , they should be corrected by multiplying a factor of @xmath78 to the `` full sizes '' . following
3.2 , we get @xmath79 pc . in 3.1 , we obtained @xmath80 , @xmath81 k and @xmath82 pc .
this size agrees with the above to @xmath83% . considering that these four dsbrs ,
while being the dominant sources , might not contribute 100% of the total @xmath2 , the actual agreement could be even better . for comparison , here
we also discuss the case of arp 220 , which is the representative of the classic ( low - redshift ) ulirg population .
arp 220 has two nuclei , which have recently been directly resolved in sub - mm by @xcite using the atacama large millimeter / submillimeter array ( alma ) . based on their highest resolution observations at ( half - power bandwidth of @xmath84 )
, these authors derived the fwhm sizes of the two nuclei ( assuming gaussian profiles ) as @xmath85 and @xmath86 pc , respectively . using spheres for approximation
, we obtain the equivalent radii of 106 and 126 pc , respectively .
they also suggest that these two nuclei account for @xmath87 - 76% of the total fir continuum , and hence we adopt the middle value of 74% .
similar to the calculation above , we correct the measured fwhm sizes by a factor of @xmath88 to the full sizes , and obtain @xmath89 pc .
as arp 220 is very close ( we adopt 77 mpc as its luminosity distance ) , using the spire bands alone would not result in a reasonable fit to its sed because these bands are too far away from the peak of its fir emission .
therefore , we use the photometry at @xmath90 , retrieved from the nasa / ipac extragalactic database .
we obtain @xmath91 , @xmath92 k and @xmath93 pc . considering the uncertainties ( especially the fractional contribution from the two nuclei to the total fir emission )
, the agreement is very reasonable .
our mbb model fixes @xmath26 to , which is about the smallest choice adopted in the literature .
as discussed in my15 ( see appendix a ) , the difference in @xmath26 impacts the derived @xmath3 significantly but has little effect on @xmath2 .
if the sed fitting can achieve the similar quality , adopting a larger @xmath26 will result in a higher @xmath3 , and hence a smaller @xmath4 . for a given @xmath26 ,
@xmath4 still sensitively depends on @xmath3 .
for example , if we take a crude approximation and ignore the temperature dependence of @xmath94 , we have @xmath95 .
therefore , it is difficult to derive @xmath4 to an accuracy better than a factor of a few , especially at @xmath96 where the mbb s - b equivalent curves are highly degenerated .
however , our main conclusions still hold regardless of such limitations .
first of all , @xmath4 can not be increased arbitrarily . even in the hylirg regime ,
it is still mostly confined to @xmath6 kpc . while in the low luminosity regime
the increasing of @xmath2 can be achieved by increasing @xmath4 ( e.g. , increasing the number of dsbrs ) , in the high luminosity regime this can only be achieved by increasing the strength of the starburst ( i.e. , reflected in the rapid increase of @xmath3 ) .
my15 has already reached this conclusion , and here we reinforce it with an enlarged sample that consists of objects from different initial selections , such as smgs , high - z radio galaxies , etc .
second , the @xmath0@xmath1 relation reflects the equivalent of the stefan - boltzmann law in case of mbb and the aforementioned limit to @xmath4 .
for this reason , this relation outlines the boundary of minsfr3d , which provides new clues in understanding dusty starbursting environment .
the value of minsfr3d should be quite close to the actual sfr volume density , although strictly speaking it is only the lower limit . in the simplest case where the dsbrs are all spheres of the same radius and dust temperature , @xmath97 , where @xmath23 is the total number of dsbrs in the galaxy .
third , the small @xmath4 values suggest that the dsbrs in high - z ulirgs are as physically compact as their counterparts in the local universe . it is true that a small @xmath4 could still be the result of an extended dsbr with a low filling factor .
however , our test cases in 3.4 strongly supports the scenario that dsbrs are universally compact . | using the far - infrared data obtained by the _ herschel space observatory _ , we study the relation between the infrared luminosity ( @xmath0 ) and the dust temperature ( @xmath1 ) of dusty starbursting galaxies at high redshifts ( high - z ) .
we focus on the total infrared luminosity from the cold - dust component ( @xmath2 ) , whose emission can be described by a modified black body ( mbb ) of a single temperature ( @xmath3 ) .
an object on the ( @xmath2 , @xmath3 ) plane can be explained by the equivalent of the stefan - boltzmann law for a mbb with an effective radius of @xmath4 .
we show that @xmath4 is a good measure of the combined size of the dusty starbursting regions ( dsbrs ) of the host galaxy . in at least one case where the individual dsbrs are well resolved through strong gravitational lensing ,
@xmath4 is consistent with the direct size measurement .
we show that the observed @xmath5 relation is simply due to the limited @xmath4 ( @xmath6 kpc )
. the small @xmath4 values also agree with the compact sizes of the dsbrs seen in the local universe .
however , previous interferometric observations to resolve high - z dusty starbursting galaxies often quote much larger sizes .
this inconsistency can be reconciled by the blending effect when considering that the current interferometry might still not be of sufficient resolution . from @xmath4
we infer the lower limits to the volume densities of the star formation rate ( `` minsfr3d '' ) in the dsbrs , and find that the @xmath0-@xmath1 relation outlines a boundary on the ( @xmath2 , @xmath1 ) plane , below which is the `` zone of avoidance '' in terms of minsfr3d . |
the solution of the numerically ill - posed linear system of equations @xmath0 is considered .
we suppose that the matrix @xmath1 has large dimension , may be over or underdetermined , @xmath2 , or @xmath3 , resp . and is severely ill - conditioned ; the singular values of @xmath1 decay exponentially to zero , or to the limits of the numerical precision .
vectors @xmath4 and @xmath5 describe the data and model parameters respectively .
noise in the data is represented by @xmath6 , i.e. @xmath7 for exact but unknown data @xmath8 .
we assume components of @xmath9 are independently sampled from a gaussian distribution and have known diagonal covariance matrix , denoted by @xmath10 . given @xmath1 and @xmath4
the aim is to compute an approximate solution for @xmath5 .
discrete ill - posed problems of the form may be obtained by discretizing linear ill - posed problems such as fredholm integral equations of the first kind and arise in many research areas including image deblurring , geophysics , etc .
the presence of the noise in the data and the ill - conditioning of @xmath1 means that regularization is needed in order to compute an approximate solution for .
probably the most well - known method is that of tikhonov regularization in which @xmath5 is estimated as @xmath11 here @xmath12 is the weighted data fidelity term and @xmath13 is a regularization term .
the product @xmath14 predicts the data , @xmath15 is a regularization matrix and @xmath16 allows specification of a given reference vector of _ a priori _ information for the model @xmath5 . in @xmath17 is an unknown regularization parameter which trades - off between the data fidelity and regularization terms .
the noise in the measurements @xmath18 is whitened when @xmath19 where @xmath20 approximates the standard deviation of the noise @xmath21 in the @xmath22th datum . introducing @xmath23 , @xmath24 , and shifting by the prior information through @xmath25 , yields @xmath26 under the assumption that the null spaces of @xmath27 and @xmath15 do not intersect @xmath28 is explicitly dependent on @xmath17 and is given by @xmath29
it is well - known that when the matrix @xmath15 is invertible , the alternative but equivalent formulation uses @xmath30 yielding , with right preconditioned matrix @xmath31 and regularized inverse @xmath32 , @xmath33 although equivalent analytically , numerical techniques to solve and differ . for small scale problems , for example , we may solve using the generalized singular value decomposition ( gsvd ) , e.g. @xcite , for the matrix pair @xmath34 $ ] , but would use the singular value decomposition ( svd ) of @xmath35 for , e.g. @xcite , as given in [ svd ] .
the solutions depend on the stability of these underlying decompositions , as well as the feasibility of calculating @xmath36 .
still , the use of the svd or gsvd is not viable computationally for large scale problems unless the underlying operators possess a specific structure .
for example , if the underlying system matrix , and associated regularization matrix are expressible via kronecker decompositions , e.g. @xcite , then the gsvd decomposition can be found via the gsvd for each dimension separately .
here we consider the more general situation and turn to consideration of iterative krylov methods to estimate the solutions of and .
forthwith we assume for simplicity of notation and the initial discussion that we solve the system @xmath37 , equivalent to using @xmath38 , weighting of @xmath4 and @xmath1 by @xmath39 , and right preconditioning of @xmath1 by @xmath36 , dependent on the context , i.e. we solve with @xmath40 , @xmath41 and @xmath38 .
specifically , we assume that the components of the error @xmath21 are independently sampled from a normal distribution with variance @xmath42 , @xmath43 . in principle
iterative methods such as conjugate gradients ( cg ) , or other krylov methods , can be employed to solve for large scale problems in the presence of noise .
results presented in @xcite demonstrate , however , that minres and gmres should not be used as regularizing krylov iterations due to the early transfer of noise to the krylov basis .
recently , there has also been some interest in the lsmr modification of lsqr which is an implementation based on minres , @xcite . here
, because our goal is to understand how to find regularization parameters for a well - studied reduced problem , we use the well - known golub - kahan bidiagonalization ( gkb ) , also known as the lsqr iteration which has been well - studied in the context of projected solutions of the least squares problem @xcite and for which the noise regularizing properties of the iteration are better understood , @xcite .
effectively the gkb projects the solution of the inverse problem to a smaller subspace , say of size @xmath44 . applying @xmath44 steps of the gkb on matrix @xmath1 with initial vector @xmath45 , of norm @xmath46 , and defining @xmath47 to be the unit vector of length @xmath48 with a @xmath42 in the first entry , lower bidiagonal matrix @xmath49 and column orthonormal matrices @xmath50 , @xmath51
are generated such that , see @xcite , @xmath52 for @xmath53 , we define the full , @xmath54 , and projected , @xmath55 , residuals via @xmath56 by the column orthonormality of @xmath57 @xmath58 theoretically , therefore , an estimate for the solution @xmath5 with respect to a reduced subspace may be found by solving the normal equations for the projected problem and then projecting back to the full problem . by the courant fischer minimax theorem and noting @xmath59 it is immediate that the eigenvalues of @xmath60 , the ritz values , are interlaced by those of @xmath61 and are bounded above and below by the largest and smallest non zero eigenvalues of @xmath61 , ( * ? ? ?
* section 5 ) .
likewise , the singular values , @xmath62 , of @xmath63 interlace the singular values @xmath64 of @xmath1 , @xmath65 and the singular vectors of @xmath1 are approximated , via @xmath66 and @xmath67 , for the singular value decomposition @xmath68 , given sufficient precision , e.g. @xcite .
thus , for large enough @xmath44 , dependent on the spectrum for @xmath1 , the normal equations for the projected system of equations inherit the ill - conditioning of the normal equations for and regularization is also needed for the noise - contaminated projected problem , @xcite , despite the regularizing impact of the krylov iterations . by the column orthonormality of @xmath69
, we have @xmath70 . explicitly introducing regularization parameter @xmath71 , distinct from @xmath17 in order to emphasize regularization on the projected problem , yields the projected tikhonov problem @xmath72 the solution of has two equivalent forms
, it will be helpful for theoretical analysis to give both , dependent on whether derived directly for , or without factoring out @xmath57 in @xmath73 in practice , one uses to find @xmath74 via the svd for @xmath63 , under the assumption that @xmath75 , noting that an explicit solution for @xmath76 is immediately available , see e.g. [ svd ] .
as already observed in @xcite , the regularized lsqr algorithm now poses the problem of both detecting the appropriate number of steps @xmath44 as well as of finding the optimal parameter @xmath77 .
one method of regularization is simply to avoid the introduction of the regularizer in and find an optimal @xmath44 at which to stop the iteration , equivalently regarding the lsqr iteration itself as sufficiently regularizing .
here we assume that regularization is needed , and thus with respect to the regularization , it is evident that the problem may be considered in two ways , namely regularize and project , or project and then regularize , e.g. @xcite .
if the regularization is applied for each step of the reduction , the method is regarded as a hybrid of the two techniques , e.g. @xcite .
the problem of first determining the appropriate size @xmath44 for the projected space is discussed in e.g. @xcite and more recently for large scale geophysical inversion in @xcite .
on the other hand , for a given @xmath44 the krylov subspace for @xmath74 is the same for all @xmath71 , i.e. the krylov subspace is invariant to shifts , which is useful for determining @xmath77 , @xcite .
although the solutions obtained from the regularize then project , and project then regularize , for a given @xmath44 and @xmath78 are equivalent , ( * ? ? ?
* theorem 3.1 ) which also points to ( * ? ? ?
* p 301 ) , this does not mean that for given @xmath44 finding @xmath77 for the subspace problem will provide @xmath79 that is optimal for the full problem , @xcite . determining to which degree certain regularization techniques provide a good estimate for @xmath79 from the subspace problem estimate , and the conditions under which this will hold , is the topic of this work and is the reason we denote regularization parameter on the subspace by @xmath71 distinct from @xmath17 .
we should note that in our discussion we explicitly assume invertibility of the regularization operator @xmath15 in order to allow the right preconditioning of @xmath1 .
practically we note that in applying the gkb reduction , applications of forward operations @xmath80 should be performed via solves for the systems @xmath81 to find @xmath82 .
typically @xmath15 is sparse and structured , and such solves are efficient given a potential initial factorization for @xmath15 .
the gkb also requires forward operations with @xmath83 which are again efficiently implemented via solution solves with system matrix @xmath84 .
here we assume that such efficient solves are possible , and do not address this aspect of algorithmic development .
we also refer to recent work in @xcite in which the most often used differential operators , themselves not invertible , are replaced by invertible versions by use of suitable boundary conditions . when @xmath15 is not invertible , the projected tikhonov problem lacks the immediate preservation of the regularization norm , yielding the subspace problem @xmath85 which unfortunately requires projecting the subspace solution back to the full space via @xmath69 , which is avoided in . on the other hand
the regularizer @xmath86 may be achieved , as noted in @xcite , by finding the @xmath87 factorization @xmath88 , yielding @xmath89 by the column orthogonality of @xmath90 .
the formulae to immediately find @xmath74 for small @xmath44 use in this case the gsvd , see e.g. @xcite .
having now set the background for our investigation , we reiterate that a main goal of this work is to theoretically analyze in which cases determining @xmath77 from the projected problem will effectively regularize the full problem . the presented analysis is independent of whether the originating problem is over or under determined . for the full problem
the question of determining an optimal parameter @xmath79 is well - studied , see e.g. @xcite , for a discussion of methods including the morozow discrepancy principle ( mdp ) , the l - curve ( lc ) , generalized cross validation ( gcv ) and unbiased predictive risk estimation ( upre ) .
the use of the mdp , lc and gcv is also widely discussed for the projected problem , particularly starting with the work of kilmer et al , @xcite and continued in @xcite .
further , extensions for windowed regularization , and hence multi - parameter regularization , @xcite are also applied for the projected problem @xcite .
our attention is initially on the use of the upre , for which we find a useful connection between the full and projected formulations .
this is not so immediately clear , particularly for the gcv and is , we believe , the reason why a weighted gcv ( wgcv ) was required in @xcite . in our work we are also able to heuristically explain the weighting parameter in the wgcv .
we stress that the approach assumes that the projected system is calculated with full reorthogonalization , a point often overlooked in many discussions , although it is apparent than many references implicitly make this assumption .
although our analysis should also be relevant for the case of windowed regularization , this is not a topic for this paper , and will be considered in future work .
instead we extend the hybrid approach for use with an iteratively reweighted regularizer ( irr ) , which sharpens edges within the solution , @xcite , hence demonstrating that edge preserving regularization can be applied in the context of regularized lsqr solutions of the least squares problem on a projected subspace .
the paper is organized as follows .
the regularization parameter estimation techniques of interest are presented in section [ sec : parameter estimation ] .
the discussion in section [ sec : parameter estimation ] is validated with one dimensional simulations in section [ sec : simulationoned ] .
image restoration problems presented in section [ sec : simulationtwod ] illustrate the relevance for the two dimensional case .
we then go further and demonstrate the use of irr @xcite , an approach for approximating the total variation regularization , in section [ sec : irr ] . finally in section [ sec : walnut ] we also illustrate the algorithms in the context of sparse tomographic reconstruction of a walnut data set , @xcite , demonstrating the more general use of the approach beyond deblurring of noisy data .
our conclusions are presented in section [ conclusions ] .
it is of particular interest that our analysis applies for both over and under determined systems of equations and is thus potentially of future use for other algorithms also in which alternative regularizers are imposed and also require repeated tikhonov solves at each step .
further , this work extends our analysis of the upre in the context of underdetermined but small scale problems in @xcite , and demonstrates that irr can be applied for projected solutions .
although it is well - known that the mdp always leads to an over estimation of the regularization parameter , e.g. @xcite , it is still a widely used method for many applications , and is thus an important baseline for comparison . on the other hand ,
the upre is less well - accepted but often leads to a better estimation of the regularization parameter , e.g. @xcite . in order to use any specific regularization parameter estimation method for the projected problem it is helpful to understand the derivation on the full problem . in the discussion that follows we explicitly assume that lsqr is implemented using sufficient precision , namely with full reorthogonalization of the columns of @xmath69 and @xmath57 so that holds .
the predictive error , @xmath91 , for the solution @xmath92 , is defined by @xmath93 where @xmath94 is the influence matrix , and compares with the full residual @xmath95 in both equations the first term is deterministic , whereas the second is stochastic , through the assumption that @xmath96 is a random vector . for completeness
, we give the trace lemma e.g. ( * ? ? ?
* lemma 7.2 ) , as required for the following discussion . for deterministic vector @xmath97 , random vector @xmath9 with diagonal covariance matrix @xmath98 , matrix @xmath99 , and expectation operator @xmath100 @xmath101 using @xmath102 to denote the trace of matrix @xmath1 .
applying to both and with the assumption that @xmath103 , due to whitening of noise @xmath9 , and using the symmetry of the influence matrix , we obtain @xmath104 where @xmath105 is the defined to be the expected value of the risk of using the solution @xmath92 .
although the first term on the right hand side in each case can not be obtained , we may use @xmath106 in .
thus using linearity of the trace and eliminating the first term in the right hand side of gives the upre estimator for the optimal parameter @xmath107 typically , @xmath79 is found by evaluating for a range of @xmath17 , for example by the svd see e.g. [ appb ] , with the minimum found within that range of parameter values , as suggested in @xcite for the gcv .
see also e.g. ( * ? ? ?
* appendix , ( a.6 ) ) for the formulae for calculating the functional in terms of the svd of matrix @xmath1 . for the projected case we consider two different approaches for minimizing the predictive risk using the solution of the projected problem .
first observe that from @xmath108 and by the column orthogonality of @xmath69 , we have @xmath109 the residual , with respect to the solution of the projected problem explicitly depending on the regularization parameter @xmath71 , is now given by @xmath110 where @xmath111 , consistent with the definition of the influence matrix @xmath112 .
similarly the predictive error is given by @xmath113 comparing with , and with , gives the upre functional for finding the regularization parameter for the solution for the full problem with the solution found with respect to the projected subspace @xmath114 expanding @xmath115 gives @xmath116 where the last equality follows from the cycle property of the trace operator for consistently sized matrices .
hence @xmath117 can be evaluated without reprojecting the solution for every @xmath71 back to the full problem .
when estimated by upre , the optimal @xmath71 for the solution on the projected space can be found from the projected solution alone .
it remains to question whether has any relevance with respect to the projected solution , i.e. does this appropriately regularize the projected solution , otherwise it may not be appropriate to find @xmath71 to minimize this functional on the subspace .
observe for @xmath118 , right hand side vector @xmath119 consists of a deterministic and stochastic part , @xmath120 , where for white noise vector @xmath9 and column orthogonal @xmath57 , @xmath121 is a random vector of length @xmath48 with covariance matrix @xmath122 .
thus from the derivation of the upre for the full problem defined by system matrix @xmath1 , right hand side @xmath4 and white noise vector @xmath9 , we may immediately write down the upre for the projected problem with system matrix @xmath63 , right hand side @xmath123 and white noise vector @xmath121 . in particular
, defining @xmath124 and comparing with , it is immediate that minimizing to minimize the risk for the projected solution , also minimizes the risk for the full solution , with respect to the given subspace .
the shift by @xmath48 as compared to @xmath125 is irrelevant in the minimization of the functional .
note that this does not immediately minimize the predictive risk ( [ optalpha2 ] ) for the full problem , i.e. @xmath126 , because @xmath79 is needed with respect to solutions in @xmath127 and not just restricted to @xmath128 . by the linearity and cycle properties of the trace operator @xmath129 and @xmath130 in exact arithmetic the large singular values of @xmath63
provide a good approximation of the large singular values of @xmath1 , ( * ? ? ?
* section 9.3.3 ) .
thus suppose @xmath44 is such that the first @xmath44 singular values are well - approximated by those of @xmath63 , and that the ill - conditioning of @xmath1 is effectively captured so that there is clear separation between @xmath131 and @xmath132 .
then for regularizing the full problem , for which @xmath133 , @xcite , and with filter factor @xmath134 , @xmath135 , we have @xmath136 comparing and , with @xmath137 in , we see that we may interpret the determination of @xmath77 as giving a good estimate for @xmath79 if @xmath138 .
further , if @xmath139 provides a good estimate for @xmath140 we may interpret the determination of @xmath77 as giving a good estimate for @xmath79 in the case in which the filter factors in the tikhonov regularization are determined for the truncated singular value decomposition ( tsvd ) of @xmath1 , with truncation at @xmath44 .
this observation follows theorem 3.2 @xcite which connects the use of the tsvd of @xmath63 for the solution with the solution obtained using the tsvd of @xmath1 . to summarize : if @xmath44 is such that @xmath141 for @xmath135 , so that @xmath142 and @xmath139 approximates @xmath140 , then @xmath143 when obtained using the upre
further , the estimate is found without projecting the solution back to the full space , namely by minimizing . the premise of the mdp , @xcite , to find @xmath17 is the assumption that the norm of the residual , @xmath144 follows a @xmath145 distribution with @xmath146 degrees of freedom , @xmath147 .
heuristically , the rationale for this choice is seen by re - expressing @xmath148 so that if @xmath149 has been found as a good estimate for @xmath150 then the residual should be dominated by the whitened error vector @xmath9 . for white noise @xmath151
is distributed as a @xmath145 distribution with @xmath125 degrees of freedom , from which @xmath152 , with variance @xmath153 .
thus we seek a residual such that @xmath154 using a newton root - finding method , see [ appb ] , where we take safety parameter @xmath155 to handle the well - known over smoothing of the mdp .
alternatively , we note @xmath156 , where the size of @xmath157 depends on the percentiles of the cumulative @xmath145 distribution with @xmath125 degrees of freedom . the larger @xmath157
the less confidence we have in the distribution for @xmath9 , and of @xmath92 as an approximation to @xmath158 . for the projected residual , @xmath159 where @xmath121 follows a @xmath145 distribution with @xmath48 degrees of freedom .
this suggests setting @xmath160 a number of other suggestions for a projected discrepancy principle have been presented in the literature , but all imply using @xmath161 dependent on the noise level of the full problem , e.g. @xcite , with @xmath155 .
it is reported in @xcite , however , that while the theory predicts choosing @xmath155 , numerical experiments support reducing @xmath162 .
alternatively this may be seen as reducing the degrees of freedom , instead of reducing @xmath162 , consistent with .
we deduce that the interpretation for finding the regularization parameter based on the statistical property of the projected residual in contrast to the full residual should be important in determing the size of @xmath163 . for the mdp the degrees of freedom change from @xmath125 to @xmath48 when the residual is calculated on the full space as compared to the projected space .
thus @xmath77 is not a good approximation for @xmath164 when obtained using @xmath163 as a guide for the actual size of the projected residual . if the full problem is effectively singular , so that @xmath165 for @xmath135 , the degrees of freedom for the full problem are reduced and again @xmath143 . unlike the upre and mdp , the method of generalized cross validation ( gcv ) for finding the regularization parameter @xmath17 does not require any estimate of the noise level @xmath151 .
it is , however , a statistical technique based on leave one out validation and has been well - studied in the context of tikhonov regularization , @xcite .
the optimal parameter @xmath17 is found as the minimizer of the functional @xmath166 ignoring constant scaling of @xmath167 by @xmath168 .
the use of the gcv for finding the optimal parameter for the projected problem , as well for finding the subspace parameter @xmath44 , has also received attention in the literature , @xcite .
the obvious implementation for is the exact replacement in using the projected system @xmath169 as indicated in @xcite .
it was recognized in ( * ? ? ?
* section 5.4 ) , however , that this formulation , tends to lead to solutions which are over smoothed .
there it was suggested instead to use the weighted gcv dependent on parameter @xmath170 @xmath171 experiments illustrated that @xmath170 should be smaller for high noise cases , but in all cases @xmath172 is required to avoid the potential of a zero in the denominator . although the choice for @xmath170 is argued heuristically , and an adaptive algorithm to find @xmath170 is given , no theoretical analysis for finding an optimal @xmath170 is discussed .
moreover , there is apparently no study of the use of for projection of underdetermined problems .
consider now the two denominators in and .
first of all it is not difficult to show from , with the not very restrictive requirement @xmath173 , @xmath174 hence @xmath175 . thus picking @xmath17 to minimize the projected gcv
will not minimize the full gcv term .
for the weighted gcv , however , @xmath176 moreover , with @xmath141 for @xmath135 @xmath177 and factoring for @xmath178 and @xmath179 in and , respectively , gives the scaled denominators @xmath180 ignoring constant scaling the denominators are equilibrated by taking @xmath181 this result suggests that we need @xmath182 in order for @xmath77 to estimate @xmath79 found with respect to the projected space .
if @xmath44 is such that @xmath141 for @xmath135 , and @xmath183 then @xmath143 when obtained using the wgcv .
the estimate is found without projecting the solution back to the full space .
note that without reorthogonalization of the columns of @xmath69 and @xmath57 clustering of the singular values means we should not expect the equilibration of the denominators to yield the correct weighting parameter .
it is interesting that this reorthogonalization was regarded as less significant in @xcite , although it is clear from our discussion that it is useful for suggesting the choice @xmath184 .
all formulae apply using the svd for @xmath63 replacing that for matrix @xmath1 .
the upre functional is given by @xmath364 the mdp functional is given by @xmath365 for the projected case @xmath366 replaces @xmath146 .
using the svd for @xmath63 the wgcv functional is given by @xmath367 with @xmath368 this reduces to the expression for the projected gcv , .
to illustrate the discussion in section [ sec : parameter estimation ] we investigate the properties of the projected system matrices @xmath63 in the context of the solution of ill - posed one dimensional problems with known solutions . because the regularization parameter estimation techniques rely on the determination of the subspace @xmath44 and on the properties of the spectra of @xmath63 we look at the spectra and at methods to estimate @xmath44 . in all experiments we use matlab 2014b with examples from the regularization tool box @xcite and the code
` bidiag_gk ` associated with the software for the paper @xcite , for finding the factorization of matrix @xmath1 .
for a given problem defined by without noise , noisy data are obtained as @xmath185 for noise level @xmath186 and with @xmath187 the @xmath188 column of error matrix @xmath189 with columns sampled from a random normal distribution using the matlab function ` randn`@xmath190 .
examples presented here use the test problems ` phillips ` , ` shaw ` and ` gravity ` from the regularization toolbox , @xcite , all of which are discretizations of fredholm integral equations of the first kind . to show the impact of the ill - posedness of the problem we take problem ` phillips ` for which the picard condition does not hold ,
problem ` shaw ` which is severely ill - posed , and problem ` gravity ` that depends on a parameter @xmath191 determining the conditioning of the problem , here we use @xmath192 for severe ill - conditioning and @xmath193 which is better conditioned .
simulations for over and under sampled data are obtained by straightforward modification of the relevant functions in @xcite . in the examples we only show @xmath194 , @xmath195 illustrating the results for the under sampled cases , with @xmath196 samples . for the given method for calculating the noise level
we observe that the signal to noise ratio for the data , given by @xmath197 is independent of the test problem .
in particular @xmath198 and @xmath199 .
the condition of each problem depends on the condition of the matrix @xmath1 , see table [ conditiontable ] , and thus for the same choice of @xmath186 noise propagates differently for each test problem .
.condition of the test matrices [ cols="^,^,^,^,^",options="header " , ] these results are substantiated by consideration of figure [ figrelerrs ] .
it is well - known that the choice of estimator is non - trivial with different methods performing better under different conditions , these results do not contradict that conclusion .
overall we deduce that gcv may work well when the subspace is detected by minimizing and also for problems not satisfying the picard condition .
we consider two image deblurring problems , ` grain ` and ` satellite ` , both of size @xmath200 from restoretools @xcite .
restore tools also provides overloaded matrix operations for calculation of matrix vector products @xmath201 and @xmath202 , where @xmath1 describes a point spread function blurring operation .
the calculation of the factorization can be immediately obtained once the object @xmath1 is defined as a psf operator with the overloaded matrix operations . as indicated by the results for the one dimensional simulations it is important to note that all presented results use reorthogonalization in obtaining the factorization .
our main point here is to first demonstrate the use of the regularization techniques pmdp , wgcv and upre for increasing @xmath44 , and then to examine a stabilizing technique using an irr , section [ sec : irr ] .
results without irr are presented for completeness in section [ sec : lsqr ] and with irr in section [ sec : resirr ] .
for contrast with the results presented in @xcite we use noise levels @xmath203 and @xmath204 in which corresponds to noise levels @xmath205 and @xmath206 , respectively , in @xcite with @xmath207 , yielding @xmath208 . these correspond to bsnr @xmath209db and @xmath210db as calculated by . for immediate comparison with @xcite
we indicate the results using the noise level @xmath211 rather than @xmath186 . in figure [ fig2dexample ]
we give the true solution , blurred and noisy data and the point spread function .
one can observe in fact that even a noise level of @xmath212 is quite low , the main problem here being the blurring .
+ in finding the restorations for the data indicated in figure [ badgrain ] and [ badsatellite ] we note first that the matrices @xmath1 for the psfs indicated in figures [ psfgrain ] and [ psfsatellite ] do not satisfy the picard condition .
as illustrated in figure [ rhotwod ] , @xmath213 does not show the increase within the shown range of @xmath44 as is clear for @xmath213 with large @xmath44 obtained for the one dimensional examples in figure [ fig : rho ] .
still , @xmath213 for ` grain ` does attain a minimum within this range and then exhibits a gradual increase .
this suggests that noise is entering the data after the minimum and that one may use @xmath214 where again we advance @xmath215 steps under the assumption that noise enters after the minimum .
when the picard condition is not satisfied we may also use to find @xmath216 . in figure [ rhotwod ]
the vertical lines indicate the positions of @xmath217 , @xmath218 and @xmath219 . for ` grain `
, @xmath218 becomes quickly independent of the number of terms used , already stabilizing at @xmath220 with just @xmath221 terms , with no change even out to a maximum size of @xmath222 in the calculation . for ` satellite ` @xmath218 is less stable only reaching @xmath223 when @xmath224 terms are used in the estimation , but increasing to @xmath225 if @xmath222 terms are required .
stability in the choice of @xmath217 with respect to @xmath226 also follows lack of stability in choice of @xmath218 , suggesting that it is preferable to use @xmath219 . in our experiments
we have deduced that it is important to examine the characteristic shape of @xmath213 in determining the optimal choice for the size of the subspace , and will show results using @xmath219 , @xmath217 and @xmath218 .
the range for the regularization parameter is also important as is indicated through the windowing approach based on .
because there is no clear distinction between the singular values as they decay , we use a single window defined by @xmath227 and apply a filtered truncated svd for the solution which is dominant for the first @xmath227 terms , ie with filter factors @xmath228 for @xmath229 and @xmath230 for @xmath231 . with @xmath232 , @xmath233 . in our results
we use @xmath234 , @xmath235 , and impose @xmath236 , for the range of @xmath71 in finding the optimal regularization parameter for each of the regularization parameter techniques . to find the _ optimal _ solution , denoted by min in the legends , solutions are found using @xmath237 sampled at @xmath238 points logarithmically on this range and scaled by the mean of the standard deviation of the noise in the data , consistent with the inverse covariance scaling of the problems , and consistent with the range for the regularization parameter used in @xcite . in assessing the solutions we adopt the approach in @xcite and use both the relative error ( re ) and the mean structured similarity ( mssm ) of the obtained images , @xcite , for which larger values correspond to better solutions .
the re and mssm for all cases are illustrated in figure [ fig2derrorsstep1 ] .
the results are consistent with the literature in terms of the semi convergence behavior of the lsqr and the difficulty for both the mdp and the gcv in estimating a useful regularization parameter . on the other hand ,
the results with wgcv , pdmd and upre are consistent with each other and verify the analysis in section [ sec : regest ] , providing a stable solution for increasing @xmath44 .
the solutions do not achieve the minimal error of the projected solution without regularization , which depends on knowing the optimal @xmath44 for stability .
solutions found at the noted @xmath239 as compared to the _ optimal _ solution with minimum error are illustrated in figure [ fig2dsolutionsgrain_step1 ] , demonstrating that the restorations are inadequate at this level of noise .
+ iteratively reweighted regularization provides a cost effective approach for sharpening images e.g. @xcite , and has been introduced and applied for focusing geophysical inversion , in this context denoted as minimum support regularization , @xcite .
regularization operator @xmath15 is replaced by a solution dependent operator @xmath240 , initialized with @xmath241 , yielding iterative solution @xmath242 . for @xmath243 @xmath244 where @xmath245 is a focusing parameter which assures that @xmath240 is invertible .
immediately @xmath246 and @xmath247 moves entries for which @xmath248 away from @xmath249 .
given that @xmath240 is invertible , we can use with system matrix @xmath250 , to obtain the iterative solution @xmath251 , @xmath243 .
furthermore , it is straightforward to modify the algorithm for calculation of the gkb factorization for matrix @xmath252 still using the overloaded matrix operations for matrix multiplication by @xmath1 and @xmath253 , and noting that operations with the diagonal matrix are simple component - wise products .
the update given by is equivalent to regularizing the system of equations , here dropping the dependence on iteration @xmath254 , @xmath255 suppose that @xmath248 , for @xmath256 and @xmath257 , then @xmath258 for @xmath256 and we may solve the reduced system @xmath259 where @xmath260 is @xmath261 but with column @xmath22 removed for @xmath256 and all other columns scaled by the relevant diagonal entries from @xmath262 .
matrix @xmath260 is of size @xmath263 where @xmath264 and vector @xmath265 is vector @xmath266 with entries @xmath256 removed .
we can then solve for @xmath265 in @xmath267 yielding the update @xmath268 , where @xmath269 is obtained from @xmath15 with the same diagonal entries @xmath256 removed .
the update for @xmath5 is therefore obtained using with entries @xmath270 , for @xmath271 and @xmath272 , for @xmath273 .
therefore , to avoid any need to discuss the choice of @xmath245 , we simply use @xmath257 and obtain the gkb factorization for the reduced system with system matrix @xmath260 .
the approach for the iteration requires some explanation as to how the range of @xmath44 is obtained at irr iterations @xmath243 , and requires consideration of @xmath274 , which depends on the subspace size @xmath275 from the prior step @xmath254 , and current subspace size @xmath44 .
thus calculation of @xmath217 , @xmath219 and @xmath218 are all dependent on @xmath275 as well as @xmath276 and @xmath277 , ie given a specific subspace size at @xmath275 the minimum and maximum sizes to use at step @xmath254 need to be specified . because the update costs for irr should be kept minimal , the subspace size is maintained less that @xmath239 from the previous step , i.e. we pick @xmath278 . because we anticipate that further noise enters with increasing @xmath254 , we expect @xmath279 , which can be determined by examination of @xmath280
. we will examine the choices for the case with @xmath206 noise .
for each iteration the range for @xmath71 is constrained using the current singular values , by @xmath281 , where at step @xmath249 , @xmath282 and @xmath283 for the irr updates .
having already noted that the pmdp , wgcv and upre give consistent solutions , while gcv and mdp generally lead to larger errors , the experiments reported with irr are given for the projected , _ optimal _ and upre solutions only . to demonstrate the effectiveness of the irr for stabilizing the solution obtained using tikhonov regularization with upre and to describe a manual approach for determining the sizes of the optimal subspaces we examine the process first for problem ` grain ` and then ` satellite ` with @xmath206 noise .
function @xmath213 at the first step @xmath284 does not differ significantly from the case with @xmath212 noise , shown in figure [ rhotwod ] .
we reiterate that the calculation of @xmath216 depends on the maximum subspace considered , here we use @xmath285 .
because the update costs for irr should be kept minimal , the subspace size is maintained less that @xmath239 from the first step , here @xmath286 .
figure [ grainrhostepk ] shows @xmath280 for the choices of @xmath44 in the legend .
it is clear that @xmath280 is almost independent of @xmath275 for the first steps , but that noise enters for @xmath287 .
this is also reflected in the re in figure [ grainre ] , the re stabilizes for increasing @xmath44 , and decreases for the first three steps of irr , but increases at step @xmath288 .
cropped images are shown in figure [ grainstep3]-[grainstep4 ] . at @xmath289
the contrast is increased and some small features not present without the irr become apparent .
an equivalent process is detailed for problem ` satellite ` with @xmath206 noise , and for maximum subspace @xmath290 .
using @xmath291 gives a choice very close to @xmath218 , but @xmath292 depends very much on identifying which peaks should be included , here we imposed @xmath293 .
the plot of @xmath213 with increasing @xmath44 in figure [ satelliterhostepk ] show that noise enters the solution sooner for smaller @xmath294 , but that the various choices for @xmath239 demonstrate similar characteristics , and suggest that no more than @xmath215 steps of irr may be needed . from these plots of @xmath213
we select @xmath295 , yielding a subspace selection of @xmath296 for the first update and in subsequent updates with @xmath297 the subspace is maintained small just of size @xmath288 .
the relative errors for the two simulations are detailed in table [ tab : errortwodfive ] , in comparison to the _ optimal _ relative error .
these results show that the stabilization leads to results which are comparable to those that are _ optimal _ but practically unknown .
moreover , it is clear that one may not conclude that finding @xmath239 using @xmath219 is preferable to using @xmath218 or @xmath217 .
provided that the solutions are stabilized with the irr , improvements in the solutions are obtained in a limited number of steps using iterative reweighting .
further , effective irr steps can be obtained using relatively small subspaces for the iterative updates .
iteration & @xmath291 & @xmath218 & min for upre&overall min + @xmath42&@xmath298 & @xmath299 & @xmath300 & @xmath301 + @xmath215&@xmath302 & @xmath303 & @xmath304 & @xmath305 + @xmath306&@xmath307 & @xmath308 & @xmath309 & @xmath310 + @xmath288&@xmath311 & @xmath312 & @xmath313 & @xmath314 + & + iteration & @xmath292 & @xmath216 & min for upre&overall min + @xmath42 & @xmath315 & @xmath316 & @xmath317 & @xmath318 + @xmath215&@xmath319 & @xmath320 & @xmath321 & @xmath322 + @xmath306&@xmath323 & @xmath324 & @xmath325 & @xmath326 + @xmath288&@xmath327 & @xmath328 & @xmath329 & @xmath326 + the solutions at the first step are illustrated in figure [ fig : twodfive ] , demonstrating that using @xmath217 leads to best solutions in one case , and @xmath218 in the other , although effectively the quality is comparable .
the graphs of @xmath213 with increasing @xmath254 in figures [ grainrhostepk ] - [ satelliterhostepk ] indicate that the properties of @xmath280 can be used to determine effective termination of the irr , based on the iteration @xmath254 when noise enters into @xmath280 .
our experience has shown that the optimal solution in terms of image quality is achieved not at the step before noise enters in @xmath280 but two steps before . in figure [ fig2derrorsgrain ]
we contrast the relative errors obtained by the upre , projected and _ optimal _
solutions for two steps of irr as compared to the first step @xmath284 .
then in figure [ fig2dmeasuresgrain ] we illustrate how the re and the mssm change with the iteration count .
the vertical lines in figures [ grainreten]-[satelliessim ] demonstrate that using @xmath217 or @xmath218 makes little difference to the quality of the solution when measured with respect to relative error or mssm .
example solutions are given in figure [ fig2dsolutionsgrain_irrstep ] . + to contrast the success of the regularization parameter estimation techniques in the context of a @xmath215d projection problem , we present results for the reconstruction of projection data obtained from tomographic x ray data of a walnut , used for edge preserving reconstruction in @xcite and with the description of the data described in @xcite .
the data are available at @xcite .
datasets ` datan ` correspond to resolution @xmath330 in the image , and use @xmath331 projections , corresponding to @xmath332 sampling .
data are provided with @xmath333 , @xmath334 and @xmath335 .
we use resolution @xmath334 with @xmath331 projections , and then downsampled to @xmath336 projections , @xmath337 projections and @xmath338 projections , ie angles @xmath332 , @xmath339 , @xmath340 and @xmath341 .
( results with resolutions @xmath342 and @xmath335 are comparable ) .
results are presented using the solution at @xmath217 for the projected solution without regularization , and regularized using upre and gcv for comparison , figures [ walnut16460]-[walnut16415 ] , where we do not show results for @xmath331 projections , for which the solutions are almost perfect due to the apparent limited noise in the provided data .
the impact of using a reduced number of projections is first evident with just @xmath336 projections . in all the presented results the parameters @xmath217 are determined automatically , after manually picking @xmath343 from manual consideration of the plot for @xmath213 , see figure [ walnutphi ] .
all other parameters are estimated in the same way as for the image restoration cases .
@xmath213 for increasing sparsity for resolution @xmath334 for the walnut data . ] the results in figure [ walnut16460]-[walnut16415 ] , which show results for one set of data at increasing sparsity , compare with ( * ? ? ?
* figure 6.6 and figure 6.7 ) , which give results with resolution for @xmath344 and @xmath345 , respectively , and angle separation @xmath346 , @xmath347 , @xmath339 and @xmath340 .
results there use selected choices for the regularization parameter based on a sparsity argument with prior information and seek to support the use of the sparsity argument for reconstruction of sparse data sets , although exhibiting the rather standard total variation blocky structures when applied for truly sparsely sampled data .
our results show robust reconstructions with the automatically determined solutions , after first examining the plot for @xmath213 .
irr generates marginally improvements in qualitative solutions , more so for the projected case without regularization . to show the impact of the correct choice of @xmath239 on the solution we show a set of results at iteration @xmath284 using @xmath348 in figure [ walnut164tmin18 ] , with the positive constraint .
there the projected solution is already significantly noise contaminated for all levels of sparsity , while the upre yields solutions qualitatively similar to the case with @xmath343 . in the examples here
we do impose an additional positivity constraint on the solutions at each step , before calculating the iterative weighting matrix .
the results demonstrate that the projected problem with automatic determination of @xmath77 can be used to reconstruct sparsely sampled tomographic data , provided that an initial estimate for @xmath226 is manually determined by consideration of the plot of @xmath213 .
further , irr can stabilize the solution when @xmath239 has not been appropriately estimated for the non regularized solution .
for the sparse data sets the solutions do not exhibit the characteristic blocky reconstructions of total variation image reconstructions , as seen in @xcite , although as there the solutions would be inadequate for precise usage .
+ + + + + + + + + + +
we have demonstrated that regularization parameter estimation by the method of upre can be effectively applied for regularizing the projected problem .
our results also explain the use of the weighting parameter in the wgcv , as well as the reduced safety parameter in the mdp when applied for the projected problem .
further , edge preserving regularization via the iteratively reweighted regularizer can be applied to stabilize regularized solutions of the projected problem .
our results suggest manual estimation of a minimal subspace size can then lead to useful estimates for an optimal projected space , with the use of the irr leading to improvements in the solutions when @xmath239 is found by different methods , including the use of @xmath217 , @xmath219 and @xmath218 , hence making the determination of this @xmath239 less crucial in providing an acceptable solution
. future work on this topic should include extending the windowed regularization parameter techniques for finding a multiply weighted gcv for projected regularization , and use of more general iteratively reweighted regularizers accounting for edges in more than one direction in conjunction with the projected solutions .
these are topics for further study .
suppose the svd of matrix @xmath1 , @xmath349 , is given by @xmath350 , where the singular values are ordered @xmath351 and occur on the diagonal of @xmath352 with @xmath353 zero columns ( when @xmath354 ) or @xmath355 zero rows ( when @xmath356 ) , and @xmath357 , and @xmath358 are orthogonal matrices , @xcite .
then @xmath359 for the projected case @xmath360 , i.e. @xmath356 , and the expression still applies with @xmath361 replacing @xmath4 , @xmath71 replacing @xmath17 , @xmath362 replacing @xmath64 and @xmath363 in .
00 bjrck a 1986 _ numerical methods for least squares problems _ society for industrial and applied mathematics philadelphia pa chung j m easley g and oleary d p 2011 windowed spectral regularization of inverse problems _ siam j. sci .
comput . _ * 33 * 6 3175 - 3200 chung j m nagy j and oleary d p 2008 a weighted gcv method for lanczos hybrid regularization _ etna _ ,
* 28 * , 149 - 167 chung j m kilmer m e and oleary d p 2015 a framework for regularization via operator approximation _
siam j. sci .
* 37 * 2 b332-b359 donatelli m and reichel l 2014 square smoothing regularization matrices with accurate boundary conditions _ j computational and applied mathematics _
* 272 * 334 - 349 fong d c - l and saunders m a 2011 lsmr : an iterative algorithm for sparse least - squares problems _
siam j. sci .
comput . _ * 33 * 5 , 2950 - 2971 golub g h heath m and wahba g 1979 generalized cross validation as a method for choosing a good ridge parameter
_ technometrics _ * 21 * 2 215 - 223 .
golub g h and van loan c 1996 _ matrix computations _ john hopkins press baltimore 3rd ed .
hmlinen k harhanen l kallonen a kujanp a niemi e and siltanen s 2015 tomographic x - ray data of a walnut arxiv:1502.04064v1 , http://www.fips.fi/dataset.php .
hmlinen k kallonen a kolehmainen v lassas m niinimki k and siltanen s 2013 sparse tomography , _ siam journal of scientific computing _ * 35 * 3 , b644- b665 hanke m and hansen p c 1993 regularization methods for large scale problems _ surveys math .
indust . _ * 3 * 253 - 315 hansen p c 1998 _ rank - deficient and discrete ill - posed problems : numerical aspects of linear inversion _
siam monographs on mathematical modeling and computation * 4 * philadelphia hansen p c 2007 regularization tools : a matlab package for analysis and solution of discrete ill - posed problems version 4.0 for matlab 7.3 , _ numerical algorithms _ * 46 * , 189 - 194 , and http://www2.imm.dtu.dk/~pcha/regutools/ hansen p c and jensen t k 2008 noise propagation in regularizing iterations for image deblurring _ etna _ * 31 * 204 - 220 hansen p c nagy j g and oleary d p 2006 _ deblurring images matrices spectra and filtering _
siam philadelphia hntynkov i plesinger m and strakos , z 2009 the regularizing effect of the golub - kahan iterative bidiagonalization and revealing the noise level in the data _ bit numerical mathematics _ *
49 * 4 669 - 696 hochstenbach m e and reichel l 2010 an iterative method for tikhonov regularization with general linear regularization operator _
j. integral equations appl . _
* 22 * 463 - 480 kilmer m e and oleary d p 2001 choosing regularization parameters in iterative methods for ill - posed problems _ siam journal on matrix analysis and applications _ * 22 * 1204 - 1221 morozov v a 1966 on the solution of functional equations by the method of regularization _ sov .
dokl . _ * 7 * 414 - 417 nagy j g palmer k and perrone l 2004 iterative methods for image deblurring : an object oriented approach _ numerical algorithms _ * 36 * 73 - 93 neelamani r choi h and baraniuk r g 2004 forward : fourier - wavelet regularized deconvolution for ill - conditioned systems _ ieee transactions on signal processing _ * 52 * 2 418 - 433 paige c c and saunders m a 1981 towards a generalized singular value decomposition _
siam journal on numerical analysis _ * 18 * 3 398 - 405 paige c c and saunders m a 1982 lsqr : an algorithm for sparse linear equations and sparse least squares _ acm trans .
math . software _
* 8 * 43 - 71 paige c c and saunders m a 1982 algorithm 583 lsqr : sparse linear equations and least squares problems _ acm trans .
math . software _ * 8 * 195 - 209 portniaguine o and zhdanov m s 1999 focusing geophysical inversion images _ geophysics _ * 64 * 874 - 887 paoletti v hansen p c hansen m f and maurizio f 2014 a computationally efficient tool for assessing the depth resolution in large - scale potential - field inversion _ geophysics _ * 79 * 4 a33a38 reichel l sgallari f and ye q 2012 tikhonov regularization based on generalized krylov subspace methods _ appl . numer .
_ , * 62 * 1215 - 1228 renaut r a hnetynkov i and mead j l 2010 regularization parameter estimation for large scale tikhonov regularization using a priori information _ computational statistics and data analysis _ * 54 * 12 3430 - 3445 doi:10.1016/j.csda.2009.05.026 vatankhah s ardestani v e and renaut r a 2014 automatic estimation of the regularization parameter in 2-d focusing gravity inversion : application of the method to the safo manganese mine in the northwest of iran _ journal of geophysics and engineering _ * * * 11 * 045001 vatankhah s ardestani v e and renaut r a 2015 application of the @xmath145 principle and unbiased predictive risk estimator for determining the regularization parameter in 3-d focusing gravity inversion _ geophysical j international _ * 200 * 265 - 277 doi : 10.1093/gji / ggu397 vatankhah s renaut r a and ardestani v e 2014 regularization parameter estimation for underdetermined problems by the @xmath145 principle with application to 2d focusing gravity inversion _ inverse problems _ * 30 * 085002 vogel c r 2002 _ computational methods for inverse problems _ siam frontiers in applied mathematics siam philadelphia u.s.a
. wohlberg b and rodriguez p 2007 an iteratively reweighted norm algorithm for minimization of total variation functionals _ ieee signal processing letters _ * 14 * 948951 wang z bovik a c sheikh h r simoncelli e p 2004 image quality assessment : from error visibility to structural similarity _ ieee trans . image process .
_ * 13 * 600 - 612 www.cns.nyu.edu/~lcv/ssim zhdanov m s 2002 _ geophysical inverse theory and regularization problems _ elsevier amsterdam . | tikhonov regularization for projected solutions of large - scale ill - posed problems is considered .
the golub - kahan iterative bidiagonalization is used to project the problem onto a subspace and regularization then applied to find a subspace approximation to the full problem .
determination of the regularization parameter using the method of unbiased predictive risk estimation is considered and contrasted with the generalized cross validation and discrepancy principle techniques .
examining the unbiased predictive risk estimator for the projected problem , it is shown that the obtained regularized parameter provides a good estimate for that to be used for the full problem with the solution found on the projected space .
the connection between regularization for full and projected systems for the discrepancy and generalized cross validation estimators is also discussed and an argument for the weight parameter in the weighted generalized cross validation approach is provided .
all results are independent of whether systems are over or underdetermined , the latter of which has not been considered in discussions of regularization parameter estimation for projected systems . numerical simulations for standard one dimensional test problems and two dimensional data for both image restoration and tomographic image reconstruction support the analysis and validate the techniques .
the size of the projected problem is found using an extension of a noise revealing function for the projected problem .
furthermore , an iteratively reweighted regularization approach for edge preserving regularization is extended for projected systems , providing stabilization of the solutions of the projected systems with respect to the determination of the size of the projected subspace .
* keywords : * large - scale inverse problems , golub - kahan bidiagonalization , regularization parameter estimation , unbiased predictive risk estimator , discrepancy principle , generalized cross validation , iteratively reweighted schemes |
unlike the cuprates , where the magnetic state owes its existence to mott insulator physics , in iron - based superconductors magnetism results from an instability of the delocalised fe @xmath0-band electrons which gives rise to a spin - density wave @xcite .
the pnictide parent compounds display metallic , antiferromagnetic spin - density wave ground states where the spins are periodically modulated in space but where the outermost electrons can be delocalised , typical of the collective effect that emerges from an instability of the paramagnetic fermi surface .
a key question remains how the parent magnetic state in pnictides evolves across the phase diagram and how the properties of the doped magnet can compete , promote or coexist with superconductivity .
the simplest of all iron arsenide systems , the monoarsenide feas , may provide some insights into these questions since , it has been argued @xcite , its itinerant magnetism is related to the magnetic ground states of iron - based superconductors .
feas crystallises in the b31 ( mnp - type ) structure ( space group @xmath1 ) @xcite which consists of distorted feas@xmath2 octahedra which are face sharing along the @xmath3-axis and edge sharing along the @xmath4- and @xmath5-axes ( fig .
[ fig : feas : cryst ] ) .
it therefore has similar fe - fe linkages to the layered fe - based superconductors ( such as lafeaso , bafe@xmath6as@xmath6 and nafeas ) , but is distinguished from them by being surrounded by six ( octahedral ) rather than four ( tetrahedral ) arsenic atoms .
the iron atoms sit at the _ 4c _ wyckoff site , giving rise to four positions in the unit cell : fe@xmath7 at@xcite ( _ x _ , @xmath8 , _ z _ ) , fe@xmath6 at ( _ _ @xmath9__+@xmath10 , @xmath11,__z__+@xmath10 ) , fe@xmath12 at ( _ @xmath9 _ , @xmath11 , _ @xmath13 _ ) and fe@xmath14 at ( _ _ x__+@xmath10 , @xmath8 , _ _
@xmath13__+@xmath15 , where @xmath16 and @xmath17 as shown in fig .
[ fig : robert1 ] .
initial neutron powder diffraction measurements @xcite showed that feas undergoes a transition to a long - range antiferromagnetically ordered state at @xmath18 k. it was suggested that the system adopts a helical magnetic structure with a wavevector @xmath19 $ ] and an ordered magnetic moment of 0.5 @xmath20 .
however , a more recent magnetic susceptibility and transport study @xcite , gave results that indicated the presence of significant magnetocrystalline anisotropy , raising doubts about the occurrence of such a simple spin helix structure .
specifically , single crystal susceptibility shows a kink at 70 k in the @xmath3- and @xmath4-directions but not in the @xmath5-direction .
( the lack of features in the @xmath5-axis direction suggesting the magnetic moment is fixed in the @xmath3-@xmath4 plane . )
the key observation is that the susceptibility along the @xmath4-axis is lower than that along @xmath3 , and only the @xmath4-axis displays a magnetic field splitting , suggesting the presence of anisotropy . following this , a polarised single crystal neutron diffraction study @xcite suggested a slightly elliptical helical structure comprising a non - collinear spin - density wave arising from a combination of itinerant and localised behaviour , with the spin amplitude along the @xmath4-axis direction being 1.5(5)% larger than that along the @xmath3-direction . in terms of its electronic properties ,
feas lies between two well - understood regimes : the delocalized magnetic metal and the localized magnetic insulator .
resistivity measurements,@xcite confirm the itinerant behaviour of feas : resistivity decreases below 150 k with a kink observed at 70 k. electronic structure calculations on feas have not fully elucidated the mechanism that generates the magnetic structure .
non - polarized , collinear and non - collinear spin calculations have been carried out , but find the lowest energy state to be antiferromagnetic , in which nearest - neighbour iron spins antialign with a resultant @xmath21 symmetry @xcite .
the study by parker and mazin @xcite calculated the static lindhard function in feas and the nesting of the fermi surface in the afm phase , and concluded that some form of nesting did not drive the magnetic order .
griffin and spaldin @xcite compared the use of different dft functionals in feas , and found that gga / hybrid gga gives values for the structure in closest agreement with experiment , but that a negative hubbard-@xmath22 calculation would be most likely to reproduce the spin spiral as it would increase competition between afm and fm interactions and increase the energy of the afm state .
however , there is no other physical justification for a negative hubbard-@xmath22 parameter , implying a larger failing in these functionals .
griffin and spaldin also performed non - collinear spin calculations imposing a variety of spirals on the system , but found that the afm state was lower in energy than all of them .
in this paper we present a refinement of the magnetic structure of iron arsenide using x - ray resonant elastic scattering ( xres ) and calculations of the electronic structure using density functional theory ( dft ) . from analysis of a magnetic satellite reflection ,
our xres results strongly suggest that the magnetic spiral is considerably more elliptical than was previously believed , has a right - handed chirality and has an ordered spin component in the propagation direction of the helix .
dft calculations show that spin - orbit interactions and the local iron environment provide an explanation for this new ordering component .
we conclude that the spin ordering is linked to localized orbital restructuring and changes in electronic density , and therefore , as might be expected , is not well described by simple stoner or ising - type models .
our sample of feas was grown by an iodine vapour transport method@xcite .
the growth method resulted in single crystals of typical dimensions @xmath23 m .
several samples were characterized using a four - circle diffractometer , and a single crystal selected , with a natural _ c_-axis facet and a sharp @xmath24 $ ] reflection with a rocking width of just @xmath25 .
xres measurements were carried out at both the soft fe @xmath26 and the hard fe _ k _ absorption edges .
for the fe _
edges experiments the beamlines id08 ( at esrf ) and i10 ( at diamond ) were used . for
the _ k _ edge experiments the beamline p09 ( at petra iii ) was used .
all three beamlines are situated on an undulator insertion device . for the id08 and i10 experiments
the sample was mounted with the @xmath4-axis in the scatter plane .
for the p09 experiment the @xmath27 $ ] reciprocal direction was used as the azimuthal reference vector .
dft calculations were run with the castep electronic structure code using the pbe exchange - correlation functional@xcite@xcite .
energy differences between spin configurations were converged to 1 part in 10,000 , and to generate the fermi - surface a monkhorst - pack @xmath28-point grid of @xmath29 was used . to account for core state contributions on atoms an ultrasoft core - corrected iron pseudopotential with 8 valence electrons and an arsenic pseudopotential with 15 valence electrons were used .
a non - magnetic configuration and a range of collinear ordered spin - structures were considered .
at the fe @xmath30 energy ( @xmath31 ev ) the radius of the ewald sphere limits access along the @xmath32 reciprocal direction to @xmath33 . within this limit
two resonant reflections were found at positions @xmath34 and @xmath35 .
a scan along the [ 00@xmath36 direction is shown in fig . [
fig : feas : l3lscan ] .
the observed peaks are asymmetric ( most likely due to the energy profile of the undulator ) , with the peak at @xmath37 having the reverse asymmetry to the @xmath34 reflection .
this suggests that [ 0,0,0.611 ] is a satellite of the forbidden @xmath38 $ ] bragg peak .
the two reflections can be indexed as @xmath39+[0,0,\tau]$ ] ( denoted @xmath40 hereafter ) , and @xmath38-[0,0,\tau]$ ] ( denoted @xmath41 hereafter ) , where @xmath42 .
energy resonances of the two reflections , without post - scatter polarisation analysis ( fig .
[ fig : feas : linenergysoft ] ) , were performed by decreasing the energy of the incident x - ray whilst maintaining the diffraction condition for the magnetic peak .
the resonances were measured with both @xmath43- and @xmath44-polarised incident light .
assuming a dipole - dipole transition ( @xmath45 ) is responsible for the resonant feature , the transition is from the fe 2@xmath46 orbital to the fe 3@xmath0 band .
exciting into this fe 3@xmath0 band leads to the sensitivity of the technique to the local magnetism , as it is the @xmath47 orbitals that are the magnetically active spin - polarised band in iron .
we also note that the temperature behaviour of the [ 0,0,@xmath40 ] magnetic bragg peak shows critical behaviour consistant with that observed previously@xcite .
the @xmath40 reflection shows a marked difference between the two polarization channels , which is sufficient to rule out either charge scattering or simple collinear magnetic spin structures along the @xmath3- , @xmath4- , or @xmath5-directions , assuming a @xmath45 origin to the scattering .
the @xmath41 reflection shows very different behaviour , giving equal intensity with incident @xmath43- and @xmath44-polarised light ( fig .
[ fig : feas : linenergysoft ] ) .
this indicates a different origin for the two peaks .
the @xmath40 and @xmath41 reflections occur at @xmath49 angles of @xmath50 and @xmath51 , respectively .
these angles are not close to @xmath52 or @xmath53 , which might cause a suppression of scattering due to the @xmath49 dependences of the scattering amplitude . for the @xmath40 reflection
the intensity in the circular - positive channel is roughly twice that of the circular - negative channel , indicative of a noncollinear spin structure .
the @xmath41 reflection has equal intensity in the circular - positive and circular - negative channels .
this again shows very different behaviour to the @xmath40 peak . a full linear polarisation analysis ( flpa )
was also carried out on both peaks using the ultra - high vacuum diffractometer , rasor at i10 , and the results are shown in fig .
[ fig : feas : softpolanalysis ] . in this measurement
the incident linear light is rotated through a full 180@xmath54 , and at each incident polarisation angle , the polarisation state of the scattered beam is measured .
[ fig : feas : softpolanalysis ] shows the incident polarisation angle against the outgoing polarisation using poincar - stokes parameters , @xmath55 and @xmath56 .
the results show a different polarisation analysis for the @xmath40 and @xmath41 reflections , confirming they have different origins .
xres measurents at the hard x - ray energy of the fe _ k _ absorption edge allow for a wider field of access to reciprocal space than measurements at the soft energies , but at a cost of sensitivity to the magnetism . a typical @xmath45 transition at the fe _ k _ absorption edge , excites a fe @xmath57 electron into the empty fe @xmath58 band .
the sensitivity to magnetism arises from any overlap , or hybridisation , between the fe @xmath47 and fe @xmath58 bands .
a survey of resonant reflections was carried out and satellite reflections were found at @xmath59 $ ] , @xmath60 $ ] , @xmath61 $ ] , @xmath62 $ ] , as well as at @xmath63 $ ] , and @xmath64 $ ] .
an off - axis reflection was also observed at the @xmath65 $ ] position .
no reflections were found at positions away from the odd forbidden bragg peaks ( @xmath66 $ ] and @xmath67 $ ] ) .
figure [ fig : feas : restau ] shows the resonances and reciprocal space scans of the @xmath59 $ ] peak .
all types of satellite reflection show a sharp resonant feature at 7110 ev .
the reciprocal space scans show the @xmath59 $ ] peak to be the sharpest with a width of @xmath68 r.l.u .
, while the @xmath69 and @xmath70 reflections are wider with widths of @xmath71 and @xmath72 r.l.u .
the @xmath40 and @xmath69 reflections were found only in the @xmath73 channel and not in the @xmath74 channel , while the @xmath70 reflection was found in both channels but was stronger in the @xmath74 channel .
an @xmath45-type transition can produce @xmath40 and @xmath69 reflections , but a quadrupole - quadrupole type transition ( e2e2 , involving an excitation from the @xmath57 orbital into the magnetically active @xmath47 spin - polarised band ) is required to explain the presence of a @xmath70 reflection .
an azimuthal measurement was performed on the @xmath59 $ ] reflection ( fig .
[ fig : feas : azimuthp09 ] ) which involves a rotation of the sample around the scattering vector , maintaining the diffraction condition .
the zero point on the azimuthal axis is defined as when the @xmath27 $ ] reciprocal vector is in the scattering plane away from the incident beam .
qualitatively , these azimuth data rules out a simple non - elliptical helix . for the @xmath59 $ ] reflection ,
the scattering vector is parallel to the magnetic propagation direction , an azimuthal measurement rotates around the magnetic propagation vector .
if the magnetic helix was circular , then there would be no change in moment direction upon an azimuthal rotation , and constant intensity would be expected .
the observation of not just the [ 0,0 , @xmath40 ] but also the [ 0,0 , 2@xmath40 ] and [ 0,0,3@xmath40 ] reflections is very reminiscent of x - ray scattering studies of chromium metal .
chromium is the canonical example of an itinerant , incommensurate antiferromagnet below its nel temperature @xmath75 k @xcite . this transverse spin density wave ( sdw )
is well understood and arises from a nesting effect between the 3@xmath0 electron pocket centered at the @xmath76 point and the hole pocket centered at the edge of the brillouin zone at the @xmath77 point .
the pairing is between momentum states separated by wavevector @xmath78 and a spiral density wave is formed . the ground state is then formed from two spiral waves of opposite helicity resulting in a long period , linearly polarised sdw .
associated with the sdw ordering , there is a distortion of the lattice with twice the wavevector of the sdw , causing a charge density wave ( cdw ) .
this results in differing satellites surrounding the bragg peaks with odd multiples of @xmath78 being magnetic satellites resulting from the sdw and even multiples of @xmath78 caused by the cdw .
thus @xmath79 and @xmath80 magnetic satellites were first observed by neutron diffraction @xcite and @xmath81 and @xmath82 charge satellites observed by x - ray scattering @xcite .
the observation of the fourth harmonic suggests that the cdw is not perfectly sinusoidal .
it is not yet clear as to the exact mechanism for producing a density wave in the charge distribution , with both a magnetostriction effect ( coupling the elasticity to the magnetism ) @xcite , or a purely electronic effect based on nesting between electronic bands @xcite , being claimed .
this is similar , but slightly different to our observations , suggesting that in feas the @xmath40 and @xmath83 satellites are magnetic in character but the @xmath84 satellite has both magnetic and charge characteristics .
further studies of the temperature dependence of these satellites may help unravel their origin .
to determine the magnetic structure of the material , we consider an elliptical magnetic helix pointing along the @xmath5-axis with spin components @xmath85 and @xmath86 in the @xmath3- and @xmath4-directions respectively and a chirality @xmath87 . the magnetic moment on the @xmath88th atom in the @xmath89th unit cell for a spin helix with propagation vector @xmath90 is given by @xmath91 where @xmath92 is the phase shift caused by the orbit of fe@xmath93 .
for such a helix propagating along the @xmath94 $ ] direction , satellite peaks only appear around even bragg reflections with structure factor @xmath95 where @xmath96 is a polarization factor , @xmath97 , @xmath98 is the fractional coordinate of fe@xmath7 and @xmath32 is a miller index .
we fit the ellipticity to the flpa analysis of the @xmath99 $ ] reflection in fig .
[ fig : feas : softpolanalysis ] ( top ) and the azimuthal measurement of the @xmath59 $ ] reflection shown in fig .
[ fig : feas : azimuthp09 ] , which shows the predicted azimuthal dependences for perfectly circular helical structure , and elliptical structures .
a circular structure gives a constant intensity as a function of azimuthal angle , as expected . as the structure is made more elliptical , the azimuth changes from a constant to a sinusoidally - changing intensity .
if the magnetic structure has the long axis of the ellipse pointing down the @xmath4-axis , then the azimuth goes through a minimum at 180@xmath54 .
however , we find that the azimuth of the @xmath59 $ ] reflection goes through a minimum at around 157@xmath54 . in our magnetisation model
, the azimuth intensity can only go to zero at 0@xmath54and 180@xmath54 or 90@xmath54 and 270@xmath54 depending on whether the long axis of the ellipse is along the @xmath3- or @xmath4-axis . to account for this ,
the ellipse is allowed to rotate such that long- and short - axes no longer point along the @xmath3- and @xmath4-axes . in order to fit the azimuth of the @xmath59 $ ]
, the long axis of the ellipse has to be placed along the _ b_-axis and rotated by @xmath100 , as shown in fig .
[ fig : feas : azimuthp09 ] .
as the magnetic structure is elliptical rather than circular , the form of the flpa of the @xmath99 $ ] should be highly dependent on the azimuth at which the measurement is taken . in the flpa of the @xmath99 $ ] reflection
, the rotation of the ellipse has a similar effect to changing the azimuth position at which the calculation is performed .
the effect of ellipticity on the flpa measurement is quite dramatic , and is shown in fig .
[ fig : feas : softpolanalysis ] . for example , for a circular magnetic structure , the predicted @xmath55 remains negative for all incident angles .
the flpa does not require a rotation of the magnetic ellipse to fit the data , but the rotation can be accommodated by correcting for a potential offset in the azimuth position .
combining both the azimuth of the @xmath59 $ ] reflection and the flpa of the @xmath99 $ ] reflection allows a fit requiring only three parameters , the ellipticity , @xmath101 , the rotation , @xmath102 and the azimuth offset of the polarisation analysis measurement @xmath103 .
the final fit results in an ellipticity of @xmath104 , which is far more substantial than the ellipticity of 1.15 proposed by the neutron experiment of rodriguez _ _ et al.__@xcite .
the azimuthal measurement suggests a rotation of the ellipse of @xmath105 .
( as a consequence the azimuthal position of flpa needs to be @xmath106 .
this value of the azimuthal offset from the _ b_-axis is within the experimental uncertainty of mounting the sample . )
the differences between our results and those derived from the neutron experiments @xcite probably result from the very different techniques used to estimate the magnitude of the ellipticity .
the study of rodriguez _ et al . _
used the intensity profiles of 16 relatively low intensity nuclear and magnetic reflections from polarised neutron diffraction measurements .
our study however has used the intensity variation of the azimuthal dependence of a resonantly enhanced x - ray magnetic satellite as well as the full linear polarised analysis of a separate magnetic satellite , which were combined and fitted with a model involving just three adjustable parameters .
the @xmath107 $ ] reflection is predicted to have zero intensity from the above structure factor . in order to explain the origin of this reflection ,
a spin helix along the @xmath5-axis with moments restricted to lie within the @xmath3-@xmath4 plane is not sufficient .
the phase difference brought about by the two - orbit structure is the origin for the predicted extinction of the @xmath107 $ ] reflection , and not the direction of the magnetisation vector .
this means that changing the magnetic structure to a cycloid or collinear spin density wave will not change the extinction of @xmath107 $ ] , whilst the two - orbit structure remains .
it is also the case that adding a canting in the @xmath5-direction will not contribute to a satellite peak as adding a @xmath5-axis component to the magnetic moment that oscillates with a periodicity of @xmath40 will contribute only to the satellite of allowed bragg peaks .
if we assume the existence of an easy - axis for the magnetic moment tied to the crystal geometry , it is reasonable to assume that this lies in the @xmath3-@xmath5 plane . the black line in fig .
[ fig : feas : cryst ] shows an example direction for the easy - axis , and the green and blue arrows show the canting effect on the @xmath3 component of the moment towards the easy - axis .
for the moments on the fe@xmath108 and fe@xmath109 sites that make up one orbit of the helix , the magnetic easy - axis on the fe@xmath109 site will be a reflection by the @xmath110 mirror plane of the easy - axis on site fe@xmath108 , resulting in a canting in the opposite direction along the @xmath5-axis .
the same relationship holds for the canting effects between the fe@xmath111 , fe@xmath112 sites . as fe@xmath108 and fe@xmath109 are half a unit cell apart the oscillation in the @xmath5-axis canting
can be described by a co - sinusoid with a periodicity of the unit cell , with a phase shift between the two orbits , proportional to the difference in the @xmath113-component of the positions of fe@xmath108 and fe@xmath111 .
the resulting @xmath5-axis component of the magnetic moment is dependent on both its position along the @xmath5-direction in the unit cell and the position around the magnetic helix , as only the @xmath3-axis component of the moment experiences a canting effect .
a @xmath5-axis component of the @xmath89th magnetic moment is given by @xmath114 where @xmath115 is the magnitude of the @xmath3 component of the spin , and @xmath116 is a constant that is determined by the strength of the canting effect , and @xmath117 takes the value @xmath118 depending on the atomic site , ( i.e. fe@xmath108 and fe@xmath111 take the value + 1 and fe@xmath109 and fe@xmath112 the value @xmath119 ) .
the specific direction of magnetic easy - axis is included in the the value @xmath117 .
for example , if the easy - axis is perpendicular to the example shown in fig . [
fig : feas : cryst ] , then @xmath116 takes a negative value ; if the easy - axis is entirely along the _ a_- or _
c_-axis then @xmath116 will be zero . including the @xmath5-axis component in the structure factor
allows the simulation of the full polarisation analysis of the @xmath107 $ ] ( solid lines in fig .
[ fig : feas : softpolanalysis ] ) .
the predicted structure factor for the @xmath107 $ ] is only dependent on the @xmath5-axis component of the magnetization vector .
there are no parameters to fit in simulating the flpa , as the only parameter @xmath116 , controls the strength of the tilting , i.e. the magnitude of the @xmath5-axis component with respect to the helical component , and consequently has no impact on the polarisation dependence of the @xmath107 $ ] .
the simulation of the polarisation dependence of the @xmath107 $ ] reflection gives good agreement with a collinear @xmath5-axis moment .
this confirms the canted model , where only the component of the moment in the @xmath5-direction experiences a canting effect , and the fe@xmath120 and fe@xmath121 sites have opposite canting effects .
however , the full polarisation analysis of the @xmath107 $ ] reflection does not contain information about the magnitude of the canting , while the @xmath99 $ ] reflection only contains information about the @xmath5-axis component .
we note that it is therefore not possible to extract the size of the canting from polarisation analysis of reflections along the @xmath94 $ ] direction .
however , we note here that the results of our dft measurements described in section [ dft_section ] , allow us to provide physical motivation for the origin of the canting and therefore further evidence that our model is appropriate . finally , the @xmath65 $ ] reflection was measured and an azimuthal dependence of the scattering collected . the structure factor for this reflection
is dominated by the helical magnetic term , unless the phase difference between the orbits falls within the range 1.8 and 2.0 radians , canting strength dependent , in which case the canting component becomes the more dominant .
figure [ fig : feas : azimuthoap09 ] shows the results of the azimuthal dependence of the @xmath65 $ ] satellite reflection .
three different models are shown in fig .
[ fig : feas : azimuthoap09 ] : the predicted azimuth for a circular helical structure with spins restricted within the @xmath3-@xmath4 plane ; a rotated elliptical structure ; and the prediction using the _
c_-component only .
the data does not agree quantitatively with any of the three models , while qualitatively it most resembles the elliptical model .
c_-component and circular models are significantly different from the measured result . this result is sufficient to further rule out the non - elliptical case , but the measurement is not sufficient to gain any information about the phase difference between the two orbits , nor the magnitude of the canting .
the chirality of the magnetic structure has no effect on the simulations for a full linear polarisation analysis , but is important when circular incident light is used@xcite . for a chiral magnetic structure , incident circular
positive and incident circular negative light can be used to establish the chirality@xcite .
energy scans of the @xmath99 $ ] and @xmath107 $ ] reflections were performed at the fe @xmath26 edges ( fig .
[ fig : feas : cirenergysoft ] ) the predicted intensities of the @xmath99 $ ] and @xmath107 $ ] peaks were made using the structure factor calculations for both chiralities ( fig .
[ fig : feas : predictedcirc ] ) . the intensity of the @xmath107 $ ] reflection does not change between circular positive and circular negative incident light , as this peak is sensitive only to the _
c_-axis component which does not have a chiral nature .
the @xmath99 $ ] peak does show a variation with incident circular light , and predictions show that one circular channel is expected to be over twice as intense as the other .
this is observed to be the case .
the predictions also show that for a right - handed chiral helix the positive circular channel is expected to be the most intense , and for a left - handed chiral helix the negative circular channel is expected to be more intense .
the non chiral case , where the two orbits have opposite chirality , is predicted to show equal intensity in the circular positive and negative channels .
the energy scans show the positive circular channel was the most intense channel , ruling out the non - chiral case and strongly suggesting that the magnetic helix is right - handed . as a check
, the linear polarisation analyser was used to examine the scattered beam from the @xmath99 $ ] satellite reflection , with both circular positive and circular negative incident light .
using the helical structure factor , with the parameters from the fit of the linear polarisation and azimuth measurements , the analyser scans were simulated for both chiral cases .
for the left and right - handed chiral cases the positive and negative incident lights are predicted to show opposite behaviour .
we find that our measurement resembles the right - handed chiral structure , providing further confirmation that the magnetic helix has right - handed chirality .
the resonant x - ray scattering results have shown that a double circular helical magnetic structure is insufficient , and that the helix maps out an ellipse in the @xmath3-@xmath4 plane .
this ellipse has been shown to have an major axis 2.58 times longer than the minor axis .
the azimuthal measurement , showed that major axis of the ellipses is rotated @xmath122 away from the @xmath4-axis .
the full polarisation analysis of the unexpected @xmath107 $ ] peak , requires a canting of the @xmath3-axis component of the moment into the @xmath5-direction with a periodicity of the unit cell .
the absolute magnitude of the canting can not be found from the measurements taken , just its presence .
the phase difference between the two magnetic orbits has not been found .
[ fig : feas : magresult ] shows the @xmath3- , @xmath4- , and @xmath5-axis components of the magnetic moment for the canted spin helix , for one orbit .
this @xmath3-@xmath5 canting relation , in which the @xmath3-axis component of the helix is canted with the periodicity of the unit cell @xmath5 parameter , results in a total magnetic structure with a periodicity longer than given by @xmath40 . if we assign @xmath123 , such that the commensurate position is @xmath124 , then the effect of the canting is to make the magnetic helix repeat every 18 unit cells along the @xmath5-axis .
this can be seen in fig .
[ fig : feas : magresult ] where the moment rotates around the helix seven times before returning to its starting position .
it should be noted that we have assumed that the canting relation is between the @xmath3- and @xmath5-moment directions , as canting in the @xmath4-direction would break the reflection symmetry .
this unusual canting effect which only occurs along one direction of the helix combined with the ellipticity explains the unusual magnetic susceptibility measured by segawa _
et al._@xcite .
whilst an elliptical helical structure goes some way to explain why the susceptibility along the @xmath4-direction is lower than along the @xmath3-direction , it does not explain the presence of the field splitting in only one direction .
zero - spin dft calculations on iron arsenide converged to a state with four bands crossing the fermi surface , as shown in fig . [ fig : robert2](a ) .
hole and electron curvature is present at the fermi surface , in agreement with previous calculations and experiment @xcite .
no symmetry was enforced in the calculation and both the lda and the gga calculations converged onto a @xmath1 symmetry .
.relative energies and spin magnitudes for different states calculated with the gga , which show an almost linear relation between the two .
* fmsf state has varying spin magnitudes , the mean value is stated here . [
cols="<,>,>,>,>,>,>",options="header " , ] collinear spin - polarized calculations were performed for all spin parallel - antiparallel pairings in the unit cell .
there are three antiferromagnetic states , a ferromagnetic state ( fm ) , and a ferrimagnetic state where one spin is flipped from the fm state ( fmsf ) .
the antiferromagnetic states are identified by the iron atoms which have parallel spins : @xmath125 ( afm1 ) , @xmath126 ( afm2 ) , and @xmath127 ( afm3 ) .
the relative energies and ordered spin moment for these states and the zero - spin state using the gga are shown in table [ tab : pberes ] .
these agree with previous calculations@xcite .
the lda results follow the same trends as the gga results , but with lower ordered moments .
the magnitude of the ordered moment is found to increase with the number of antiferromagnetically aligned pairs of fe moments .
we also find that the the energy of the states varies linearly with ordered moment .
the energy relative to the afm3 state is best fitted by @xmath128 , where the sum is over spins @xmath129 , @xmath130 mev/@xmath20 and @xmath131 mev .
the observed linear energy dependence on spin magnitude contrasts sharply with the heisenberg and ising models which have a quadratic energy dependence , and is instead reminiscent of a stoner instability as found in ferromagnetic metals@xcite .
examination of the electron density in the system also shows that it changes with the transition to the ordered spin - state . to explore the origin of this linear dependence we calculated differences in total electron density between states .
total density differences between the zero - spin state and afm1 state are shown in fig.[fig : robert1](c ) and ( d ) .
these show that when the system is relaxed into the afm1 state , the electron density descreases between iron nearest - neighbours ( fe@xmath108-fe@xmath109 , fe@xmath111-fe@xmath112 ) and increases along fe - as bonds ( fe@xmath132-as@xmath132 ) in a pattern that forms strings along the @xmath3-direction .
the arsenic atoms were found to have a complicated local arrangement of spin density with no overall magnetic moment . from our calculations we are able to provide an insight into the @xmath3-@xmath5 canting relation of the magnetic order arises .
the helical magnetic ordering wavevector is found to change with temperature @xcite , and this could imply that the system is sensitive to small perturbations .
in fact , there are several frustrated non - equivalent fe - as bonds in the material which are likely to be responsible for the sensitivity of the ground state
. there are three inequivalent fe - fe bonds , although one bond ( fe@xmath108-fe@xmath111 ) is seperated by a much larger distance than the others .
we find that the spin moment magnitude on individual iron sites is correlated with the number of antiferromagnetically aligned nearest neighbour fe atoms . in our calculations , an iron atom with no antiferromagnetic short bonds has a moment of @xmath133 ; those with two antiferromagnetic short bonds have a moment of @xmath134 , and those with all four antiferromagnetic short bonds have a magnitude of @xmath135 . in the lowest energy state ( afm3 ) , iron atoms linked by a long bond ( fe@xmath108-fe@xmath111 ) must be ferromagnetically aligned .
the trend in table i shows that the structures with more antiferromagnetic bonds have a larger spin magnitude and lower total energy .
an explanation for the occurence of the helical magnetic state rather than afm3 is that it allows the energetically unfavourable ferromagnetic pairing along the long bond in the afm3 structure to reduce energy by canting . due to the sensitivity of the ground state ,
we expect weak mechanisms such as spin - orbit coupling to be decisive in realizing the ground state magnetic structure , and here we show that spin - orbit coupling explains the presence of the ordered component in the @xmath5-direction . in order to see the effect of the spin orbit interaction on the ordered magnetic structure , we calculate the energetic perturbation of the spin - orbit interaction between the iron electron spins and the projected atomic orbitals .
this is used to estimate the preferred direction of fe spin alignment .
the perturbation to the ground state energy can be calculated using the minimization of the energy density functional given by @xmath136+\varepsilon h_{\mathrm{spin\text{-}orbit } } [ \rho],\ ] ] where @xmath137 $ ] is our unperturbed density functional for the energy and @xmath138 is the electron density . in the kohn - sham representation we can use the usual electronic formulation for the spin - orbit interaction , including it as a correction term to the electrostatic field around a given atom .
this is given by @xmath139 where the sum is over the kohn - sham orbitals @xmath140 , @xmath141 is the electrostatic potential around the atomic nucleus , @xmath142 is the distance to the centre of the atom , and @xmath143 is the band mass of the ks orbital .
we approximate the electric field as radial , and this is valid as long as the core orbitals that shield the nuclear charge are not affected by the bonding .
this radial symmetry leads to the condition @xmath144=0 $ ] , and further for feas we only consider the fe @xmath47 orbitals which are involved in the magnetic structure .
this permits a significant simplification of the calculation . for zero - spin calculations where @xmath145 , the energetic perturbation @xmath146 is second - order in the perturbation parameter@xcite @xmath147 , and is given by @xmath148 where @xmath149 is the direction of the spin moment on an iron atom and @xmath150 is the component of angular momentum in this direction .
the integral is over the orbitals @xmath140 evaluated at the fermi surface , and is dependent on the gradient of the energy @xmath22 of the kohn - sham orbitals at the fermi surface @xmath151 .
we find that the diamagnetic contribution of the orbitals is small compared to @xmath146 and has therefore been neglected .
a perturbation to the lowest nonzero order does not change the electron density from that of the ground state , and so we are free to choose the value of @xmath149 and calculate a physically meaningful energetic perturbation for this chosen spin orientation @xcite .
we are therefore able to create a full map of the energetic perturbation for different spin alignments .
this allows us to assess the spin anisotropy of a specific fe atom .
the principle parts of a plane - wave pseudopotential calculation are the projections of the kohn - sham orbitals onto the atomic basis set .
the atomic orbital projections do not necessarily obey crystal symmetries , to generate the full set of projections the relevant local symmetry operators are calculated from the wigner-@xmath0 matrices @xcite and , if required , the application of a reflection .
from this complete orbital projection , projection amplitudes at the fermi surface are calculated using a @xmath152-spline interpolation@xcite .
this projection of the fermi surface is then used as the basis for the energetic perturbation computation .
the final result is calculated by applying eq .
[ equation : fincalc ] to each fermi surface point , and @xmath153 is calculated using a mulliken orbital projection @xcite , and the use of wigner-@xmath0 matrices to include a rotation from the @xmath113-direction to @xmath149@xcite .
this is performed successively for each value of @xmath149 to generate a full map ( in energy ) of the perturbation , which is chosen to be a polar map with regular intervals in both @xmath49 and @xmath154 coordinates .
we take the unperturbed state ( corresponding to @xmath137 $ ] ) to be the zero spin configuration .
this state was chosen for the calculation as it has the largest fermi surface .
results were obtained for an fe@xmath7 test spin aligned along different directions , and these energies are shown , for each band in fig .
[ fig : robert1 ] .
the calculations show clearly the effects of anisotropy , which causes a large difference in energy for different iron spin orientations .
in general , we note that the extrema in energy do not lie along a crystal axis . crossing bands 3 and 4 ( fig .
[ fig : robert1 ] ) makes a significantly larger contribution to @xmath146 than the other bands , as they have the largest fermi surfaces and the highest density of @xmath0-orbitals . on band 3
the highest energy perturbation occurs when the fe spin points in the @xmath3-@xmath5 plane , at an angle of 23@xmath155 from the @xmath5-direction ( towards @xmath3 ) .
we find that spin alignment along the @xmath4-axis is energetically unfavourable .
the results for band 3 compare favourably with the measured susceptibilities in the high temperature paramagnetic spin state , in which @xmath156 @xcite .
finally , the anisotropy of the fe @xmath0-orbitals is calculated here from the second order correction energy @xmath157 as @xmath158 .
this quantity , on band 3 , is high in the @xmath3-@xmath5 plane , at 81% , meaning that the local environment strongly affects the fe @xmath0-orbitals in this band .
the other bands have anisotropies 77% ( band 1 ) , 51% ( band 2 ) and 97% ( band 4 ) . for crossing band 2 the lowest energy spin direction
lies exactly midway between @xmath3 and @xmath159 ; and in crossing band 4 it lies along @xmath5 .
the moment on the iron can not satisfy all of these conditions simultaneously , but by far the largest proportion of @xmath0-orbitals lie on crossing band 3 .
however , it is notable that the optimal direction of spin alignment lies off - axis in the @xmath3-@xmath5 plane , and that spin - orbit effects will couple ordering in the @xmath3 and @xmath5 directions , with the relative orientation dependent on the iron site .
in conclusion , we have used polarised resonant x - ray scattering measurements to investigate the incommensurate nature of the magnetic helix along the @xmath5-axis in feas as well as its ellipticity . we have found evidence of a much greater ellipticity to that inferred previously , as well as a @xmath3-@xmath5 canting relation in which moments are canted out of the @xmath3-@xmath4 plane in addition by use of circular polarised x - rays we have demonstrated the existence of a right - handed chiral structure . we have combined our experimental measurements with dft calculations which have quantified the relative energies of different antiferromagnetic states and showed that the origin of the spin canting effect we have measured may be accounted for by considering the spin - orbit coupling .
finally , we note that the observation of both the fundamental magnetic satellite and also higher - order harmonics in feas suggest a different behaviour to modulated antiferromagnetism in elemental metals such as chromium and the rare - earths .
we are grateful to epsrc and the john templeton foundation for financial support .
we acknowledge diamond light source for time on beamline i10 under proposal si6422 . parts of this research were carried out using po9 at the light source petra iii at desy , a member of the helmholtz association ( hgf ) ( proposal i-20110275 ec ) .
work at king fahd university of petroleum and minerals was supported by national science technology and innovation plan ( nstip ) under project no .
11-adv1631 - 04 .
we acknowledge the uk national supercomputing facility ( archer ) , durham hpc ( hamilton ) , the facilities of n8 hpc and the ukcp consortium ( epsrc grant no .
ep / f037481/1 ) .
data presented in this paper will be made available via doi.org/10.15128/r36h440s444 .
l. m. corliss , j. m. hastings , and r. j. weiss , phys . rev . lett . * 3 * , 211 ( 1959 ) ; v.n .
bykov , v. s. golovkin , n. v. ageev , v. a. levdik , and s. i. vinogradov , dokl .
nauk sssr * 128 * , 1153 ( 1959 ) [ sov .
. dokl . * 128 * , 1070 ( 1960 ) ] . | we present the results of resonant x - ray scattering measurements and electronic structure calculations on the monoarsenide feas .
we elucidate details of the magnetic structure , showing the ratio of ellipticity of the spin helix is larger than previously thought , at 2.58(3 ) , and reveal both a right - handed chirality and an out of plane component of the magnetic moments in the spin helix .
we find that electronic structure calculations and analysis of the spin - orbit interaction are able to qualitatively account for this canting . |
let @xmath0 be a discrete set of nonzero reals such that the sum @xmath1 is finite and define the real entire function @xmath2 of exponential type zero by @xmath3 for a given sequence @xmath4 ( referred to as _ coupling constants _ or _ data _ ) , where we denote with @xmath5 the one - point compactification of @xmath6 , we consider the following task . find a pair of real entire functions @xmath7 of exponential type zero such that the three conditions listed below are satisfied . 1 .
coupling condition : whenever @xmath8 . ] @xmath9 2 .
growth and positivity condition : @xmath10 3 .
normalization condition : @xmath11 let us first assume that the pair @xmath7 is a solution of the coupling problem with data @xmath12 .
the growth and positivity condition ( g ) means that the function @xmath13 is a so - called herglotz
nevanlinna function ( * ? ? ?
* chapter vi ) , @xcite , ( * ? ? ?
* chapter 5 ) . upon invoking the open mapping theorem ,
this first of all guarantees that all zeros of the functions @xmath14 and @xmath15 are real .
it furthermore entails that the zeros of the function in the numerator of and the zeros of the function in the denominator of are interlacing ( after possible cancelations ) ; see ( * ? ? ?
* theorem 27.2.1 ) . from this
we may conclude that the functions @xmath14 and @xmath15 are actually of genus zero and satisfy the bound @xmath16 in fact , this inequality follows essentially from roughly estimating the individual factors in the corresponding hadamard representation , with the normalization condition ( n ) taken into account , and employing the interlacing property mentioned above .
we should emphasize here that this upper bound is always independent of the actual coupling constants @xmath12 . on the other side
, the condition ( g ) also tells us that the residues of the function in at all poles are negative . in conjunction with the coupling condition ( c ) , this implies @xmath17 for all those @xmath18 for which the coupling constant @xmath19 is finite .
unless it happens that @xmath20 is a zero of the function @xmath15 , this constitutes a necessary restriction on the sign of the coupling constant @xmath19 in order for a solution of the coupling problem to exist . roughly speaking
, the coupling constants are expected to have alternating signs beginning with non - negative ones for those corresponding to the smallest ( in modulus ) positive and negative element of @xmath0 .
motivated by these considerations and the nature of our applications , we introduce the following terminology .
coupling constants @xmath4 are called admissible if the inequality @xmath21 holds for all those @xmath18 for which @xmath19 is finite .
the main purpose of the present article is to prove that this simple condition is sufficient to guarantee unique solvability of the corresponding coupling problem .
if the coupling constants @xmath4 are admissible , then the coupling problem with data @xmath12 has a unique solution .
apart from this result , we will also establish the fact that the solution of the coupling problem depends in a continuous way on the given data .
let @xmath22 be a sequence of admissible coupling constants that converge to some coupling constants @xmath12 ( in the product topology ) .
then the solutions of the coupling problems with data @xmath23 converge locally uniformly to the solution of the coupling problem with ( admissible ) data @xmath12 . in the simple case when the set @xmath0
consists of only one point , it is possible to write down solutions explicitly in terms of the single coupling constant .
suppose that @xmath24 for some nonzero @xmath25 so that @xmath26 from the very definition , we readily see that some @xmath4 is admissible if and only if the coupling constant @xmath27 is not a negative real number . in this case , the unique solution @xmath7 of the coupling problem with data @xmath12 is given by @xmath28 which has to be interpreted in an appropriate way when @xmath27 is equal to zero or not finite . otherwise , when the coupling constant @xmath27 is a negative real number , the coupling problem with data @xmath12 has no solution at all .
the following observation sheds some light on what happens in the general situation , when the coupling constants are not necessarily admissible .
let @xmath4 be coupling constants and define the sequence @xmath29 by @xmath30 where the set @xmath31 consists of all those @xmath18 for which @xmath19 is finite and @xmath32 since the coupling constants @xmath33 are admissible , there is a unique solution @xmath7 of the coupling problem with data @xmath33 .
now one can show that the coupling problem with data @xmath12 is solvable if and only if the function @xmath15 vanishes on the set @xmath31 . in this case , the solution of the coupling problem with data @xmath12 is unique and coincides with the solution of the coupling problem with data @xmath33 .
before we proceed to the proofs of our results , let us point out two applications that constitute our main motivation for considering this coupling problem for entire functions .
first and foremost , the coupling problem is essentially equivalent to an inverse spectral problem for second order ordinary differential equations or two - dimensional first order systems with trace class resolvents .
this circumstance indicates that it is not likely for a simple elementary proof of our theorem to exist , as the uniqueness part allows one to effortlessly deduce ( generalizations of ) results in @xcite , @xcite , @xcite , @xcite , @xcite , which had to be proven in a more cumbersome way before . on the other side , the coupling problem is also of relevance for certain completely integrable nonlinear wave equations ( with the camassa
holm equation @xcite , @xcite and the hunter
saxton equation @xcite being the prime examples ) when the underlying isospectral problem has purely discrete spectrum . for these kinds of equations
, the coupling problem takes the same role as riemann
hilbert problems do in the case when the associated spectrum has a continuous component ; see @xcite , @xcite , @xcite .
in particular , the stability result for the coupling problem enables us to derive long - time asymptotics for solutions of such nonlinear wave equations @xcite . as a prototypical example
, we are going to discuss the spectral problem for an inhomogeneous vibrating string @xmath34 on the interval @xmath35 , where @xmath36 is a complex spectral parameter and @xmath37 is a positive borel measure on @xmath35 representing the mass distribution of the string .
we impose a growth restriction on the measure @xmath37 to the extent that the integral @xmath38 is finite . despite both endpoints being potentially singular
, these conditions guarantee that the associated dirichlet spectrum @xmath0 is a discrete set of positive reals such that the sum is finite ( we refer to ( * ? ? ?
* section 2 ) for details ) .
this fact is reflected by the existence of two solutions @xmath39 and @xmath40 of the differential equation with the asymptotics @xmath41 such that @xmath42 and @xmath43 are real entire functions of genus zero .
because the spectrum @xmath0 consists precisely of those @xmath36 for which the solutions @xmath39 and @xmath40 are linearly dependent , we may infer that the function @xmath2 defined by is nothing but the wronskian of these solutions , that is , one has @xmath44 where we take the unique left - continuous representatives of the derivatives .
our interest here lies in a particular associated inverse spectral problem which consists in recovering the borel measure @xmath37 from the spectrum @xmath0 and the sequence of accompanying norming constants @xmath45 defined by @xmath46 in order to work out the connection to the coupling problem , we first mention that for every eigenvalue @xmath18 one has the relation @xmath47 which is somewhat reminiscent of the coupling condition .
furthermore , the growth and positivity condition will be due to the fact that the function @xmath48 is a herglotz
nevanlinna function for all @xmath49 . as a final ingredient
, it remains to note the identity @xmath50 upon simply normalizing the functions @xmath42 and @xmath43 at zero , we are now in the position to observe the following : _ for every given @xmath49 one has @xmath51 where the pair @xmath7 is the unique solution of the coupling problem with ( admissible ) data @xmath12 given by _
hence we are able to retrieve the measure @xmath37 from the spectrum and the norming constants by means of solving a family of coupling problems .
in particular , this guarantees that @xmath37 is uniquely determined by the given spectral data , a fact that usually requires considerable effort @xcite , @xcite , @xcite , @xcite , @xcite .
more generally , the coupling problem can also be employed to solve analogous inverse spectral problems for indefinite strings as in @xcite or canonical systems with two singular endpoints .
let us consider the camassa
holm equation @xmath53 which arises as a model for unidirectional wave propagation on shallow water @xcite .
associated with a solution @xmath54 is the family of spectral problems @xmath55 whose significance lies in the fact that their corresponding spectra are independent of the time parameter @xmath56 . in the case
when @xmath54 is real - valued and such that the integral @xmath57 is finite for one ( and hence for all ) @xmath56 , the common spectrum @xmath0 is a discrete set of nonzero reals such that the sum is finite . apart from this , these assumptions also guarantee the existence of two solutions @xmath58 and @xmath59 of the differential equation with the spatial asymptotics @xmath60 such that @xmath61 and @xmath62 are real entire functions of genus zero .
the function @xmath2 defined by is precisely the wronskian of these solutions ; @xmath63 independent of time @xmath56 . for every eigenvalue @xmath18
, we therefore may write @xmath64 with some real - valued function @xmath65 . the crucial additional fact for this to be useful
is that the time evolution for these quantities is known explicitly and given by @xmath66 of course , this simple behavior of the spectral data is highly exceptional and only due to the completely integrable structure of the camassa
holm equation .
before we are able to substantiate the importance of the coupling problem in this context , we are left to note that the function @xmath67 is a herglotz
nevanlinna function for all @xmath68 and @xmath56 with @xmath69 after taking the normalizations of the functions @xmath61 and @xmath62 at zero into account , we may now state the following : _ for any given @xmath68 and @xmath56 , we have @xmath70 where the pair @xmath7 is the unique solution of the coupling problem with ( admissible ) data @xmath12 given by _
thus we may recover the solution @xmath54 by means of solving coupling problems whose data are given explicitly in terms of the associated spectral data at an initial time .
since we are going to employ de branges theory of hilbert spaces of entire functions @xcite to establish the uniqueness part of our theorem , we begin with summarizing some necessary notation .
first , an entire function @xmath72 is called a _ de branges function _ if it satisfies the inequality @xmath73 for all @xmath36 in the open upper complex half - plane .
associated with such a function is a _ de branges space _ @xmath74 .
it consists of all entire functions @xmath75 such that the integral @xmath76 is finite and such that the two quotients @xmath77 and @xmath78 are of bounded type in the upper half - plane with non - positive mean type , where @xmath79 is the entire function defined by @xmath80 endowed with the inner product @xmath81 the space @xmath74 turns into a reproducing kernel hilbert space ; see ( * ? ? ?
* theorem 19 and theorem 21 ) . for each @xmath82 , the point evaluation in @xmath83 can be written as @xmath84 where the entire function @xmath85 is given by @xmath86 we now show how de branges spaces arise in connection with our coupling problem .
[ lem1 ] let @xmath4 be such that @xmath19 is finite and non - zero for every @xmath18 and suppose that the pair @xmath7 is a solution of the coupling problem with data @xmath12 . unless the function @xmath15 is constant , there are two de branges functions @xmath87 and @xmath88 of exponential type zero without real roots such that the following properties hold : 1 .
the de branges functions @xmath87 and @xmath88 are normalized by @xmath89 2 .
the de branges spaces @xmath90 and @xmath91 are both isometrically embedded in the space @xmath92 , where the borel measure @xmath93 on @xmath6 is given by @xmath94 and @xmath95 denotes the unit dirac measure centered at @xmath36 .
3 . the corresponding reproducing kernels @xmath96 and @xmath97 satisfy the inequality @xmath98 4 . the space @xmath90 is a closed subspace of @xmath91 with codimension at most one . if @xmath90 coincides with @xmath91 , then @xmath99 otherwise , when @xmath90 has codimension one in @xmath91 , we have @xmath100 where @xmath101 is any nontrivial function in @xmath91 which is orthogonal to @xmath90 . 1 .
the de branges function @xmath102 is normalized by @xmath103 2 .
the de branges space @xmath104 is isometrically embedded in the space @xmath92 .
3 . the corresponding reproducing kernel @xmath105 satisfies the inequality @xmath106 4 .
the space @xmath104 is one - dimensional and @xmath107 under the imposed conditions , all zeros of the functions @xmath14 and @xmath15 are simple .
in fact , if some @xmath20 was a multiple zero of @xmath14 or @xmath15 , then @xmath20 would have to be a zero of the function @xmath2 as well since the function in is a herglotz nevanlinna function . as this means that @xmath20 belongs to the set @xmath0 , the coupling condition would then imply that @xmath20 is a zero of both functions , @xmath14 and @xmath15 , so that the function in the numerator of would have a zero of order greater than two at @xmath20 , which constitutes a contradiction .
let us denote with @xmath108 the set of zeros of the entire function @xmath109 .
due to the integral representation for herglotz nevanlinna functions , we may write @xmath110 with some @xmath111 , @xmath112 and @xmath113 for every @xmath114 such that the sum @xmath115 is finite . since each @xmath114 is indeed a simple pole of the function on the left - hand side of
, one sees that the quantities @xmath116 are actually positive .
now we introduce the herglotz nevanlinna function @xmath117 by @xmath118 where we choose @xmath119 , @xmath120 , @xmath121 , @xmath122 and the quantities @xmath123 are given by @xmath124 if @xmath125 , @xmath126 if @xmath127 and @xmath128 if @xmath129 . as a consequence of this definition ,
one clearly has @xmath130 since the set of nonzero poles of the function @xmath117 is precisely @xmath108 , we may define the real entire function @xmath131 of exponential type zero via @xmath132 from the identity in , we first infer that @xmath133 by simply plugging in the definition of @xmath134 and @xmath135 .
moreover , one verifies that @xmath136 in a straightforward manner , that the function @xmath137 is a herglotz nevanlinna function by using and the normalization @xmath138 because the function @xmath117 is a non - constant herglotz
nevanlinna function , the entire function @xmath139 given by @xmath140 is a de branges function of exponential type zero .
furthermore , the function @xmath139 does not have any real roots since otherwise the functions @xmath109 and @xmath131 would have a common zero which is impossible by definition . if @xmath141 denotes the reproducing kernel in the corresponding de branges space @xmath142 , then we have @xmath143 next , we introduce the matrix - valued herglotz
nevanlinna function @xmath144 by @xmath145 for such a function ( see ( * ? ? ?
* theorem 5.4 ) for example ) , the limit @xmath146 exists and is a non - negative matrix . apart from this
, the matrix @xmath147 is symmetric by definition , which implies that all its entries are real . since the determinant of the matrix @xmath148 is equal to minus one for all @xmath36 in the upper half - plane , we have @xmath149 thus
, we may conclude that the rank of the matrix @xmath147 is at most one .
let us first suppose that the matrix @xmath147 is the null matrix , which entails that @xmath150 due to the integral representation for herglotz nevanlinna functions , we thus have @xmath151 for some @xmath152 and all @xmath36 in the open upper half - plane , where @xmath153 is given by @xmath154 upon taking the coupling condition and into account , we further compute @xmath155 it now follows from ( * ? ? ?
* theorem 32 ) , that for every function @xmath156 one has @xmath157 that is , the de branges space @xmath158 is isometrically embedded in the space @xmath92 . upon choosing @xmath159
if the function @xmath15 is not a constant and @xmath160 otherwise , one readily verifies the claimed properties in this case .
if the matrix @xmath147 has rank one , then there is a @xmath161 and a @xmath162 such that @xmath163 we now introduce the real entire functions @xmath164 and @xmath165 of exponential type zero via @xmath166 in view of ( * ? ? ? * theorem 34 ) , the entire function given by @xmath167 is a de branges function of exponential type zero without real roots and such that the associated de branges space coincides with @xmath142 isometrically .
this also guarantees that the quotient @xmath168 is a non - constant herglotz nevanlinna function .
furthermore , one readily sees that @xmath169 as well as the identity @xmath170 in conjunction with and , we infer that @xmath171 from this we may deduce that the limits @xmath172 exist and are positive .
next , we define the real entire functions @xmath173 and @xmath174 by @xmath175 so that the functions @xmath176 given by @xmath177 are herglotz nevanlinna functions that satisfy @xmath178 as @xmath179 . as a consequence
, we may conclude that @xmath180 in order to finish the proof , let us first suppose that the function @xmath15 is not constant .
as then the function @xmath181 has at least two poles , we may infer that the herglotz
nevanlinna function @xmath182 is not constant .
since the same holds for @xmath183 in any case , we see that the entire functions @xmath87 and @xmath88 given by @xmath184 are de branges functions of exponential type zero without real roots .
furthermore , the analytic functions @xmath185 and @xmath186 defined by @xmath187 are bounded by one on the upper half - plane because the functions @xmath188 and @xmath189 are herglotz nevanlinna functions . due to the integral representation for herglotz nevanlinna functions and
we may write @xmath190 for some @xmath191 , all @xmath36 in the open upper half - plane and @xmath192 . upon noticing that @xmath193 which follows from the coupling condition and , we conclude that @xmath194 in view of ( * ? ? ?
* theorem 32 ) , we see that the de branges spaces @xmath90 and @xmath91 are isometrically embedded in the space @xmath92 .
the third item in the claim follows from the identity @xmath195 the space @xmath90 is a closed subspace of @xmath91 with codimension at most one because ( see also ( * ? ? ?
* theorem 33 and theorem 34 ) ) we have @xmath196 and therefore the corresponding reproducing kernels are related by @xmath197 the left properties in the fourth item are readily verified upon observing that the function @xmath198 in @xmath91 is orthogonal to @xmath90 in view of ( * ? ? ?
* theorem 33 ) and does not vanish at zero since positivity of the second limit in would contradict the definition of @xmath181 in this case . it remains to note that the required normalization can be achieved by redefining @xmath199 through @xmath200 which leaves the corresponding de branges space unchanged ( * ? ? ?
* theorem 34 ) .
otherwise , if the function @xmath15 is constant , then positivity of the second limit in shows that @xmath201 is necessarily equal to zero .
then the function @xmath102 given by @xmath202 is a polynomial de branges function of degree one without real roots .
it follows as in the non - constant case above that the associated de branges space @xmath104 is isometrically embedded in @xmath92 . finally , observing that @xmath203 readily yields the remaining claims .
let us for now suppose that the coupling constants @xmath4 are such that @xmath19 is finite and non - zero for every @xmath18 .
we are going to show that any two solutions , say @xmath204 and @xmath205 , of the coupling problem with data @xmath12 actually coincide . to this end
, we first note that it suffices to verify that the functions @xmath206 and @xmath207 are equal .
in fact , in this case we may conclude from the integral representation for herglotz nevanlinna functions that @xmath208 since the residues of both functions ( due to the coupling condition ) as well as their behavior at zero ( due to the normalization ) are the same , which guarantees that the functions @xmath209 and @xmath210 coincide too .
we distinguish the following three cases : _ case 2 ; precisely one of the functions @xmath211 and @xmath212 is constant . _ without loss of generality , we may assume that @xmath211 is constant but @xmath212 is not .
let @xmath213 and @xmath214 denote the corresponding de branges functions from lemma [ lem1 ] . since the associated de branges spaces are both isometrically embedded in the same space @xmath92 , we infer from the theorem in @xcite that either @xmath215 or @xmath216 . as the space
@xmath217 is one - dimensional , it is impossible that @xmath218 is a proper subspace of @xmath217 and we conclude that @xmath215 .
it follows from ( * ? ? ?
* theorem 33 ) that there are real entire functions @xmath219 , @xmath220 , @xmath221 , @xmath222 with @xmath223 and @xmath224 ( due to the normalization of our de branges functions ) as well as @xmath225 such that ( see also ( * ? ? ?
* section 1 ) ) the quotient @xmath226 is a herglotz
nevanlinna function and the corresponding reproducing kernels satisfy @xmath227 where @xmath228 and @xmath229 are real entire functions such that @xmath230 . differentiating with respect to @xmath36 and evaluating at zero
then gives @xmath231 because lemma [ lem1 ] and the inclusion @xmath215 guarantee the inequality @xmath232 on the other side , we see that @xmath233 . as this means that the herglotz
nevanlinna function @xmath226 has a multiple root at zero , we may conclude that @xmath220 vanishes identically .
due to , this also shows that @xmath222 has no zeros at all and thus is identically equal to one ( since it is of cartwright class ( * ? ? ?
* proposition 1.1 ) ) . in conjunction with the remaining properties of the kernels in lemma [ lem1 ] ,
we thus get @xmath234 _ case 3 ; neither of the functions @xmath211 and @xmath212 is constant .
let us denote with @xmath235 , @xmath236 and @xmath214 , @xmath237 the respective corresponding de branges functions from lemma [ lem1 ] .
since the associated de branges spaces are all isometrically embedded in the same space @xmath92 , we see from the theorem in @xcite that they are totally ordered .
if one of the inclusions , @xmath238 or @xmath239 , holds , then we may deduce that the functions @xmath211 and @xmath212 are equal by literally following the lines of the argument in the previous case .
for this reason , it remains to verify the claim when @xmath240 and @xmath241 . because @xmath218 has codimension at most one in @xmath242
, we see that @xmath243 and analogously also @xmath244 , which results in @xmath245 . after a similar argument
, we furthermore see that @xmath246 as well .
now the claim follows from the properties of the corresponding reproducing kernels in lemma [ lem1 ] . in order to prove uniqueness also under general assumptions ,
let @xmath4 be arbitrary and consider two solutions @xmath247 and @xmath248 of the coupling problem with data @xmath12 .
we first define the sets @xmath249 as well as the entire function @xmath250 by @xmath251 and the sequence @xmath252 via @xmath253 then for any @xmath254 , the pair of real entire functions @xmath255 of exponential type zero defined such that @xmath256 satisfies first of all the coupling condition @xmath257 furthermore , we readily see that the function @xmath258 is a herglotz
nevanlinna function as well as the normalization @xmath259 in other words , the pairs @xmath260 and @xmath261 are solutions of the coupling problem with data @xmath33 when the set @xmath0 is replaced with @xmath262 .
since @xmath263 is finite and non - zero for every @xmath264 , we may invoke the first part of the proof to infer that @xmath265 this shows that solutions to the coupling problem are always unique .
[ lem2 ] if @xmath266 is a rational herglotz
nevanlinna function with a pole at zero , then there is an @xmath267 , positive constants @xmath268 , real constants @xmath269 and non - negative real constants @xmath270 such that @xmath271 where the polynomials @xmath272 and @xmath273 are defined recursively via @xmath274 for all @xmath275 and @xmath276 .
if the function @xmath266 has precisely one pole , then it admits the representation @xmath277 for some @xmath219 , @xmath220 , @xmath278 with @xmath112 and @xmath279 . upon setting @xmath280 , @xmath281 , @xmath282 and @xmath283
, we readily obtain the claim in this case .
now let @xmath284 , suppose that the claim holds for all functions with at most @xmath285 poles and assume that the function @xmath266 has exactly @xmath286 poles .
we still have @xmath287 for some @xmath219 , @xmath288 with @xmath112 and a rational herglotz nevanlinna function @xmath289 that satisfies @xmath290 as @xmath179 .
since @xmath289 is not identically zero , we may write @xmath291 for some positive constant @xmath221 and a rational herglotz nevanlinna function @xmath292 which satisfies @xmath293 as @xmath179 and has less poles than @xmath266 .
the function @xmath292 does not vanish identically because otherwise the function @xmath266 would have only one pole .
for this reason , the function @xmath294 defined by @xmath295 is a rational herglotz
nevanlinna function with a pole at zero but at most @xmath285 poles altogether . due to our induction hypothesis
, there is an @xmath267 , positive constants @xmath268 , real constants @xmath269 and non - negative real constants @xmath270 such that @xmath296 where the polynomials @xmath272 and @xmath273 are given recursively by . upon defining the quantities @xmath297 , @xmath298 and @xmath299 as well as the polynomials @xmath300 and @xmath301 via setting @xmath302 for all @xmath275
, we readily compute that @xmath303 which establishes the claimed representation .
put differently , the previous lemma says that every rational herglotz
nevanlinna function @xmath266 with a pole at zero admits a continued fraction expansion of the form @xmath304 in turn , any function that can be written as such a continued fraction is a rational herglotz
nevanlinna function with a pole at zero .
_ step 1 ; the coupling problem with data @xmath12 is solvable when @xmath0 is a finite set and @xmath19 is finite and non - zero for every @xmath18 .
_ consider the function @xmath266 defined by @xmath305 due to the admissibility of the coupling constants @xmath12 , the function @xmath266 is a rational herglotz
nevanlinna function with a pole at zero .
it follows from lemma [ lem2 ] that there is an @xmath267 , positive constants @xmath268 , real constants @xmath269 and non - negative real constants @xmath270 such that @xmath271 where the polynomials @xmath272 and @xmath273 are defined recursively via . because @xmath306 and @xmath307 must not have any common zeros , we may conclude that @xmath308 upon also taking the residue of @xmath266 at zero and the fact that @xmath309 into account .
moreover , by means of evaluating the residue of @xmath266 at a pole @xmath18 , we get @xmath310 we now deduce from the recursion in that the quotient @xmath311 is a non - constant herglotz nevanlinna function for all @xmath276 . therefore , also the function @xmath312 is a herglotz
nevanlinna function that is not constant if and only if @xmath313 .
next , we define the polynomials @xmath314 and @xmath315 recursively via @xmath316 for all @xmath275 and @xmath317 .
one notes again that the quotient @xmath318 is a herglotz
nevanlinna function for all @xmath319 .
since both sets of polynomials satisfy the same recursion , we readily compute using that @xmath320 independent of @xmath321 . apart from this
, we infer that for each @xmath18 one has @xmath322 which is obvious for @xmath323 due to and then follows for all @xmath317 by repeatedly employing the recursion relation .
since the sum over all @xmath324 is equal to two , we may pick an @xmath325 such that @xmath326 .
\end{aligned}\ ] ] with these definitions , we introduce the real polynomials @xmath14 and @xmath15 such that @xmath327 for all @xmath275 and first note that due to we have @xmath328 from the considerations above , we see that the two functions @xmath329 are herglotz nevanlinna functions . whereas the latter one is never constant
, the former one is constant if and only if @xmath330 and @xmath331 .
however , as this case would contradict the definition of @xmath222 , we see that neither of the functions is actually constant .
thus , a computation reveals that also the function @xmath332 is a non - constant herglotz
nevanlinna function .
it remains to evaluate @xmath333 to see that the pair @xmath7 is a solution of the coupling problem with data @xmath12 .
_ step 2 ; the coupling problem with data @xmath12 is solvable when @xmath0 is a finite set .
_ let us define the finite sets @xmath334 as well as the polynomial @xmath250 by @xmath335 and the sequence @xmath252 via @xmath253 for every @xmath264 , the coupling constant @xmath263 is finite and non - zero with @xmath336 due to the admissibility of @xmath12 .
thus it follows from the first part of the proof that there is a pair of real entire functions @xmath337 of exponential type zero with @xmath338 such that the function @xmath339 is a herglotz nevanlinna function and such that @xmath340 it is now straightforward to verify that the pair @xmath7 defined by @xmath341 is a solution of the coupling problem with data @xmath12 .
_ step 3 ; the coupling problem with data @xmath12 is solvable .
_ for each @xmath284 , let us define the finite set @xmath342 $ ] , the polynomial @xmath343 via @xmath344 and the sequence @xmath345 by @xmath346 for every @xmath347
. then the inequality @xmath348 holds for all those @xmath347 for which @xmath349 is finite .
more precisely , this is due to admissibility of the coupling constants @xmath12 and the fact that @xmath350 when @xmath351 . as we have seen in the second part of the proof , this guarantees that there is a pair of real entire functions @xmath352 of exponential type zero such that @xmath353 such that the function @xmath354 is a herglotz
nevanlinna function and such that @xmath355 because the estimate in gives rise to the locally uniform bound @xmath356 we may choose a subsequence @xmath357 such that the pairs @xmath358 converge locally uniformly to a pair of real entire functions @xmath7 . due to the above bound , both of these functions are of exponential type zero . furthermore , it follows readily that the pair @xmath7 satisfies the coupling condition .
in fact , for every @xmath18 we have @xmath359 as long as @xmath360 is large enough and it suffices to take the limit @xmath361 .
of course , this has to be interpreted appropriately when @xmath19 is infinite .
we are left to note that @xmath362 as well as that we have the normalization @xmath363 to conclude that @xmath7 is a solution of the coupling problem with data @xmath12 .
let @xmath22 be a sequence of coupling constants that converge to some @xmath12 in the product topology and suppose that the pairs @xmath364 are solutions of the coupling problems with data @xmath23 . from the inequality in ,
we get the locally uniform bound @xmath365 if a subsequence @xmath366 converges locally uniformly to a pair @xmath367 , then the functions @xmath368 and @xmath369 are real entire and of exponential type zero due to .
when @xmath18 is such that @xmath19 is finite , then the coupling condition yields @xmath370 in a similar manner , we see that @xmath371 when @xmath18 is such that @xmath372 is not finite . upon noting that @xmath373 as well as verifying the normalization @xmath374 we see that the pair @xmath367 is a solution of the coupling problem with data @xmath12 .
since such a solution is unique , we may conclude by means of a compactness argument , using the bound and montel s theorem , that the pairs @xmath352 converge locally uniformly to the unique solution of the coupling problem with data @xmath12 .
a. boutet de monvel and d. shepelsky , _ riemann hilbert problem in the inverse scattering for the camassa
holm equation on the line _ , in _ probability , geometry and integrable systems _
, 5375 , math .
publ . , 55 , cambridge univ . press , cambridge , 2008 .
p. a. deift , a. r. its and x. zhou , _ long - time asymptotics for integrable nonlinear wave equations _
, in _ important developments in soliton theory _ , 181204 , springer ser
. nonlinear dynam . , springer , berlin , 1993 . | we establish the unique solvability of a coupling problem for entire functions which arises in inverse spectral theory for singular second order ordinary differential equations / two - dimensional first order systems and is also of relevance for the integration of certain nonlinear wave equations . |
we thank a. georges , e. janod , a. j. millis and m. j. rozenberg for insightful discussions . a.a . and
are financed by the european union under fp7 erc starting grant no .
240524 `` superbad '' .
part of this work was supported by european union , seventh framework program fp7 , under grant no .
280555 `` go fast '' .
g.m . acknowledges support of the european research council ( erc-319286 `` qmac '' ) . 3 c. zener . , * 145 * , 523 - 529 ( 1934 )
. l. d. landau . .
, * 2 * , 46 ( 1932 ) .
d. m. newns , j. a. misewich , c. c. tsuei , a gupta , b. a. scott , and a. schrott . ,
* 73 * , 780 - 782 ( 1998 ) .
hidenori takagi and harold y. hwang . .
, * 327 * , 1601 - 1602 ( 2010 ) .
v. guiot , l. cario , e. janod , b. corraze , v. ta phuoc , m. rozenberg , p. stoliar , t. cren , and d. roditchev . , * 4 * , 1722 ( 2013 ) .
p. stoliar , l. cario , e. janod , b. corraze , c. guillot - deudon , s. salmon - bourmand , v. guiot , j. tranchant , and m. rozenberg . , * 25 * , 3222 - 3226 ( 2013 ) .
j. kim , c. ko , a. frenzel , s. ramanathan , and j.e .
, * 96 * , ( 2010 ) .
f. nakamura , m. sakaki , y. yamanaka , s. tamaru , t. suzuki , and y. maeno .
, * 3*. 2536 ( 2013 ) .
a. a. fursina , r. g. s. sofin , i. v. shvets , and d. natelson . , * 79 * , 245131 ( 2009 ) .
a. a. fursina , r g s sofin , i v shvets , and d natelson . , * 14 * , 013019 ( 2012 ) .
s. gunon , s. scharinger , siming wang , j. g. ramrez , d. koelle , r. kleiner , and ivan k. schuller . , * 101 * , 57003 ( 2013 ) .
j. s. brockman , l. gao , b. hughes , c.t .
rettner , m. g. samant , k.p .
roche , and parkinstuart s. p. , * 9
* , 453 - 458 , ( 2014 ) .
e. janod , j. tranchant , b. corraze , m. querr , p. stoliar , m. rozenberg , t. cren , d. roditchev , v ta phuoc , m .-
besland , and l. cario . , * 25 * , 62876305 ( 2015 ) .
m. yoshida , r. suzuki , y. zhang , m. nakano , and y. iwasa .
arxiv:1505.04038 , ( 2015 ) .
j. tranchant , e. janod , b. corraze , p. stoliar , m. rozenberg , m - p .
besland , and l. cario . , * 212 * , 239244 ( 2015 ) .
t. oka , r. arita , and h. aoki .
, * 91 * , 066406 , ( 2003 ) . s. okamoto . , * 101 * , 116807 ( 2008 ) .
m. eckstein , t. oka , and p. werner . , * 105 * , 146404 ( 2010 ) .
j. li , c. aron , g. kotliar , and j.e .
, 226403 ( 2015 ) .
g. mazza , a. amaricci , m. capone , and m. fabrizio .
, * 91 * , 195124 ( 2015 ) .
inoue and m. rozenberg .
, * 18 * , 2289 - 2292 ( 2008 ) .
m. sandri and m. fabrizio . , * 91 * , 115102 ( 2015 ) .
a. georges , g. kotliar , w. krauth , and m.j .
rozenberg . ,
* 68 * , 13 ( 1996 ) .
m. potthoff and w. nolting .
, * 59 * , 25492555 ( 1999 ) . p. werner and a.j .
, * 99 * , 126405 ( 2007 ) . supplementary material .
j. zaanen , g. a. sawatzky , and j. w. allen . , * 55 * , 418421 ( 1985 ) .
a. camjayi , c. acha , r. weht , m.g .
rodrguez , b. corraze , e. janod , l. cario , and m.j .
rozenberg . , * 113 * , 086404 ( 2014 ) . * supplementary informations : + field - driven mott gap collapse and resistive switch in correlated insulators * + g. mazza , a. amaricci , m. capone and m. fabrizio
we consider the following two - bands model hamiltonian : @xmath31 where @xmath32 and @xmath33 , while @xmath34 is the intra - band dispersion on a three - dimensional cubic lattice .
we also add a non - local hybridization @xmath35 which allows for inter - orbital charge fluctuations but leaves the local single - particle density matrix diagonal in the orbital index .
we set the energy unit such that the hopping @xmath1 , the hybridization @xmath36 .
in addition we consider a constant electric field directed along the @xmath37-axis : @xmath38 .
working in the coulomb gauge we express the electric field in terms of a linearly varying potential : @xmath39 and we assume that the field is imposed on a system with finite extension along the @xmath37-direction , mimicking a sample between two external leads kept at finite voltage difference @xmath40 . to this extent
we introduce a three - dimensional layered structure ( fig .
[ fig1](b ) main text ) and fix the reference potential value @xmath41 imposing a symmetric voltage drop @xmath42 respect to its center : @xmath43 the hamiltonian for the slab structure is obtained performing a discrete fourier transform along the @xmath44direction of the fermionic operators defined in momentum space @xmath45 where the explicit form of the basis functions takes into account the open boundary conditions which we impose on the system . using a vector representation for the orbitals fermionic operators : @xmath46 and setting to unity the elementary charge @xmath47 we obtain @xmath48 where @xmath0 is the crystal - field splitting , @xmath10 is the local coulomb interaction strength , @xmath49 is the scalar potential defined by eq .
[ eq : scalar ] and the matrices @xmath50 and @xmath51 contain respectively the intra- and inter- layer hopping amplitudes @xmath52 the model is solved using the extension of the dmft formalism to in - homogeneous systems @xcite , based on the assumption that the self - energies encoding the effect of all the many - body correlation are completely local in space while retaining an explicit dependence on the layer index @xmath37 @xmath53 following the standard dmft approach , the layer - dependent self - energies are then extracted from a set of layer - dependent single - site effective problems to be self - consistently determined imposing that the lattice local green s function @xmath54 computed using such local self - energy is equal to the green s function of the effective single - site problem . indicating with @xmath55 the bare propagators of the effective single - sites problems , the problem is practically solved using the following equations which implicitly relate @xmath54 and @xmath55 @xmath56 @xmath57^{-1}_{zz'}\ ! & = \delta_{zz ' } \left [ i \omega_n\mathbb{i}- \bm{h}_{{\mathbf{k}_{\| } } } - \bm{\sigma}_z(i
\omega_n ) \right ] - \delta_{z , z \pm 1 } \bm{t}_{{\mathbf{k}_{\| } } } , \end{split}\ ] ] where we indicate with @xmath58 the @xmath59 matrix constructed with all the @xmath60 @xmath61 matrices . in the present case
we map the effective single - sites problems onto interacting anderson impurity models which we solve using a finite bath discretization and an exact diagonalization scheme based on the lanczos method @xcite . across the first - order mott transition . on
the metallic side ( left panel ) the two spectral functions overlaps with a partial orbital polarization ( @xmath29 ) . on the insulating side ( right panel )
the two spectral functions are separated with an almost completed orbital polarization ( @xmath62).,title="fig:",width=207 ] across the first - order mott transition . on
the metallic side ( left panel ) the two spectral functions overlaps with a partial orbital polarization ( @xmath29 ) . on the insulating side ( right panel )
the two spectral functions are separated with an almost completed orbital polarization ( @xmath62).,title="fig:",width=207 ]
in this section we provide few additional information concerning the zero bias metal - insulator transition . as described in the main text the transition is driven by the combined effects of the interaction enhancement of the polarization strenght and the shrinking of the coherent quasiparticle peak . as a consequence
, the mott transition appears as a sharp charge redistribution between the two orbitals .
we show this plotting in fig .
[ fig1s](a ) the orbital - resolved spectral functions @xmath63 for two values of the interaction parameter just before and after the metal - insulator critical value @xmath64 . for @xmath65
the two spectral functions overlap with a sizable spectral weight at the fermi level displaying a two - orbitals character .
the abrupt separation of the orbital resolved spectral weight occours for @xmath66 leading to a mott insulator in which the lower band is fully occupied and the upper one empty ( panel b fig . [ fig1s ] ) .
we notice that due to the finite hybridization in the kinetic part of the hamiltonian the orbital polarization is not exaclty complete with less than one percent residual occupation in the upper orbital @xcite .
a more exhaustive description of the sharp differences between the resistive- and the tunnel - breakdown described in the main text can be appreciated looking at the layer distributions of the charge density and of the bias - induced spectral density at the fermi ( see fig . [ fig1s](b ) ) . in the case of the resistive switch , namely the insulator - to - metal transition followed by the abrupt mott gap collapse close to the mott transition , we observe an infinitesimal tilting of the charge distribution ( red squares in fig .
[ fig2s](b ) left ) and an homogenous distribution of the spectral weight ( red squares in fig .
[ fig2s](b ) right ) . on the contrary far away from the mott transition ,
as expected by the strong tilting of the layer - resolved spectral density ( main text ) , a very strong charge redistribution is needed in order to observe the formation of conducting states ( gray to black circles in fig .
[ fig2s](b ) left ) .
this leads to the gradual formation of two metallic regions at the boundaries which are separated by a bulk region in which the spectral weight results exponentially suppressed ( gray to black circles in fig .
[ fig2s](b ) right ) .
such evanescent spectral weight represents the tunnel through the bulk mott insulating region of the carriers from the two doped metallic regions at the slab boundaries @xcite .
this confirms the tunnel - like scenario for the field - induced formation of conducting paths deep in the mott insulating phase ( fig.1e in the main text ) . in the inset
we show the field evolution of the spectral density at the fermi level for the bulk layer .
the spectral density increases linearly up to a threshold value where a clear jump is observed signaling the disappearence of the bulk insulating region .
we take this value to extract the breakdown threshold field for @xmath30 reported in the main text . | mott insulators can be portrayed as unsuccessful metals " : systems in which a strong coulomb repulsion prevents charge conduction notwithstanding the metal - like density of conduction electrons .
the possibility to unlock such large density of frozen carriers with an electric field offers a tantalizing opportunity to realize new mott - based microelectronic devices .
here we explicitly unveil how such unlocking happens by solving a simple , yet generic , model for correlated insulators using dynamical mean - field theory .
specifically , we show that the electric breakdown of a mott insulator can occur via a first - order insulator - to - metal transition , characterized by an abrupt gap - collapse in sharp contrast to the zener tunneling mechanism .
the switch - on of charge conduction is due to the energetic stabilization of a metallic phase that preexists as metastable state in equilibrium and is disconnected from the stable insulator .
our findings rationalize recent experimental observations and offer a guideline for future technological research .
the conventional description of electric breakdown , i.e. the field driven formation of a conductive state in an otherwise insulating system , is based on the well - known landau - zener mechanism of quantum tunnelling across the insulating gap @xcite .
because of the collective nature of their gap , mott insulators recently emerged as ideal candidates to replace semiconductors @xcite , whose gap is , on the contrary , fixed by their chemistry and lattice structure .
recent experiments in several mott insulators @xcite have reported a whole novel scenario for the electric breakdown , including for instance anomalously small threshold fields , which can not be reconciled with the smooth carrier activation predicted by quantum tunnelling .
these findings rather suggest a genuine field - induced insulator - to - metal mott transition , which can be phenomenologically described @xcite through the competition between a stable insulating phase and a metastable metallic one .
however , theoretical studies within the single - band hubbard model , the paradigm of strongly correlated systems , have only shown a breakdown due to tunnelling across the mott gap , as if the latter was as rigid as the band gap in semiconductors @xcite , in disagreement with the experimental observations . here
we reveal a novel mechanism for the electric breakdown , which is compatible with the above experimental evidences and can be realized as soon as extra degrees of freedom commonly active in real materials are taken into account @xcite . in particular
, we disclose this scenario in the non - perturbative solution of a simple two - orbital model for mott insulators , which it is found to undergo a discontinuous transition from an insulator to a gap - collapsed metal at threshold fields much smaller than those expected in a zener breakdown .
this phenomenon arises on the insulating side of the first - order equilibrium mott transition where the stable insulator coexists with a metastable metal , and it is evidently the more pronounced the greater the difference in physical properties of the two coexisting phases .
we argue the latter is the actual crux for this phenomenon to be observable , as it entails wider coexistence regions and stronger metastability .
our model features two orbitals lifted by a crystal - field splitting @xmath0 that are broadened by intra- and inter - orbital hopping into two bands of equal width .
the electrons interact via a local coulomb repulsion .
the density is set to two electrons per site ( half - filling ) .
the energy unit is such that the intra - orbital hopping is @xmath1 . we solve the model by means of dynamical mean - field theory @xcite ( see supplementary informations ) . at equilibrium
the model undergoes a first - order mott transition for a critical value of @xmath2 ( see fig .
[ fig1](a ) ) . for the sake of definiteness , in the following we shall fix @xmath3 , for which @xmath4 .
for @xmath5 the insulating solution is stable , but the metal continues to exist as a metastable solution up to the spinodal point @xmath6 .
the state variable that better characterizes the transition is the orbital polarization @xmath7 .
at @xmath8 , the model describes a partially polarized metal ( @xmath9 ) .
a finite interaction @xmath10 induces a repulsion between occupied and unoccupied states , leading to an effective crystal field larger than the bare value @xmath11 , i.e. increasing @xmath12 . at the first - order transition
the metal turns abruptly into a fully polarized ( @xmath13 ) insulator @xcite , with a finite gap separating the two bands @xcite : a sort of mott insulator `` disguised '' as a conventional band insulator . in order to study the effect of an applied electric field we consider a layered slab of our idealized material subject to a static and uniform electric field @xmath14 ( coulomb gauge ) directed along the slab direction ( fig.[fig1](b ) ) . starting from the equilibrium insulator at @xmath15 we increase the electric field @xmath16 until a _ conducting _
state is established at a threshold value @xmath17 .
this field - induced conducting state has strikingly different properties depending on the distance of the zero - field system from the first - order mott transition point , see fig.[fig1](c)-(d ) .
deep in the mott insulating region ( @xmath18 ) we observe the gradual formation of a conducting path through the slab .
this is the result of the progressive bending of the gapped spectral density eventually leading to the formation of two metal regions at the two ends of the slab .
as the field grows , the two regions extend towards the center until they touch and a conductive path through the sample is established . during such evolution the insulating gap
is rigidly preserved and the conduction happens because the field leads to a sizeable hole- and electron - doping on the two sides .
this picture is not dissimilar from the landau - zener mechanism in band insulators , as already discussed for correlated systems in different contexts @xcite .
a completely novel scenario is encountered when the system is perturbed near the equilibrium mott transition ( @xmath19 ) . in this case
a sharp insulator - to - metal transition occurs at the threshold field @xmath17 .
this is illustrated by the sudden change of the layer - resolved spectral density across the resistive transition reported in panel ( d ) of fig .
[ fig1 ] . as the @xmath17 value is crossed
the gap abruptly collapses : a metallic state characterized by a sizeable spectral weight at the fermi level suddenly appears without any precursor on the insulating side .
these qualitative differences reflect on the quantitative aspects of the transition . from fig .
[ fig1](c)-(d ) it is possible to observe that the field strength needed to induce a conductive channel through the slab is much smaller for the sharp resistive transition with respect to the tunnel breakdown case .
this is readily appreciated in panel ( e ) , showing the threshold field @xmath20 in units of the zero - field insulating gap @xmath21 as a function of the distance from the _ equilibrium _ mott transition .
a clear break is observed in the curve around @xmath22 separating the two breakdown regimes .
remarkably , the strong variations of @xmath17 are independent on the gap value , which weakly varies in the whole region . of the metallic ( blue circles and line ) and the insulating ( red diamonds and line ) solutions , as a function of the electric field @xmath23 .
the filled / open symbols mark the stable / metastable character of the solution for each value of @xmath16 .
( b ) hysteresis loop for the orbital polarization @xmath7 across the zero - bias mott transition .
the red and blue arrows define the directions of continuous evolution for the insulating and metallic phases respectively .
the big green arrow indicates the electric - field induced switch between the stable insulator and the metastable metal .
_ inset _ : effective crystal field @xmath24 in units of the bandwidth @xmath25 as a function of the correlation strength @xmath10 .
the metallic state is destabilized for @xmath26 .
( c ) polarization profile for the metastable metal for increasing electric field strength ( grey arrow ) .
red diamonds / blue circles refer to a system with an insulating / metallic ground state . ] our findings for the resistive transition are directly related to the competition between the two phases that characterize the equilibrium metal - insulator transition , as we explicitly show below . at @xmath27
the insulating solution has a lower internal energy and is separated from the metastable metal by a small energy difference @xmath28 ( fig .
[ fig2](a ) ) .
however , the metallic solution is more charge - polarizable hence it rapidly gains energy as @xmath16 is increased , whereas the energy of the incompressible mott insulator remains constant . as a consequence ,
the two energy curves eventually cross at the threshold field @xmath17 and the metallic phase becomes stable .
in fig .
[ fig2](b ) we report the equilibrium orbital polarization of the two phases near the mott transition point . an hysteresis loop is realized as a function of @xmath10 .
the two solutions , characterized by different values of @xmath12 for any value of the model parameters , are not adiabatically connected across the mott transition .
thus , the applied field can transform the stable insulator into the metastable metal @xcite only through a genuine first - order transition , as highlighted by the sudden collapse of the insulating gap reported in fig .
[ fig1](d ) . in fig .
[ fig2](c ) we show that , as long as the system is able to sustain a metallic solution with partial orbital polarization , @xmath29 , the field drives the resistive switch reducing the orbital charge imbalance of the metastable metal .
thus , the net effect of the electric field is essentially to move in the phase diagram of fig .
[ fig1](e ) as if @xmath10 or better the crystal field @xmath0 were reduced by the field . on the contrary ,
when correlations destabilize any partially polarized metal ( @xmath30 ) the field can only continuously modify the insulating state leading to the formation of the tunnel conducting paths observed in fig .
[ fig1](c ) . besides unveiling the above unconventional breakdown mechanism
, these results also explain why , at least within dmft , the single - band model fails in describing a true resistive transition towards a gap - collapsed metal . in this model metallic and insulating solutions around the mott transition
do not differ by any extensive observable that can be tuned by means of an external field , but simply for the existence ( metal ) or not ( insulator ) of a very narrow quasiparticle peak at fermi that can accommodate only a few percent of carriers .
for that same reason , the region of parameters where an insulator is stable while a metal does exist as a metastable state is extremely narrow , actually vanishing at zero temperature , and essentially the whole stability region of the insulator behaves like our solution in the deep mott insulator beyond the spinodal point . on the contrary ,
our results clearly demonstrate that the coexistence of two distinct phases with a sharp difference in the value of the orbital polarization is the key to observe a genuine resistive transition .
a similar mechanism could be active in a wide class of models for which the `` polarization '' of another , model dependent , degree of freedom starkly distinguishes the metal from the insulator .
the application of an external field would change the polarization favoring the metallic state , ultimately driving the resistive transition .
we thus argue that the results we obtained for a specific model are in fact generic to essentially every `` narrow gap '' mott insulator as long as it is sufficiently close to a first - order mott transition to host a metastable metallic solution competing with the insulating ground state . |
the ising model is one of the most studied models in statistical physics .
we now have a well developed mathematical theory for it s behaviour in 2-dimensions @xcite , extensive numerical work in 3-dimensions , and it is known that the upper critical dimension for the model if @xmath1 .
for @xmath2 it is known that the model takes on it s mean field critical exponents in the thermodynamic limit . for finite systems far less is known rigorously and there has been a long running debate on differences in scaling for systems with free and cyclic boundaries @xcite . in a previous paper @xcite we found an hitherto overlooked way in which for @xmath3 the behaviour of the model depends strongly on the boundary condition , in fact so strongly that at the critical point the difference is not believed to vanish as the side @xmath0 of the system grows .
this conclusion has been questioned @xcite by a group of authors which has earlier been promoting an alternative scaling picture for the case with free boundary conditions . however , in a recent paper we revisited this question with much larger system sizes and found a good agreement with our earlier results , and the standard scaling picture , where the susceptibility of the model with free boundary grows as @xmath4 and for cyclic boundary grows as @xmath5 . in view of these results
a natural question is why we should see a different behaviour for the case with cyclic boundary , and what the exact form of the scaling behaviour for this case should be for @xmath6 above the critical dimension .
our approach to the answer of this question is via the random - cluster model .
the fortuin - kasteleyn random cluster model , or rc - model for short , is a natural extension of the ising , potts and edge percolation models , all captured by varying @xmath7 , one of its two parameters , the other @xmath8 corresponding to the temperature in the first two models and the edge probability in the last .
many properties of the ising model have direct interpretations in terms of the numbers and sizes of connected clusters in the rc - model . in particular the susceptibility corresponds to the average cluster size .
for percolation , the case @xmath9 of the rc - model , aizenman @xcite conjectured that the largest clusters should scale as @xmath10 for @xmath11 with cyclic boundary , instead of @xmath12 for free boundary .
one of the reasons for this conjecture was that this gives the same scaling as for the largest connected in the erds - renyi , or er for short , random graph with @xmath13 vertices at it s critical point , where the largest component has size proportional to @xmath14 .
note that the er - random graph can be seen as the percolation model on the complete graph on @xmath13 vertices .
this conjecture was proved in @xcite , and in @xcite it was proved that asymptotics of the same type as on the complete graph can be expected on a wider class of finite graphs as well . as we have pointed out
percolation is a special case of the rc - model and the latter has also been studied on the complete graph by mathematicians . in @xcite
the critical probability was identified , the exponential asymptotics of the partition function was studied , and it was proved that the phase transition is of second order for @xmath15 and of first order for @xmath16 . later @xcite a more detailed study of the cluster structure was carried out , and it was found that there are three ranges of @xmath7 with distinct behaviour @xmath17 , @xmath16 and @xmath18 .
our aim in this paper is to compare the behaviour of the largest and second largest cluster in the rc - model for @xmath18 , corresponding to the ising model , for 5-dimensional lattices with side @xmath0 and cyclic boundary .
for the case @xmath19 @xcite identified no less than five distinct scaling regions near the critical probability for the complete graph , each with its own asymptotic behaviour for @xmath20 and @xmath21 , the sizes of the largest and second largest clusters respectively .
some of the regions are difficult to study , since in order to obtain correct scaling they would require far larger graphs than those used here .
thus we have focused on three cases : i ) the high - temperature case ( fixed @xmath22 for @xmath23 ) , ii ) near @xmath24 at fixed @xmath25 , and iii ) near @xmath24 at fixed @xmath26 , where @xmath26 and @xmath27 are different @xmath0-dependent couplings .
as we shall see , for these regions we have an excellent agreement between the scaling for the complete graph and for the 5-dimensional ising model with cyclic boundary .
for a graph @xmath28 on @xmath29 vertices and @xmath30 edges the random cluster model s partition function is @xmath31 where @xmath32 is the number of ( connected ) components , or clusters as we will call them , of the graph @xmath33 , i.e. the graph with vertex set @xmath34 and edge set @xmath35 .
note that @xmath36 and @xmath37 are parameters to the distribution . the ising model without an external field , on the other hand , has the partition function @xmath38 where the sum is taken over all functions @xmath39 from @xmath34 to @xmath40 .
thus vertex @xmath41 has spin @xmath42 .
here the energy is @xmath43 , summed over all edges in @xmath44 .
the parameter @xmath45 is the dimensionless coupling , or inverse temperature .
the two models are actually equivalent for @xmath18 when setting @xmath46 .
assuming we have a state @xmath47 from the random cluster distribution in equation , we can obtain a state @xmath39 from the ising distribution in the following way : for each component , pick a spin @xmath48 uniformly at random and assign this spin to all vertices of the component .
going in the other direction is also easy .
starting with a state @xmath39 from the ising distribution at coupling @xmath22 , let @xmath49 .
now add each satisfied edge , i.e. edges @xmath50 with @xmath51 , to the set @xmath35 with probability @xmath46 . for more information on this
see @xcite .
this second scheme is a surprisingly efficient way of obtaining random cluster data .
just start up your trusted ising state generator , whether it be metropolis , heat - bath or wolff cluster @xcite , and convert the ising states to correctly distributed random cluster states .
all graphs studied here are 5-dimensional grid graphs with periodic boundary conditions , i.e. cartesian products of five cycles on @xmath0 vertices , so that @xmath52 and @xmath53 , and we have used @xmath54 , 6 , 8 , 10 , 12 , 16 , 20 and 24 .
we collected the data by generating ising states with the wolff cluster method @xcite and then converting them to random cluster states using the scheme described above . throughout
we use the critical coupling @xmath55 @xcite , thus corresponding to the random cluster critical probablity @xmath56 , and denote one scaled temperature by @xmath57 and another by @xmath58 . recall that the critical temperature for a complete graph on @xmath59 vertices approaches zero as @xmath60 and hence @xmath61 .
we will denote a scaled probability by @xmath62 .
the @xmath63th central moment of a distribution is denoted by @xmath64 , the mean value by @xmath65 and the variance by @xmath66 .
the standard deviation is then @xmath67 .
the median is written @xmath68 .
given a random cluster state @xmath35 the largest and second largest cluster size , i.e. the number of vertices in these clusters , is denoted @xmath20 and @xmath21 respectively .
we will here give a very brief description of some of the relevant scaling results from @xcite for the complete graph on @xmath59 vertices .
where relevant further details will be given in later sections .
as proven in @xcite the critical probability is given by @xmath69 .
now if @xmath70 then the model is not critical and depending on whether @xmath71 is less than 1 or greater than 1 we see behaviour corresponding to the high- and low - temperature regions respectively , with only small clusters in the first case and a large cluster , linear in @xmath59 sized , plus some small ones in the latter case . if @xmath72 we are inside the critical window and need a finer parameterization of @xmath8 .
we thus assume that @xmath73 , where @xmath74 can depend on @xmath59 .
we now see five distinct regions inside the critical window : 1 .
if @xmath75 then , asymptotically , all clusters are trees and small .
2 . if @xmath76 , where @xmath77 is a constant , then @xmath20 is roughly of order @xmath78 .
3 . if @xmath79 but @xmath80 there exists a unique largest component and it s size is @xmath81 , for a value @xmath82 .
if @xmath83 , where @xmath77 is a constant , then @xmath20 is of order @xmath84 and @xmath21 is bounded by @xmath85 .
if @xmath86 and @xmath87 then @xmath20 is of order @xmath88 and @xmath21 of order @xmath89 .
strictly speaking , the complete graph version of the high - temperature case only requires that @xmath90 . of course
, a fixed @xmath91 will satisfy this condition .
hence , for the 5d case we will simply test the case of a fixed @xmath22 for @xmath23 . in @xcite it was shown that @xmath20 is distributed as an extreme - value distribution @xcite .
these have a density function of the form @xmath92 given some parameters @xmath93 .
extreme value distributions have skewness @xmath94 and kurtosis @xmath95 . in figure
[ fig : laskhi ] we show how the skewness and excess kurtosis scales with @xmath0 and how they approach these values for two different temperatures .
the right plot of figure [ fig : laskhi ] shows the distribution of @xmath20 at @xmath96 for @xmath97 . in short
, we have good reason to think that the complete graph behaviour also holds for 5d in this case .
, pointing to @xmath98 , and excess kurtosis @xmath99 , pointing to @xmath100 , of the distribution of @xmath20 at @xmath101 ( blue circles ) and @xmath96 ( red squares ) plotted versus @xmath102 for @xmath54 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
the fitted curves are second degree polynomials .
the black points at @xmath110 are the extreme value distribution values .
right : distribution of @xmath20 for @xmath97 at @xmath96 ( points ) and a fitted density function of equation with parameters @xmath111 and @xmath112.,scaledwidth=99.0% ] , pointing to @xmath98 , and excess kurtosis @xmath99 , pointing to @xmath100 , of the distribution of @xmath20 at @xmath101 ( blue circles ) and @xmath96 ( red squares ) plotted versus @xmath102 for @xmath54 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
the fitted curves are second degree polynomials .
the black points at @xmath110 are the extreme value distribution values .
right : distribution of @xmath20 for @xmath97 at @xmath96 ( points ) and a fitted density function of equation with parameters @xmath111 and @xmath112.,scaledwidth=99.0% ]
here we consider the case of a fixed negative scaled temperature , @xmath113 .
this falls under the complete graph case @xmath114 . in
@xcite it was shown that for this particular case the two largest clusters behave almost surely as @xmath115 for all functions @xmath116 tending to infinity .
since @xmath116 is allowed to grow as slowly as we like , thus boxing in @xmath20 and @xmath21 , it is quite possible that in fact @xmath117 , though with different prefactors .
let us test this for the 5d case . in figure
[ fig : lablambda ] we show the normalised mean @xmath20 and @xmath21 for different negative @xmath27 .
the right plot shows a zoom - in for @xmath21 .
note here that @xmath118 when @xmath119 whereas the @xmath21 counterpart actually has a local maximum .
the right panel of figure [ fig : lablambda ] shows a zoom - in for @xmath21 , both the sampled data for @xmath120 and also an estimated limit function based on fitting second degree polynomials to the values at @xmath27 versus @xmath102 .
the @xmath121 has a distinct local maximum for all @xmath122 and its location may possibly have zero as limit , approaching it very slowly .
it would take considerably bigger graphs to settle that question .
however , a rough estimate gives that the limit has the local maximum @xmath123 at @xmath124 . to conclude this section , we find that the 5d behaviour matches that of the complete graph case .
( above ) and @xmath121 ( below ) versus @xmath27 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
right : @xmath121 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and @xmath126 ( see text).,scaledwidth=99.0% ] ( above ) and @xmath121 ( below ) versus @xmath27 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
right : @xmath121 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and @xmath126 ( see text).,scaledwidth=99.0% ]
the next case is that of @xmath127 , for constant @xmath26 .
this fits under the complete graph case in @xcite of @xmath128 for real constants @xmath77 . in this region
the complete graph has different scalings for @xmath20 and @xmath21 and we treat them separately in the following two subsections . in the complete graph case
it was shown @xcite that @xmath129 and in fact @xmath71 can be replaced by @xmath130 with some positive limit value @xmath71 .
note that the @xmath20-distribution essentially has the same form as the ising magnetisation distribution of a complete graph , which is @xmath131 , see @xcite , except that @xmath20 has support only for @xmath132 .
this follows since the spin - state coming from an rc - state is obtained by assigning a random spin value to each cluster in the rc - state . above all this means that @xmath133 for constant @xmath26 . in figure
[ fig : lakappa1 ] we show the normalised mean @xmath134 and the normalised variance @xmath135 over the interval @xmath136 .
the plot also shows the complete graph values computed from equation when setting @xmath137 .
note how close the complete graph values are to their 5d counterparts .
the limit of the mean @xmath20 is easily found by plotting them for some @xmath26 versus @xmath102 and then fitting a polynomial to the points .
there is some very mild correction to scaling at work for @xmath138 but it is easily captured by a second degree polynomial . a slightly more careful analysis on the case @xmath139 , i.e. at @xmath24 , based on fitting second degree polynomials to all 4-subsets of the data points for @xmath122 , results in a median value @xmath140 , and the error corresponds to the interquartile range .
the complete graph value at @xmath139 is @xmath141 .
plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case .
right : normalised variance of the largest cluster size @xmath135 plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case.,scaledwidth=99.0% ] plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case .
right : normalised variance of the largest cluster size @xmath135 plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case.,scaledwidth=99.0% ] what about the distribution of @xmath20 ?
figure [ fig : ladist ] shows a normalised form of the @xmath20-distribution , @xmath142 where @xmath143 , for a few values of @xmath26 with @xmath144 . between roughly @xmath145 the distribution actually goes through a bimodal phase and this property is not found in the complete graph case of equation .
however , it agrees with the finding in @xcite that for the ising model with cyclic boundary in 5d the energy distribution at the critical temperature becomes bimodal .
the second plot in figure [ fig : ladist ] shows the distribution at @xmath146 for @xmath147 and it clearly shows the distribution retaining its bimodal form with increasing @xmath0 .
-distribution @xmath148 versus @xmath149 for @xmath144 at @xmath139 , @xmath150 , @xmath151 , @xmath152 and @xmath153 .
the black vertical line is located at @xmath154 , the asymptotic mean for @xmath139 .
right : normalised @xmath20-distribution @xmath148 versus @xmath149 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and complete graph case at @xmath146.,scaledwidth=99.0% ] -distribution @xmath148 versus @xmath149 for @xmath144 at @xmath139 , @xmath150 , @xmath151 , @xmath152 and @xmath153 . the black vertical line is located at @xmath154 , the asymptotic mean for @xmath139 .
right : normalised @xmath20-distribution @xmath148 versus @xmath149 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and complete graph case at @xmath146.,scaledwidth=99.0% ] the skewness and kurtosis of the @xmath20-distribution are shown in figure [ fig : lakappa2 ] together with the complete graph case .
clearly there are distinct limits for each @xmath26 , occasionally with some mild corrections to scaling , easily captured by a second degree polynomial .
perhaps unexpectedly , the data suggest a limit value of the skewness of about @xmath155 as @xmath156 .
there is no conflict in the existence of such a limit and our earlier claim of the skewness taking the extreme value distribution value of @xmath157 in the high - temperature case .
it is simply a matter of taking limits in the right order , i.e. @xmath158 and the kurtosis limit is approximately @xmath105 .
this is in fact also the behaviour of the complete graph case though it takes different values . as @xmath159 its skewness approaches @xmath160 and the kurtosis has the limit @xmath161 .
the two models agree well on the case @xmath139 though . here
the 5d case has skewness @xmath162 and kurtosis @xmath163 while the complete graph values are @xmath164 and @xmath165 respectively . of the @xmath20-distribution
plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and omplete graph case .
right : kurtosis @xmath166 of the @xmath20-distribution plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and complete graph case.,scaledwidth=99.0% ] of the @xmath20-distribution plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and omplete graph case .
right : kurtosis @xmath166 of the @xmath20-distribution plotted versus scaled temperature @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and complete graph case.,scaledwidth=99.0% ] in the complete graph case it was shown @xcite that @xmath167 for @xmath128 .
note that @xmath71 may be replaced by function @xmath130 having some limit @xmath71 .
this gives a density function @xmath168 which implies an infinite expectation value .
we will thus instead consider the median value .
so , when we move close enough to @xmath169 , from @xmath170 to @xmath128 , the second largest cluster drops in size from @xmath78 to @xmath171 .
this also seems to be the case for the 5d case though the corrections to scaling are quite significant for @xmath172 .
figure [ fig : lbkappa1 ] shows the normalised mediam @xmath173 versus @xmath26 for 5d and the complete graph case which demonstrates this effect .
note that for @xmath172 it would require enormous graphs to get anything close to the complete graph case . for @xmath174 , however , the two curves quickly agree on the complete graph behaviour , if not on the actual value .
the right plot of the figure shows a normalised form of the @xmath21-distribution , @xmath175 with @xmath176 , for a range of @xmath0 together with the complete graph distribution in equation at @xmath177 where the two cases largely agree . plotted versus scaled temperature @xmath26 for @xmath54 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case .
the inset shows the complete graph median for a wider range of @xmath26 .
right : normalised distribution @xmath178 ( see text ) of @xmath21 , for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and the complete graph case of equation at @xmath179.,scaledwidth=99.0% ] plotted versus scaled temperature @xmath26 for @xmath54 , @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 and the complete graph case .
the inset shows the complete graph median for a wider range of @xmath26 .
right : normalised distribution @xmath178 ( see text ) of @xmath21 , for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 and the complete graph case of equation at @xmath179.,scaledwidth=99.0% ] if the 5d @xmath21-distribution has a fat tailed distribution like the complete graph case of equation this must eventually show up in some higher moment , and indeed it does .
consider the skewness and kurtosis of the @xmath21-distribution in figure [ fig : lbkappa2 ] . both show a divergent behaviour around @xmath180 .
in fact , a very rough estimate suggests that the peak skewness grows as @xmath181 and the peak kurtosis as @xmath182 . of the @xmath21-distribution plotted versus @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
right : kurtosis @xmath166 of the @xmath21-distribution plotted versus @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
, scaledwidth=99.0% ] of the @xmath21-distribution plotted versus @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
right : kurtosis @xmath166 of the @xmath21-distribution plotted versus @xmath26 for @xmath125 , @xmath104 , @xmath105 , @xmath106 , @xmath107 , @xmath108 , @xmath109 .
, scaledwidth=99.0% ]
our aim has been to compare the scaling of the sizes of the largest and second largest clusters for the 5d random cluster model with that seen for the corresponding value of @xmath8 for the complete graph . for the complete graph
there are two non - critical regions and inside the critical window five distinct scaling regions . due to the limitations coming from the range of system sizes for which we can simulate the model we have chosen to work with the high temperature region and two of the regions from the critical window . in each of the tested regions we have found that the scaling from the complete graph agrees well with the observed values from the 5d model , and in many cases not only the scaling but
also the probability distributions agree well .
the most notable exception is the bimodal distribution of @xmath20 for @xmath26 close to 0 for the 5d model .
this feature corresponds to the bimodal energy distribution seen for the ising model on the same graphs in @xcite , which also clearly separates the 5d model with cyclic boundary from the case with free boundary and the infinite system thermodynamic limit .
we expect this agreement to hold for @xmath183 as well .
if we in the scaling for @xmath20 and @xmath21 on the complete graph case replace @xmath59 by @xmath184 we conjecture that we get the correct scaling for the @xmath6-dimensional case with cyclic boundary .
the computations were performed on resources provided by the swedish national infrastructure for computing ( snic ) at high performance computing center north ( hpc2n ) and at chalmers centre for computational science and engineering ( c3se ) . | the finite size scaling behaviour for the ising model in five dimensions , with either free or cyclic boundary , has been the subject for a long running debate .
the older papers have been based on ideas from e.g. field theory or renormalization . in this paper
we propose a detailed and exact scaling picture for critical region of the model with cyclic boundary . unlike the previous papers
our approach is based on a comparison with the existing exact and rigorous results for the fk - random - cluster model on a complete graph .
based on those results we identify several distinct scaling regions in an @xmath0-dependent window around the critical point .
we test these predictions by comparing with data from monte carlo simulations and find a good agreement .
the main feature which differs between the complete graph and the five dimensional model with free boundary is the existence of a bimodal energy distribution near the critical point in the latter .
this feature was found by the same authors in an earlier paper in the form of a quasi - first order phase transition for the same ising model . |
at present , qcd can not describe quantitatively the low - energy nucleon - antinucleon interaction , and various phenomenological approaches have been suggested in order to explain numerous experimental data , see , e.g. , refs .
@xcite and recent reviews @xcite .
however , parameters of the models still can not be extracted with a good accuracy from the experimental data @xcite .
very recently , renewed interest in low - energy nucleon - antinucleon physics has been stimulated by the experimental observation of a strong enhancement of decay probability at low invariant mass of @xmath1 in the processes @xmath2 @xcite , @xmath3 and @xmath4 @xcite , @xmath5 and @xmath6 @xcite , @xmath7 @xcite .
one of the most natural explanation of this enhancement is final state interaction of the proton and antiproton @xcite .
a similar phenomenon was observed in the investigation of the proton ( antiproton ) electric , @xmath8 , and magnetic , @xmath9 , form factors in the process @xmath10 @xcite .
namely , it was found that the ratio @xmath11 strongly depends on @xmath12 ( in the center - of - mass frame ) in the narrow region of the energy @xmath13 near the threshold of @xmath1 production .
such strong dependence at small @xmath13 is related to the large - scale interaction of proton and antiproton .
therefore , it is possible to apply the approaches of @xcite for an explanation of experimental data in the process @xmath0 . in the present paper ,
we use the paris nucleon - antinucleon optical potential @xmath14 which has the form @xcite : @xmath15 where the real part @xmath16 is the @xmath17-parity transform of the well established paris @xmath18 potential for the long- and medium - range distances ( @xmath19 ) , and some phenomenological part for the short distances .
the absorptive part @xmath20 of the optical potential takes into account the inelastic channels of @xmath21 interaction , i.e. annihilation into mesons .
it is essential at short distances and depends on the kinetic energy of the particles .
we perform calculations in the non - relativistic approximation .
the coulomb interaction between proton and antiproton is important only for the kinetic energy @xmath22 mev , where @xmath23 is the fine structure constant , and @xmath24 is the proton mass .
here we consider the process for kinetic energies @xmath25 mev , therefore we neglect the coulomb interaction . taking into account that the difference of the cross sections @xmath0 and @xmath26 is small @xcite
, we calculate the cross sections at a given isospin final states and compare them with the experimental data for @xmath27 . as a result we found that the amplitude with the isospin @xmath28 strongly dominates .
besides , our prediction for the ratio @xmath11 depends on the parameters of the paris potential but independent of the form factor at threshold .
the ratio is also in a qualitative agreement with experimental data .
in the nonrelativistic approximation , the amplitude of @xmath29 pair production in a certain isospin channel @xmath30 near threshold can be presented as follows ( in units @xmath31 ) : @xmath32\ , , \nonumber\\ & & \mathcal g_s^i=\mathcal f_1^i(q^2)+\mathcal f_2^i(q^2)+\frac{\beta^2}{6}[\mathcal f_2^i(q^2)-\mathcal f_1^i(q^2)]\,,\nonumber\\ & & \mathcal g_d^i=\mathcal f_1^i(q^2)-\mathcal f_2^i(q^2)\,,\end{aligned}\ ] ] where @xmath33 , @xmath34 is a virtual photon polarization vector , corresponding to the projection of spin @xmath35 , and @xmath36 is the spin-1 function of @xmath29 pair , @xmath37 is the projection of spin on the vector @xmath38 .
two tensor structures in eq.([2 ] ) correspond to the s - wave and d - wave production amplitudes .
the total angular momentum of the @xmath29 pair is fixed by a production mechanism .
the functions @xmath39 and @xmath40 are the dirac form factors of the @xmath29 pair which include the effects of final state interaction and have a pronounced @xmath41 behavior near the threshold .
summation over the polarization of nucleon pair and averaging over the polarization of virtual photon is performed using the equations , @xmath42 where @xmath43 is the electron momentum .
our aim is to single out the effects of final state interaction . in order to do that , we write the amplitude ( [ 2 ] ) in the form @xmath44\ , , \end{aligned}\ ] ] where
@xmath45 is the fourier transform of the function @xmath46 , the wave function of the @xmath29 pair in coordinate space .
this wave function is the solution of the schrdinger equation @xmath47 where @xmath48 is the optical potential .
note that @xmath49 is the left eigenfunction of the bi - orthogonal set of eigenfunctions of the non - hermitian operator @xmath50 .
the asymptotic form of the wave function at large distances reads @xmath51 in eq.([3 ] ) the form factors @xmath52 and @xmath53 are @xmath54 where the `` bare '' dirac form factors @xmath55 and @xmath56 do not account for the effect of final state interaction .
near threshold these form factors are smooth functions of @xmath41 and can be treated as phenomenological constants .
let us introduce the vector spherical functions @xmath57 as @xmath58 where @xmath59 are spherical harmonics , @xmath60 are clebsch - gordan coefficients , and @xmath61 . in eq .
( [ a ] ) the quantization axes is directed along the vector @xmath38 .
then the wave function @xmath46 can be written in the form @xmath62
\nonumber\\ & & + \sum_j \sqrt{4\pi(2j+3 ) } c^{j\,\lambda}_{j+1\,0,\,1\,\lambda}\big [ u^{i*}_{2j}(r)\bm y_{j\,\lambda}^{j-1}(\bm n)+w^{i*}_{2j}(r)\bm y_{j\,\lambda}^{j+1}(\bm n)\big]\ , , \end{aligned}\ ] ] here the functions @xmath63 , @xmath64 , and @xmath65 have the asymptotic form at large distances @xmath66}-e^{-i[kr - j\pi/2]}\big]\,,\nonumber\\ & & u^i_{1j}(r)=\frac{1}{2ikr}\big[s_{11}^{ij}e^{i[kr-(j-1)\pi/2]}-e^{-i[kr-(j-1)\pi/2]}\big]\,,\nonumber\\ & & w^i_{1j}(r)=\frac{1}{2ikr}s_{12}^{ij}e^{i[kr-(j+1)\pi/2]}\,,\nonumber\\ & & u^i_{2j}(r)=\frac{1}{2ikr}s_{21}^{ij}e^{i[kr-(j-1)\pi/2]}\,,\nonumber\\ & & w^i_{2j}(r)=\frac{1}{2ikr}\big[s_{22}^{ij}e^{i[kr-(j+1)\pi/2]}-e^{-i[kr-(j+1)\pi/2]}\big]\,,\end{aligned}\ ] ] where @xmath67 and @xmath68 are some functions of energy with @xmath69 and @xmath70 .
due to angular momentum conservation , only the terms with @xmath71 and @xmath72 ( i.e. , @xmath73 ) contribute to the matrix element ( [ 3 ] ) . then the amplitude ( [ 3 ] )
can be written as @xmath74\mathbf{\psi}^{i*}_{{\bm k}\lambda}(\bm r)\ , , \nonumber\\ & & \psi^{i}_{{\bm k}\lambda}(\bm r)= [ u^{i*}_{11}(r)\bm \epsilon_\lambda+w^{i*}_{11}(r)\sqrt{4\pi}\bm y_{1\,\lambda}^2(\bm n)]\nonumber\\ & & + \sqrt{5}\ , c^{1\,\lambda}_{2\,0,\,1\,\lambda}[u^{i*}_{21}(r)\bm \epsilon_\lambda+w^{i*}_{21}(r)\sqrt{4\pi}\bm y_{1\,\lambda}^2(\bm n)]\,.\end{aligned}\ ] ] finally we have @xmath75 the first term in @xmath76 contains the large factor @xmath77 while the second term in @xmath78 is small due to the proton mass @xmath24 in denominator .
thus , in the non - relativistic approximation the amplitude @xmath79 reads , @xmath80\bigg\}\ , \,,\end{aligned}\ ] ] where @xmath81 .
the interpretation of this equation is the following . as a result of re - scattering due to the tensor forces
, the pair produced at the origin in @xmath82-wave has a non - zero amplitude to transfer to @xmath83-wave .
the cross section corresponding to the amplitude ( [ 3 ] ) has the form in the center - of - mass frame ( see , e.g. , ref .
@xcite ) @xmath84 is the angle between the electron ( positron ) momentum @xmath43 and the momentum of the final particle @xmath38 . in terms of the `` dressed '' form factors @xmath85 and @xmath86 the electromagnetic sachs form factors ,
corresponding to the contribution of the amplitude with the isospin @xmath87 , have the form @xmath88 \,,\nonumber\\ & & \frac{2m}{q}g_e^i={\cal g}_s^i-\frac{\beta^2}{3}{\cal g}_d^i = g_s^i[u^{i}_{11}(0)-{\sqrt{2}}u^{i}_{21}(0 ) ] \,.\end{aligned}\ ] ] thus , in the non - relativistic approximation the ratio @xmath89 is independent of the constant @xmath90 , @xmath91 note that the electromagnetic interaction is important only in the narrow region where @xmath92 and the nucleon energy is @xmath93mev . in this paper
we will not consider this narrow region and neglect the electromagnetic interaction in the potential .
then , the amplitude of @xmath94 pair production , @xmath95 , and the amplitude of @xmath96 pair production , @xmath97 , has the form @xmath98 the contribution of the isospin @xmath87 to the total cross section of the nucleon pair production reads @xmath99\,.\end{aligned}\ ] ] thus , to describe the energy dependence of the ratio @xmath100 and the cross section @xmath101 in the non - relativistic approximation , it is necessary to know the functions @xmath102 and @xmath103 .
let us write the hamiltonian @xmath104 for the isospin @xmath87 as follows , @xmath105 where @xmath106 is the spin operator for the spin - one system of produced pair , @xmath107 is the radial part of the laplace operator , and @xmath108 denotes the orbital angular momentum . then the radial wave functions @xmath109 and @xmath110 , @xmath111 , satisfy the equations @xmath112 the asymptotic form of the solutions at large distances is , eq .
( [ asvuw ] ) , @xmath113\,,\nonumber\\ & & w^{i}_{11}(r)=-\frac{1}{2ikr}s_{12}^{i1}\,e^{ikr}\,,\nonumber\\ & & u^{i}_{21}(r)=\frac{1}{2ikr}s_{21}^{i1}\,e^{ikr}\,,\nonumber\\ & & w^{i}_{21}(r)=\frac{1}{2ikr}\big[-s_{22}^{i1}e^{ikr}+e^{-ikr}\big]\,.\end{aligned}\ ] ]
it is known that the difference of the cross section of the processes @xmath0 and @xmath26 is small @xcite .
therefore it is natural to suggest that one of the isospin amplitudes is much larger than another one .
1 shows the cross section @xmath114 at @xmath115 ( solid line ) as well as the cross section @xmath27 .
the only free parameter @xmath116 has been found by normalizing the theoretical curve to the data at the third experimental point above threshold . it is seen that the cross section for @xmath28 perfectly reproduces the shape of the data in a wide range of energy near threshold ( from threshold up to @xmath117 gev ) . besides , for this case the value of @xmath118 looks reasonable compared to the value @xmath119 obtained by assumption of isovector dominance .
2 shows the cross section @xmath120 normalized at the same data point .
it is seen that the shape of the theoretical curve does not reproduce the energy behaviour of the measured cross section .
in addition , the value of the fitting parameter @xmath121 does not look reasonable .
3 shows the ratio @xmath122 , see eq.([gegmnr ] ) , together with the experimental data ref.@xcite .
our prediction is in a qualitative agreement with the data .
isoscalar dominance in a model of final - state interaction based on the paris potential has a simple explanation . at very small distances ,
the paris potential corresponding to @xmath123 is strongly repulsive while the potential for @xmath28 is strongly attractive . as a result ,
the wave function for @xmath123 at small distances is strongly suppressed as compare to the case @xmath28 . to conclude , using the paris optical potential
, we show that the isoscalar amplitude dominates in the cross section of the process @xmath124 .
our prediction for the cross section is in good agreement with the data .
the prediction for @xmath122 obtained under assumption of the isoscalar dominance agrees qualitatively with the data . to confirm our statement on the isoscalar dominance ,
other optical potential models should be considered .
the work was supported by the ministry of education and science of the russian federation . | we use the paris nucleon - antinucleon optical potential for explanation of experimental data in the process @xmath0 near threshold .
it turns out that final - state interaction due to paris optical potential allows us to reproduce available experimental data .
it follows from our consideration that the isoscalar form factor is much larger than the isovector one . |
this paper is motivated by an open question on a system of interacting locally regulated diffusions . in @xcite
, a sufficient condition for local extinction is established for such a system . in general , however , there is no criterion available for global extinction , that is , convergence of the total mass process to zero when started in finite total mass .
the method of proof for the local extinction result in @xcite is a comparison with a mean field model @xmath1 which solves @xmath2 where @xmath3 is a standard brownian motion and where @xmath4 are suitable functions satisfying @xmath5 .
this mean field model arises as the limit as @xmath6 ( see theorem 1.4 in @xcite for the case @xmath7 ) of the following system of interacting locally regulated diffusions on @xmath8 islands with uniform migration @xmath9}}}\,dt\\ & + h{{{\bigl(x_t^n(i)\bigr)}}}dt + \sqrt{2g{{{\bigl(x_t^n(i)\bigr)}}}}\,db_t(i)\qquad i=0,\ldots , n-1 . \end { split}\ ] ] for this convergence , @xmath10 may be assumed to be independent and identically distributed with the law of @xmath11 being independent of @xmath8 .
the intuition behind the comparison with the mean field model is that if there is competition ( modeled through the functions @xmath12 and @xmath13 in ) among individuals and resources are everywhere the same , then the best strategy for survival of the population is to spread out in space as quickly as possible .
the results of @xcite cover translation invariant initial measures and local extinction . for general @xmath12 and @xmath13 ,
not much is known about extinction of the total mass process .
let the solution @xmath14 of be started in @xmath15 , @xmath16 .
we prove in a forthcoming paper under suitable conditions on the parameters that the total mass @xmath17 converges as @xmath6 .
in addition , we show in that paper that the limiting process dominates the total mass process of the corresponding system of interacting locally regulated diffusions started in finite total mass .
consequently , a global extinction result for the limiting process would imply a global extinction result for systems of locally regulated diffusions
. in this paper we introduce and study a model which we call _ virgin island model _ and which is the limiting process of @xmath14 as @xmath6 .
note that in the process @xmath14 an emigrant moves to a given island with probability @xmath18 .
this leads to the characteristic property of the virgin island model namely every emigrant moves to an unpopulated island .
our main result is a necessary and sufficient condition ( see below ) for global extinction for the virgin island model .
moreover , this condition is fairly explicit in terms of the parameters of the model .
now we define the model . on the @xmath19-th island
evolves a diffusion @xmath20 with state space @xmath21 given by the strong solution of the stochastic differential equation @xmath22 where @xmath3 is a standard brownian motion .
this diffusion models the total mass of a population and is the diffusion limit of near - critical branching particle processes where both the offspring mean and the offspring variance are regulated by the total population .
later , we will specify conditions on @xmath23 and @xmath13 so that @xmath24 is well - defined .
for now , we restrict our attention to the prototype example of a feller branching diffusion with logistic growth in which @xmath25 , @xmath26 and @xmath27 with @xmath28 .
note that zero is a trap for @xmath24 , that is , @xmath29 implies @xmath30 for all @xmath31 .
mass emigrates from the @xmath19-th island at rate @xmath32 and colonizes unpopulated islands .
a new population should evolve as the process @xmath33 .
thus , we need the law of excursions of @xmath24 from the trap zero . for this , define the set of excursions from zero by @xmath34 , \,\chi_t=0{\ensuremath{\;\;\forall\;}}t\in(-\infty,0]\cup[t_0,\infty)\bigr\}\ ] ] where @xmath35 is the first hitting time of @xmath36 .
the set @xmath37 is furnished with locally uniform convergence . throughout the paper ,
@xmath38 and @xmath39 denote the set of continuous functions and the set of cdlg functions , respectively , between two intervals @xmath40 .
furthermore , define @xmath41 the _ excursion measure _ @xmath42 is a @xmath43-finite measure on @xmath37 .
it has been constructed by pitman and yor @xcite as follows : under @xmath42 , the trajectories come from zero according to an entrance law and then move according to the law of @xmath24 .
further characterizations of @xmath44 are given in @xcite , too .
for a discussion on the excursion theory of one - dimensional diffusions , see @xcite
. we will give a definition of @xmath42 later .
next we construct all islands which are colonized from the @xmath19-th island and call these islands the first generation .
then we construct the second generation which is the collection of all islands which have been colonized from islands of the first generation , and so on .
figure [ f : excursion_tree ] illustrates the resulting tree of excursions . for the generation - wise construction ,
we use a method to index islands which keeps track of which island has been colonized from which island .
an island is identified with a triple which indicates its mother island , the time of its colonization and the population size on the island as a function of time . for @xmath45 ,
let @xmath46 be a possible @xmath19-th island .
for each @xmath47 and @xmath45 , define @xmath48 which we will refer to as the set of all possible islands of the @xmath49-th generation with fixed @xmath19-th island @xmath50 .
this notation should be read as follows .
the island @xmath51 has been colonized from island @xmath52 at time @xmath53 and carries total mass @xmath54 at time @xmath55 .
notice that there is no mass on an island before the time of its colonization .
the island space is defined by @xmath56 denote by @xmath57 the colonization time of island @xmath58 if @xmath59 for some @xmath60 .
furthermore , let @xmath61 be a set of poisson point processes on @xmath62 with intensity measure @xmath63 } } } = a{{{\bigl(\chi(t - s)\bigr)}}}\,dt\otimes { { q}_y}(d\psi)\quad \iota\in{\ensuremath{{{\ensuremath{\mathcal i}}}}}.\ ] ] for later use , let @xmath64 .
we assume that the family @xmath65 is independent for every @xmath45 .
the virgin island model is defined recursively generation by generation .
the @xmath19-th generation only consists of the @xmath19-th island @xmath66 the @xmath67-st generation , @xmath68 , is the ( random ) set of all islands which have been colonized from islands of the @xmath49-th generation @xmath69 the set of all islands is defined by @xmath70 the total mass process of the virgin island model is defined by @xmath71 our main interest concerns the behaviour of the law @xmath72 of @xmath73 as @xmath74 .
the following observation is crucial for understanding the behavior of @xmath75 as @xmath74 .
there is an inherent branching structure in the virgin island model .
consider as new `` time coordinate '' the number of island generations .
one offspring island together with all its offspring islands is again a virgin island model but with the path @xmath33 on the @xmath19-th island replaced by an excursion path .
because of this branching structure , the virgin island model is a multi - type crump - mode - jagers branching process ( see @xcite under `` general branching process '' ) if we consider islands as individuals and @xmath62 as type space .
we recall that a single - type crump - mode - jagers process is a particle process where every particle @xmath76 gives birth to particles at the time points of a point process @xmath77 until its death at time @xmath78 , and @xmath79 are independent and identically distributed .
the literature on crump - mode - jagers processes assumes that the number of offspring per individual is finite in every finite time interval . in the virgin island model , however , every island has infinitely many offspring islands in a finite time interval because @xmath44 is an infinite measure .
the most interesting question about the virgin island model is whether or not the process survives with positive probability as @xmath74 .
generally speaking , branching particle processes survive if and only if the expected number of offspring per particle is strictly greater than one , e.g. the crump - mode - jagers process survives if and only if @xmath80>1 $ ] .
for the virgin island model , the offspring of an island @xmath81 depends on the emigration intensities @xmath82 .
it is therefore not surprising that the decisive parameter for survival is the expected `` sum '' over those emigration intensities @xmath83 we denote the expression in as `` expected total emigration intensity '' of the virgin island model .
the observation that is the decisive parameter plus an explicit formula for leads to the following main result . in theorem
[ thm : extinction_vim ] , we will prove that the virgin island model survives with strictly positive probability if and only if @xmath84 note that the left - hand side of is equal to @xmath85 where @xmath86 is the speed measure of the one - dimensional diffusion .
the method of proof for the extinction result is to study an integral equation ( see lemma [ l : integral_laplace_trafo ] ) which the laplace transform of the total mass @xmath87 solves .
furthermore , we will show in lemma [ l : q_explicit_man_hours ] that the expression in is equal to the left - hand side of .
condition already appeared in @xcite as necessary and sufficient condition for existence of a nontrivial invariant measure for the mean field model , see theorem 1 and lemma 5.1 in @xcite .
thus , the total mass process of the virgin island model dies out if and only if the mean field model dies out .
the following duality indicates why the same condition appears in two situations which seem to be fairly different at first view . if @xmath88 , @xmath89 and @xmath90 with @xmath91 , that is , in the case of feller branching diffusions with logistic growth , then model is dual to itself , see theorem 3 in @xcite .
if @xmath14 indeed approximates the virgin island model as @xmath6 , then for this choice of parameters the total mass process @xmath75 is dual to the mean field model .
this duality would directly imply that in the case of feller branching diffusions with logistic growth global extinction of the virgin island model is equivalent to local extinction of the mean field model .
an interesting quantity of the virgin island process is the area under the path of @xmath87 . in theorem [ thm : expected_man_hours ]
, we prove that the expectation of this quantity is finite exactly in the subcritical situation in which case we give an expression in terms of @xmath92 , @xmath12 and @xmath13 .
in addition , in the critical case and in the supercritical case , we obtain the asymptotic behaviour of the expected area under the path of @xmath87 up to time @xmath93 @xmath94 as @xmath74 for all @xmath16 .
more precisely , the order of is @xmath95 in the critical case . for the supercritical case ,
let @xmath96 be the malthusian parameter defined by @xmath97 it turns out that the expression in grows exponentially with rate @xmath98 as @xmath74 .
the result of theorem [ thm : expected_man_hours ] in the supercritical case suggests that the event that @xmath75 grows exponentially with rate @xmath98 as @xmath74 has positive probability .
however , this is not always true .
theorem [ thm : general : xlogx ] proves that @xmath99 converges in distribution to a random variable @xmath100 .
furthermore , this variable is not identically zero if and only if @xmath101 where @xmath102 .
this @xmath103-criterion is similar to the kesten - stigum theorem ( see @xcite ) for multidimensional galton - watson processes .
our proof follows doney @xcite who establishes an @xmath103-criterion for crump - mode - jagers processes .
our construction introduces as new `` time coordinate '' the number of island generations .
readers being interested in a construction of the virgin island model in the original time coordinate for example in a relation between @xmath73 and @xmath104 are referred to dawson and li ( 2003 ) @xcite . in that paper , a superprocess with dependent spatial motion and interactive immigration
is constructed as the pathwise unique solution of a stochastic integral equation driven by a poisson point process whose intensity measure has as one component the excursion measure of the feller branching diffusion . in a special case ( see equation ( 1.6 ) in @xcite with @xmath105 , @xmath106 and @xmath107}(a)\,da$ ] ) , this is just the virgin island model with replaced by a feller branching diffusion , i.e. @xmath25 , @xmath108 , @xmath27 .
it would be interesting to know whether existence and uniqueness of such stochastic integral equations still hold if the excursion measure of the feller branching diffusion is replaced by @xmath44 .
models with competition have been studied by various authors .
mueller and tribe ( 1994 ) @xcite and horridge and tribe ( 2004 ) @xcite investigate an one - dimensional spde analog of interacting feller branching diffusions with logistic growth which can also be viewed as kpp equation with branching noise .
bolker and pacala ( 1997 ) @xcite propose a branching random walk in which the individual mortality rate is increased by a weighted sum of the entire population .
etheridge ( 2004 ) @xcite studies two diffusion limits hereof .
the `` stepping stone version of the bolker - pacala model '' is a system of interacting feller branching diffusions with non - local logistic growth .
the `` superprocess version of the bolker - pacala model '' is an analog of this in continuous space .
hutzenthaler and wakolbinger @xcite , motivated by @xcite , investigated interacting diffusions with local competition which is an analog of the virgin island model but with mass migrating on @xmath109 instead of migration to unpopulated islands .
the following assumption guarantees existence and uniqueness of a strong @xmath110-valued solution of equation , see e.g. theorem iv.3.1 in @xcite .
assumption [ a : a1 ] additionally requires that @xmath111 is essentially linear .
[ a : a1 ] the three functions @xmath112 , @xmath113 and @xmath114 are locally lipschitz continuous in @xmath110 and satisfy @xmath115 .
the function @xmath13 is strictly positive on @xmath116 .
furthermore , @xmath12 and @xmath117 satisfy the linear growth condition @xmath118 where @xmath119 denotes the maximum of @xmath120 and @xmath121 .
in addition , @xmath122 holds for all @xmath16 and for some constants @xmath123 .
the key ingredient in the construction of the virgin island model is the law of excursions of @xmath33 from the boundary zero .
note that under assumption [ a : a1 ] , zero is an absorbing boundary for , i.e. @xmath29 implies @xmath30 for all @xmath31 . as zero
is not a regular point , it is not possible to apply the well - established it excursion theory .
instead we follow pitman and yor @xcite and obtain a @xmath43-finite measure @xmath124 to be called _ excursion measure _ on @xmath37 ( defined in ) . for this
, we additionally assume that @xmath33 hits zero in finite time with positive probability .
the following assumption formulates a necessary and sufficient condition for this ( see lemma 15.6.2 in @xcite ) . to formulate the assumption ,
we define @xmath125 note that @xmath126 is a scale function , that is , @xmath127 holds for all @xmath128 , see section 15.6 in @xcite .
[ a : a2 ] the functions @xmath92 , @xmath13 and @xmath12 satisfy @xmath129 for some @xmath130 . note that
if assumption [ a : a2 ] is satisfied , then holds for all @xmath130 .
pitman and yor @xcite construct the excursion measure @xmath131 in three different ways one being as follows .
the set of excursions reaching level @xmath132 has @xmath131-measure @xmath133 .
conditioned on this event an excursion follows the diffusion @xmath33 conditioned to converge to infinity until this process reaches level @xmath134 . from this time on the excursion follows an independent unconditioned process .
we carry out this construction in detail in section [ sec : excursions_from_a_trap_of_one_dimensional_diffusions ] .
in addition pitman and yor @xcite describe the excursion measure `` in a preliminary way as '' @xmath135 where the limit indicates weak convergence of finite measures on @xmath136 away from neighbourhoods of the zero - trajectory .
however , they do not give a proof .
having @xmath131 identified as the limit in will enable us to transfer explicit formulas for @xmath137 to explicit formulas for @xmath131 .
we establish the existence of the limit in in theorem [ thm : existence_excursion_measure ] below . for this , let the topology on @xmath136 be given by locally uniform convergence . furthermore ,
recall @xmath24 from , the definition of @xmath37 from and the definition of @xmath126 from .
[ thm : existence_excursion_measure ] assume [ a : a1 ] and [ a : a2 ] .
then there exists a @xmath43-finite measure @xmath124 on @xmath37 such that @xmath138 for all bounded continuous @xmath139 for which there exists an @xmath140 such that @xmath141 whenever @xmath142 . for our proof of the global extinction result for the virgin island model , we need the scaling function @xmath126 in to behave essentially linearly in a neighbourhood of zero .
more precisely , we assume @xmath143 to exist in @xmath116 . from definition of @xmath126
it is clear that a sufficient condition for this is given by the following assumption .
[ a : s_bar ] the integral @xmath144 has a limit in @xmath145 as @xmath146 .
it follows from dominated convergence and from the local lipschitz continuity of @xmath92 and @xmath12 that assumption [ a : s_bar ] holds if @xmath147 is finite .
in addition , we assume that the expected total emigration intensity of the virgin island model is finite .
lemma [ l : finite_man_hours ] shows that , under assumptions [ a : a1 ] and [ a : a2 ] , an equivalent condition for this is given in assumption [ a : finite_man_hours ] .
[ a : finite_man_hours ] the functions @xmath92 , @xmath13 and @xmath12 satisfy @xmath148 for some and then for all @xmath130 . we mention that if assumptions [ a : a1 ] , [ a : a2 ] and [ a : finite_man_hours ] hold , then the process @xmath24 hits zero in finite time almost surely ( see lemma [ l : finite_time_extinction ] and lemma [
l : finite_man_hours ] ) .
furthermore , we give a generic example for @xmath92 , @xmath12 and @xmath13 namely @xmath149 , @xmath150 , @xmath151 with @xmath152 .
the assumptions [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] are all satisfied if @xmath153 and if @xmath154 .
assumption [ a : a2 ] is not met by @xmath25 , @xmath155 , @xmath156 and @xmath157 because then @xmath158 , @xmath159 and condition fails to hold .
next we formulate the main result of this paper .
theorem [ thm : extinction_vim ] proves a nontrivial transition from extinction to survival . for the formulation of this result ,
we define @xmath160 which is well - defined under assumption [ a : s_bar ] .
note that @xmath161 .
define the excursion measure @xmath162 and recall the total mass process @xmath75 from .
[ thm : extinction_vim ] assume [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] .
then the total mass process @xmath75 started in @xmath130 dies out ( i.e. , converges in probability to zero as @xmath74 ) if and only if @xmath163 if fails to hold , then @xmath73 converges in distribution as @xmath74 to a random variable @xmath164 satisfying @xmath165 for all @xmath16 and some @xmath166 .
the constant @xmath166 is the unique strictly positive fixed - point of a function defined in lemma [ l : fixed_point ] . in the critical case , that is , equality in
, @xmath73 converges to zero in distribution as @xmath74 .
however , it turns out that the expected area under the graph of @xmath87 is infinite .
in addition , we obtain in theorem [ thm : expected_man_hours ] the asymptotic behaviour of the expected area under the graph of @xmath87 up to time @xmath93 as @xmath74 . for this , define @xmath167 and similarly @xmath168 with @xmath169 .
if assumptions [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] hold , then @xmath170 is finite for fixed @xmath171 ; see lemma [ l : finite_man_hours ] . furthermore , under assumptions [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] , @xmath172 by the dominated convergence theorem .
[ thm : expected_man_hours ] assume [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] . if the left - hand side of is strictly smaller than one , then the expected area under the path of @xmath87 is equal to @xmath173 for all @xmath16 .
otherwise , the left - hand side of is infinite . in the critical case , that is , equality in , @xmath174 where the right - hand side is interpreted as zero if the denominator is equal to infinity . in the supercritical case ,
i.e. , if fails to be true , let @xmath96 be such that @xmath175 then the order of growth of the expected area under the path of @xmath176 up to time @xmath93 as @xmath74 can be read off from @xmath177 for all @xmath16 .
the following result is an analog of the kesten - stigum theorem , see @xcite . in the supercritical case
, @xmath99 converges to a random variable @xmath178 as @xmath74 .
in addition , @xmath178 is not identically zero if and only if the @xmath103-condition holds .
we will prove a more general version hereof in theorem [ thm : general : xlogx ] below .
unfortunately , we do not know of an explicit formula in terms of @xmath92 , @xmath12 and @xmath13 for the left - hand side of . aiming at a condition which is easy to verify , we assume instead of that the second moment @xmath179 is finite . in assumption [ a : finite_man_hours_squared ]
, we formulate a condition which is slightly stronger than that , see lemma [ l : q_explicit_man_hours ] below .
[ a : finite_man_hours_squared ] the functions @xmath92 , @xmath13 and @xmath12 satisfy @xmath180 for some and then for all @xmath130 .
[ thm : xlogx ] assume [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours_squared ] .
suppose that fails to be true ( supercritical case ) and let @xmath96 be the unique solution of .
then @xmath181 in the weak topology and @xmath182 .
theorem [ thm : existence_excursion_measure ] will be established in section [ sec : excursions_from_a_trap_of_one_dimensional_diffusions ] . note that section [ sec : excursions_from_a_trap_of_one_dimensional_diffusions ] does not depend on the sections [ sec : random_characteristics]-[sec : convergence_supercritical ] .
we will prove the survival and extinction result of theorem [ thm : extinction_vim ] in two steps .
in the first step , we obtain a criterion for survival and extinction in terms of @xmath44 .
more precisely , we prove that the process dies out if and only if the expression in is smaller than or equal to one . in this step
, we do not exploit that @xmath44 is the excursion measure of @xmath24 .
in fact , we will prove an analog of theorem [ thm : extinction_vim ] in a more general setting where @xmath44 is replaced by some @xmath43-finite measure @xmath183 and where the islands are counted with random characteristics .
see section [ sec : random_characteristics ] below for the definitions .
the analog of theorem [ thm : extinction_vim ] is stated in theorem [ thm : general : extinction ] , see section [ sec : random_characteristics ] , and will be proven in section [ sec : extinction_vim ] . the key equation for its proof is contained in lemma [ l : tree_structure ] which formulates the branching structure in the virgin island model . in the second step ,
we calculate an expression for in terms of @xmath184 and @xmath13 .
this will be done in lemma [
l : q_explicit_man_hours ] .
theorem [ thm : extinction_vim ] is then a corollary of theorem [ thm : general : extinction ] and of lemma [ l : q_explicit_man_hours ] , see section [ sec : proof_main_theorems ] .
similarly , a more general version of theorem [ thm : expected_man_hours ] is stated in theorem [ thm : general : expected_man_hours ] , see section [ sec : random_characteristics ] below .
the proofs of theorem [ thm : expected_man_hours ] and of theorem [ thm : general : expected_man_hours ] are contained in section [ sec : proof_main_theorems ] and section [ sec : proof_of_theorem_expected_man_hours ] , respectively . as mentioned in section [ sec : introduction ] , a rescaled version of @xmath75 converges in the supercritical case .
this convergence is stated in a more general formulation in theorem [ thm : general : xlogx ] , see section [ sec : random_characteristics ] below .
the proofs of theorem [ thm : xlogx ] and of theorem [ thm : general : xlogx ] are contained in section [ sec : proof_main_theorems ] and in section [ sec : convergence_supercritical ] , respectively .
in the proof of the extinction result of theorem [ thm : extinction_vim ] , we exploit that one offspring island together with all its offspring islands is again a virgin island model but with a typical excursion instead of @xmath24 on the @xmath19-th island . for the formulation of this branching property
, we need a version of the virgin island model where the population on the @xmath19-th island is governed by @xmath44 .
more generally , we replace the law @xmath137 of the first island by some measure @xmath185 and we replace the excursion measure @xmath44 by some measure @xmath183 . given two @xmath43-finite measures @xmath185 and @xmath183 on the borel-@xmath43-algebra of @xmath186 , we define the _ virgin island model with initial island measure @xmath185 and excursion measure @xmath183 _ as follows .
define the random sets of islands @xmath187 , @xmath68 , and @xmath188 through the definitions , , and with @xmath137 and @xmath44 replaced by @xmath185 and @xmath183 , respectively .
a simple example for @xmath185 and @xmath183 is @xmath189 where @xmath190 is a random variable and @xmath191 is the dirac measure on the path @xmath192
. then the virgin island model coincides with a crump - mode - jagers process in which a particle has offspring according to a rate @xmath193 poisson process until its death at time @xmath194 .
furthermore , our results do not only hold for the total mass process but more generally when the islands are counted with random characteristics .
this concept is well - known for crump - mode - jagers processes , see section 6.9 in @xcite .
assume that @xmath195 , @xmath196 , are separable and nonnegative processes with the following properties .
it vanishes on the negative half - axis , i.e. @xmath197 for @xmath198 . informally speaking our main assumption on @xmath199
is that it does not depend on the history .
formally we assume that @xmath200 furthermore , we assume that the family @xmath201 is independent for each @xmath45 and @xmath202 is measurable . as a short notation , define @xmath203 for @xmath45 . with this , we define @xmath204 and say that @xmath205 is a _ virgin island process counted with random characteristics _
@xmath206 . instead of @xmath207 ,
we write @xmath208 for a path @xmath45 and note that @xmath209 is measurable .
a prominent example for @xmath210 is the deterministic random variable @xmath211 . in this case
, @xmath212 is the total mass of all islands at time @xmath93 .
notice that @xmath75 defined in is a special case hereof , namely @xmath213 .
another example for @xmath210 is @xmath214 .
then @xmath215 is the total mass at time @xmath93 of all islands which have been colonized in the last @xmath216 time units .
if @xmath217 , then @xmath218 . as in section
[ sec : main_results ] , we need an assumption which guarantees finiteness of @xmath219 .
[ a : general : finite_moments ] the function @xmath112 is continuous and there exist @xmath123 such that @xmath220 for all @xmath16 .
furthermore , @xmath221 for every @xmath222 the analog of assumption [ a : finite_man_hours ] in the general setting is the following assumption .
[ a : general : finite_man_hours ] both the expected emigration intensity of the @xmath19-th island and of subsequent islands are finite : @xmath223 in section [ sec : main_results ] , we assumed that @xmath33 hits zero in finite time with positive probability .
see assumption [ a : a2 ] for an equivalent condition .
together with [ a : finite_man_hours ] , this assumption implied almost sure convergence of @xmath33 to zero as @xmath74 . in the general setting , we need a similar but somewhat weaker assumption .
more precisely , we assume that @xmath224 converges to zero in distribution both with respect to @xmath185 and with respect to @xmath183 .
[ a : convergence_to_zero ] the random processes @xmath225 and the measures @xmath183 and @xmath185 satisfy @xmath226 for all @xmath227 . having introduced the necessary assumptions , we now formulate the extinction and survival result of theorem [ thm : extinction_vim ] in the general setting .
[ thm : general : extinction ] let @xmath185 be a probability measure on @xmath186 and let @xmath183 be a measure on @xmath186 .
assume [ a : general : finite_moments ] , [ a : general : finite_man_hours ] and [ a : convergence_to_zero ] .
then the virgin island process @xmath228 counted with random characteristics @xmath206 with @xmath19-th island distribution @xmath185 and with excursion measure @xmath183 dies out ( i.e. , converges to zero in probability ) if and only if @xmath229 in case of survival , the process converges weakly as @xmath74 to a probability measure @xmath230 with support in @xmath231 which puts mass @xmath232 on the point @xmath233 where @xmath166 is the unique strictly positive fixed - point of @xmath234 the assumption on @xmath185 to be a probability measure is convenient for the formulation in terms of convergence in probability . for a formulation in the case of a @xmath43-finite measure @xmath185 , see the proof of the theorem in section [ sec : extinction_vim ] .
next we state theorem [ thm : expected_man_hours ] in the general setting . for its formulation ,
define @xmath235 and similarly @xmath236 with @xmath185 replaced by @xmath183 .
[ thm : general : expected_man_hours ] assume [ a : general : finite_moments ] and [ a : general : finite_man_hours ] . if the left - hand side of is strictly smaller than one and if both @xmath237 and @xmath236 are integrable , then @xmath238}}}\nu(d\chi ) = \int_0^\infty \!\!\!f^\nu(s)ds + \frac{\int_0^\infty f^q(s)\,ds\,\int \int_0^\infty a{{{\bigl(\chi_s\bigr)}}}\,ds\nu(d\chi ) } { 1-\int { { { \bigl(\int_0^\infty a{{{\bigl(\chi_s\bigr)}}}\,ds\bigr)}}}q(d\chi)}\ ] ] which is finite and strictly positive .
otherwise , the left - hand side of is infinite .
if the left - hand side of is equal to one and if both @xmath237 and @xmath236 are integrable , @xmath239}}}\nu(d\chi ) = \frac{\int_0^\infty f^q(s)\,ds\,{\ensuremath{{\displaystyle \cdot}}}\int \int_0^\infty a{{{\bigl(\chi_s\bigr)}}}\,ds\,\nu(d\chi ) } { \int_0^\infty s\int a{{{\bigl(\chi_s\bigr)}}}q(d\chi)\,ds } <
\infty\ ] ] where the right - hand side is interpreted as zero if the denominator is equal to infinity . in the supercritical case , i.e. , if fails to be true , let @xmath240 be such that @xmath241 additionally assume that @xmath236 is continuous a.e . with respect to the lebesgue measure , @xmath242 and that @xmath243 as @xmath74 .
then the order of convergence of the expected total intensity up to time @xmath93 can be read off from @xmath244}}}\nu(d\chi ) = \frac{1}{{{\ensuremath{\alpha}}}}\,{{\ensuremath{{\displaystyle \lim_{t { \rightarrow}\infty}}}}}e^{-{{\ensuremath{\alpha}}}t}\int { { \ensuremath{\mathbf{e}}}}{{{\bigl[v_t^{\phi,\chi , q}\bigr]}}}\nu(d\chi)\ ] ] and from @xmath245}}}\nu(d\chi ) = \frac{\int_0^\infty e^{-{{\ensuremath{\alpha}}}s}f^q(s)\,ds \,{\ensuremath{{\displaystyle \cdot}}}\int_0^\infty e^{-{{\ensuremath{\alpha}}}s}\int a{{{\bigl(\chi_s\bigr)}}}\nu(d\chi)\,ds } { \int_0^\infty s e^{-{{\ensuremath{\alpha}}}s}\int a{{{\bigl(\chi_s\bigr)}}}q(d\chi)\,ds}.\ ] ] for the formulation of the analog of the kesten - stigum theorem , denote by @xmath246 the right - hand side of with @xmath185 replaced by @xmath183 .
furthermore , define @xmath247 for every path @xmath45 . for our proof of theorem [ thm : general : xlogx ] , we additionally assume the following properties of @xmath183 . [ a : for_skdist ] the measure @xmath183 satisfies @xmath248 for every @xmath222 and @xmath249}}}q(d\chi)<\infty,\quad \sup_{t\geq0}\int{{\ensuremath{\mathbf{e}}}}{{{\bigl(\phi_\chi^2(t)\bigr ) } } } q(d\chi)<\infty.\ ] ] [ thm : general : xlogx ] let @xmath185 be a probability measure on @xmath186 and let @xmath183 be a measure on @xmath186 .
assume [ a : general : finite_moments ] , [ a : general : finite_man_hours ] , [ a : convergence_to_zero ] and [ a : for_skdist ] .
suppose that @xmath250 ( supercritical case ) and let @xmath96 be the unique solution of .
then @xmath251 in the weak topology where @xmath178 is a nonnegative random variable .
the variable @xmath178 is not identically zero if and only if @xmath252 where @xmath102 .
if holds , then @xmath253}}}\nu(d\chi ) , { { \ensuremath{\mathbf{p}}}}{{{\bigl(w=0\bigr)}}}=\int{{{\bigl[e^{-q\int_0^\infty a{{{(\chi_s)}}}\,ds}\bigr]}}}\nu(d\chi ) \end{split}\ ] ] where @xmath166 is the unique strictly positive fixed - point of . comparing with
, we see that @xmath254 . consequently , the virgin island process @xmath205 conditioned on not converging to zero grows exponentially fast with rate @xmath98 as @xmath74 .
we mentioned in the introduction that there is an inherent branching structure in the virgin island model .
one offspring island together with all its offspring islands is again a virgin island model but with a typical excursion instead of @xmath24 on the @xmath19-th island . in lemma
[ l : tree_structure ] , we formalize this idea . as a corollary thereof , we obtain an integral equation for the modified laplace transform of the virgin island model in lemma [ l : integral_laplace_trafo ] which is the key equation for our proof of the extinction result of theorem [ thm : extinction_vim ] .
recall the notation of section [ sec : introduction ] and of section [ sec : random_characteristics ] .
[ l : tree_structure ] let @xmath255 .
there exists an independent family @xmath256 of random variables which is independent of @xmath210 and of @xmath257 such that @xmath258 and such that @xmath259 for all @xmath260 .
write @xmath261 and @xmath262 .
define @xmath263 for @xmath260 where @xmath264 for @xmath47 . comparing and with
, we see that @xmath265 define @xmath266 for @xmath55 and for @xmath47 @xmath267 summing over @xmath68 we obtain for @xmath55 @xmath268 this is equality .
independence of the family follows from independence of @xmath269 and from independence of @xmath270 .
it remains to prove . because of assumption the random characteristics @xmath199 only depends on the last part of @xmath58 .
therefore @xmath271 summing over @xmath47 results in and finishes the proof . in order to increase readability ,
we introduce the following suggestive symbolic abbreviation @xmath272 } } } : = \int{{\ensuremath{\mathbf{e}}}}f{{{\bigl ( v_t^{\phi,\chi , q}\bigr)}}}\nu(d\chi ) \quad t\geq0 , f\in{{\ensuremath{\mathbf{c}}}}{{{\bigl([0,\infty),[0,\infty)\bigr)}}}.\ ] ] one might want to read this as `` expectation '' with respect to a non - probability measure . however , is not intended to define an operator .
the following lemma proves that the virgin island model counted with random characteristics as defined in is finite .
[ l : general : finite_moments ] assume [ a : general : finite_moments ] .
then , for every @xmath222 , @xmath273 } } } < \infty.\ ] ] furthermore , if @xmath274 then there exists a constant @xmath275 such that @xmath276}}}}\\ & \leq c_t{{{\bigl(1+\sup_{t\leq t}\int { { \ensuremath{\mathbf{e}}}}{{{\bigl(\phi_\chi^2(t)\bigr)}}}\ , ( \nu+q)(d\chi ) + \int{{{\bigl(\int_0^t a(\chi_s)ds\bigr)}}}^2\nu(d\chi)\bigr ) } } } \end { split}\ ] ] for all @xmath185 and the right - hand side of is finite in the special case @xmath277 . we exploit the branching property formalized in lemma [ l : tree_structure ] and apply gronwall s inequality . recall @xmath278 from the proof of lemma [ l : tree_structure ] . the two equalities and imply @xmath279 } } } = \int { { \ensuremath{\mathbf{e}}}}\phi_\chi(t)\,\nu(d\chi)\leq \sup_{s\leq t } \int { { \ensuremath{\mathbf{e}}}}\phi_\chi(s)\,\nu(d\chi)\ ] ] for @xmath280 and for @xmath47 @xmath281 } } } = \int { { \ensuremath{\mathbf{e}}}}{{{\bigl[\sum_{(s,\psi)\in\pi^{\chi } } { { \ensuremath{\mathbf{e}}}}{{{\bigl[v_{t - s}^{(n-1),\phi,\psi , q}\bigr]}}}\bigr]}}}\nu(d\chi)\\ & = \int { { { \biggl(\int_0^t\int { { \ensuremath{\mathbf{e}}}}{{{\bigl[v_{t - s}^{(n-1),\phi,\psi , q}\bigr]}}}q(d\psi ) a{{{(\chi_s)}}}ds\biggr)}}}\,\nu(d\chi)\\ & \leq \sup_{u\leq t}\int a{{{(\chi_u)}}}\nu(d\chi)\int_0^t { { \ensuremath{\mathbf{i}}}}{{{\bigl[v_{s}^{(n-1),\phi , q , q}\bigr ] } } } \,ds . \end { split}\ ] ] using assumption [ a : general : finite_moments ] induction on @xmath68 shows that all expressions in and in are finite in the case @xmath277 . summing over @xmath282 we obtain @xmath283 } } }
\leq \int{{\ensuremath{\mathbf{e}}}}\phi_\chi(u)\nu(d\chi ) + \int_0^t\sum_{n=0}^{n_0}{{\ensuremath{\mathbf{i}}}}{{{\bigl[v_s^{(n),\phi , q , q}\bigr]}}}\int a{{{(\chi_{t - s})}}}\nu(d\chi)\,ds \end { split}\ ] ] for @xmath280 . in the special case @xmath277 gronwall
s inequality implies @xmath284}}}\leq \sup_{u\leq t}\int{{\ensuremath{\mathbf{e}}}}\phi_\chi(u)q(d\chi ) { \ensuremath{{\displaystyle \cdot}}}\exp{{{\bigl ( t \sup_{u\leq t}\int a{{{(\chi_u)}}}q(d\chi)\bigr)}}}.\ ] ] summing over @xmath282 , inserting into and letting @xmath285 we see that follows from assumption [ a : general : finite_moments ] . for the proof of , note that with @xmath286 and imply @xmath287 for some @xmath288 .
in addition the two equalities and together with independence imply @xmath289 for @xmath55 and for @xmath47 @xmath290 in the special case @xmath277 induction on @xmath68 together with shows that all involved expressions are finite . a similar estimate as in leads to @xmath291}}}\nu(d\chi ) -\int\bigl({{\ensuremath{\mathbf{e}}}}\sum_{n=0}^{n_0 } v_t^{(n),\phi,\chi , q}\bigr)^2\nu(d\chi)}\\ & = \int { \operatorname{var}}{{{\bigl(\phi_\chi(t)\bigr)}}}+{{\ensuremath{\mathbf{e}}}}{{{\biggl(\sum_{(s,\psi)\in\pi^\chi } { \operatorname{var}}{{{\bigl(\sum_{n=1}^{n_0}v_{t - s}^{(n-1),\phi,\psi , q}\bigr)}}}\biggr)}}}\,\nu(d\chi)\\ & = \int { \operatorname{var}}{{{\bigl(\phi_\chi(t)\bigr)}}}+\int_0^t { { { \bigl(a(\chi_s ) \int { \operatorname{var}}{{{\bigl(\sum_{n=0}^{n_0 - 1}v_{t - s}^{(n),\phi,\psi , q}\bigr ) } } } q(d\psi)\bigr ) } } } ds\,\nu(d\chi)\\ & \leq \int{{\ensuremath{\mathbf{e}}}}{{{\bigl(\phi_\chi^2(t)\bigr)}}}\nu(d\chi)+ \int_0^t \int { { \ensuremath{\mathbf{e}}}}{{{\bigl[\bigl(\sum_{n=0}^{n_0}v_{s}^{(n),\phi,\psi , q}\bigr)^2\bigr ] } } } q(d\psi ) ds { \ensuremath{{\displaystyle \cdot}}}\sup_{u\leq t}\int a(\chi_u)\,\nu(d\chi ) .
\end { split}\ ] ] in the special case @xmath277 gronwall s inequality together with leads to @xmath292}}}q(d\chi)}\\ & \leq{{{\biggl(\int 3{{\ensuremath{\mathbf{e}}}}{{{\bigl(\phi_\chi^2 ( t)\bigr ) } } } + { { \ensuremath{\tilde{c}}}}_t\bigl(\int_0^t a(\chi_s)\,ds\bigr)^2q(d\chi)\biggr ) } } } \exp{{{\bigl(\sup_{u\leq t } \int a(\chi_u)\,q(d\chi)t\bigr ) } } } \end{split}\ ] ] which is finite by assumption [ a : general : finite_moments ] and assumption .
inserting into and letting @xmath285 finishes the proof . in the following lemma
, we establish an integral equation for the modified laplace transform of the virgin island model .
recall the definition of @xmath219 from .
[ l : integral_laplace_trafo ] assume [ a : general : finite_moments ] .
the modified laplace transform @xmath293}}}$ ] of the virgin island model counted with random characteristics @xmath206 satisfies @xmath294}}}}\\ & = \int{{\ensuremath{\mathbf{e}}}}{{{\bigl[1- \exp{{{\bigl(-{{\ensuremath{\lambda}}}\phi_{\chi}(t)-\int_0^\infty { { \ensuremath{\mathbf{i}}}}{{{\bigl[1-e^{-{{\ensuremath{\lambda}}}v^{\phi , q , q}_{t - s}}\bigr ] } } } a{{{(\chi_s)}}}\,ds\bigr)}}}\bigr]}}}\nu(d\chi ) \end { split}\ ] ] for all @xmath295 .
fix @xmath295 .
applying lemma [ l : tree_structure ] , @xmath296}}}}\\ & = \int{{{\bigl[1-{{\ensuremath{\mathbf{e}}}}{{{\bigl(e^{-{{\ensuremath{\lambda}}}\phi_\chi(t)}\bigr)}}}{\ensuremath{{\displaystyle \cdot}}}{{\ensuremath{\mathbf{e}}}}{{{\bigl(\prod_{(s,\psi)\in\pi^\chi } { { \ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}v^{\phi,\psi , q}_{t - s}}\bigr)}}}\bigr ] } } } \nu(d\chi)\\ & = \int{{{\bigl[1-{{\ensuremath{\mathbf{e}}}}{{{\bigl(e^{-{{\ensuremath{\lambda}}}\phi_\chi(t)}\bigr)}}}{\ensuremath{{\displaystyle \cdot}}}\exp{{{\bigl(-\int_0^\infty\int 1- { { \ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}v^{\phi,\psi , q}_{t - s}}q(d\psi)a{{{(\chi_s)}}}\,ds\bigr)}}}\bigr ] } } } \nu(d\chi)\\ & = \int{{\ensuremath{\mathbf{e}}}}{{{\bigl[1- \exp{{{\bigl(-{{\ensuremath{\lambda}}}\phi_{\chi}(t)-\int_0^\infty { { \ensuremath{\mathbf{i}}}}{{{\bigl[1-e^{-{{\ensuremath{\lambda}}}v^{\phi , q , q}_{t - s}}\bigr ] } } } a{{{(\chi_s)}}}\,ds\bigr)}}}\bigr]}}}\nu(d\chi ) .
\end { split}\ ] ] this proves the assertion .
recall the definition of @xmath228 from , @xmath237 from and the notation @xmath297 from .
we begin with the supercritical case and let @xmath96 be the malthusian parameter which is the unique solution of .
define @xmath298}}}\qquad \mu^\nu(ds):=\int a{{{\bigl(\chi_s\bigr ) } } } \,\nu(d\chi)\,ds\ ] ] for @xmath55 . in this notation ,
equation with @xmath185 replaced by @xmath183 reads as @xmath299 this is a renewal equation for @xmath300 . by definition of @xmath98 , @xmath301 is a probability measure . from lemma [ l :
general : finite_moments ] we know that @xmath302 is bounded on finite intervals . by assumption , @xmath236 is continuous lebesgue - a.e . and satisfies .
hence , we may apply standard renewal theory ( e.g. theorem 5.2.6 of @xcite ) and obtain @xmath303 multiply equation by @xmath304 , recall @xmath243 as @xmath74 and apply the dominated convergence theorem together with [ a : general : finite_man_hours ] to obtain @xmath305 insert to obtain equation .
an immediate consequence of the existence of the limit on the left - hand side of is equation @xmath306 where we used the dominated convergence theorem .
next we consider the subcritical and the critical case
. define @xmath307}}}\,ds,\quad t\geq0.\ ] ] in this notation , equation integrated over @xmath308 $ ] reads as @xmath309 in the subcritical case , @xmath236 and @xmath237 are integrable .
theorem 5.2.9 in @xcite applied to with @xmath185 replaced by @xmath183 implies @xmath310 letting @xmath74 in , dominated convergence and @xmath311 imply @xmath312 inserting results in . in the critical case , similar arguments lead to @xmath313 the last equality follows from with @xmath185 replaced by @xmath183 and corollary 5.2.14 of @xcite with @xmath314 , @xmath315 and @xmath316 .
note that the assumption @xmath317 of this corollary is not necessary for this conclusion .
recall the definition of @xmath228 from and the notation @xmath297 from . as we pointed out in section
[ sec : main_results ] , the expected total emigration intensity of the virgin island model plays an important role .
the following lemma provides us with some properties of the modified laplace transform of the total emigration intensity .
these properties are crucial for our proof of theorem [ thm : general : extinction ] .
[ l : fixed_point ] assume [ a : general : finite_man_hours ] . then the function @xmath318 is concave with at most two fixed - points .
zero is the only fixed - point if and only if @xmath319 denote by @xmath320 the maximal fixed - point .
then we have for all @xmath321 : @xmath322 if @xmath323 for @xmath183-a.a .
@xmath324 , then @xmath325 and zero is the only fixed - point .
for the rest of the proof , we assume w.l.o.g . that @xmath326 .
the function @xmath327 has finite values because of @xmath328 , @xmath329 , and assumption [ a : general : finite_man_hours ] .
concavity of @xmath327 is inherited from the concavity of @xmath330 , @xmath329 . using dominated convergence together with assumption [ a : general : finite_man_hours ]
, we see that @xmath331 in addition , dominated convergence together with assumption [ a : general : finite_man_hours ] implies @xmath332}}}q(d\chi)\,\quad z\geq0.\ ] ] hence , @xmath327 is strictly concave .
thus , @xmath327 has a fixed - point which is not zero if and only if @xmath333 . the implications and follow from the strict concavity of @xmath327 .
the method of proof ( cf .
section 6.5 in @xcite ) of the extinction result for a crump - mode - jagers process @xmath334 is to study an equation for @xmath335 .
the laplace transform @xmath336 converges monotonically to @xmath337 as @xmath338 , @xmath55 .
furthermore , @xmath339 converges monotonically to the extinction probability @xmath340 as @xmath74 .
taking monotone limits in the equation for @xmath335 results in an equation for the extinction probability .
in our situation , there is an equation for the modified laplace transform @xmath341 as defined in below .
however , the monotone limit of @xmath342 as @xmath338 might be infinite .
thus , it is not clear how to transfer the above method of proof .
the following proof of theorem [ thm : extinction_vim ] directly establishes the convergence of the modified laplace transform .
recall @xmath320 from lemma [ l : fixed_point ] . in the first step , we will prove @xmath343 } } } \to q{\ensuremath{\qquad(\text{as } t\to\infty})}\ ] ] for all @xmath344 . set @xmath345 .
it follows from lemma [
l : general : finite_moments ] that @xmath346 is bounded for every finite @xmath347 .
lemma [ l : integral_laplace_trafo ] with @xmath185 replaced by @xmath183 provides us with the fundamental equation @xmath348}}}q(d\chi)\quad{\ensuremath{\;\;\forall\;}}t\geq0.\ ] ] based on , the idea for the proof of is as follows .
the term @xmath349 vanishes as @xmath74 .
if @xmath350 converges to some limit , then the limit has to be a fixed - point of the function @xmath351}}}q(d\chi).\ ] ] by lemma [ l : fixed_point ] , this function is ( typically strictly ) concave .
therefore , it has exactly one attracting fixed - point
. furthermore , this fact forces @xmath350 to converge as @xmath74 .
we will need finiteness of @xmath352 .
looking for a contradiction , we assume @xmath353
. then there exists a sequence @xmath354 with @xmath355 such that @xmath356 .
we estimate @xmath357}}}q(d\chi)\\ & \leq k{{{(l_{t_n}\!+1)}}}+\int \exp{{{\bigl(- \int_0^\infty a{{{\bigl(\chi_s\bigr ) } } } l_{t_n}\,ds\bigr ) } } } { { { \bigl(1-{{\ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}\phi_\chi(t_n)}\bigr)}}}q(d\chi)\\ & \leq k{{{(l_{t_n}\!+1)}}}+\int { { { \bigl(1-{{\ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}\phi_\chi(t_n)}\bigr)}}}q(d\chi ) .
\end { split}\ ] ] the last summand converges to zero by assumption [ a : convergence_to_zero ] and is therefore bounded by some constant @xmath358 .
inequality leads to the contradiction @xmath359 the last equation is a consequence of and the assumption @xmath353 .
next we prove @xmath360 using boundedness of @xmath361 .
let @xmath354 be a sequence such that @xmath362 .
then a calculation as in results in @xmath363 } } } q(d\chi)\\ & \qquad+{{\ensuremath{{\displaystyle \limsup_{n { \rightarrow}\infty}}}}}\int { { { \bigl(1-{{\ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}\phi_\chi(t_n)}\bigr)}}}q(d\chi ) .
\end { split}\ ] ] the last summand is equal to zero by assumption [ a : convergence_to_zero ] .
the first summand on the right - hand side of is dominated by @xmath364 which is finite by boundedness of @xmath361 and by assumption [ a : general : finite_man_hours ] . applying dominated convergence
, we conclude that @xmath365 is bounded by @xmath366}}}q(d\chi ) = k{{{\bigl(l_\infty\bigr)}}}.\ ] ] thus , lemma [ l : fixed_point ] implies @xmath367 .
assume @xmath166 and suppose that @xmath368 .
let @xmath354 be such that @xmath369 as @xmath370 and @xmath371 . by lemma [ l : fixed_point ]
, there is an @xmath372 and a @xmath373 such that @xmath374 .
we estimate @xmath375}}}q(d\chi)\\ & \geq\int{{{\bigl[1-\exp{{{\bigl ( - c\int_0^{t_{n_0}}a{{{\bigl(\chi_s\bigr ) } } } l_{t_n}\,ds\bigr)}}}\bigr]}}}q(d\chi)\quad{\ensuremath{\;\;\forall\;}}n > n_0 . \end { split}\ ] ] using dominated convergence , the assumption @xmath376 results in the contradiction @xmath377 } } }
q(d\chi)\\ & = c\int{{{\bigl(\int_0^{t_{n_0 } } a{{{\bigl(\chi_s\bigr)}}}\,ds \bigr)}}}q(d\chi)>1 .
\end { split}\ ] ] in order to prove @xmath378 , let @xmath354 be such that @xmath379 .
an estimate as above together with dominated convergence yields @xmath380}}}q(d\chi)\\ & = \int{{{\bigl[1-\exp{{{\bigl ( -\int_{0}^\infty a{{{\bigl(\chi_s\bigr)}}}{{\ensuremath{{\displaystyle \liminf_{t { \rightarrow}\infty}}}}}\ , l_t\,ds\bigr)}}}\bigr]}}}q(d\chi ) = k(m ) . \end { split}\ ] ] therefore ,
lemma [ l : fixed_point ] implies @xmath381 , which yields . finally , we finish the proof of theorem [ thm : general : extinction ] . applying lemma [ l : integral_laplace_trafo ] , we see that @xmath382 } } } -\int{{{\bigl[1-\exp{{{\bigl(-q\int_0^\infty a(\chi_s)\,ds\bigr)}}}\bigr]}}}\nu(d\chi)\bigr|}}}}\\ & \leq\int\exp{{{\bigl(-\int_0^\infty l_{t - s}a{{{(\chi_s)}}}\,ds\bigr ) } } } { { \ensuremath{\mathbf{e}}}}{{{\bigl[1- e^{-{{\ensuremath{\lambda}}}\phi_\chi(t)}\bigr]}}}\nu(d\chi)\\ & \qquad\,+\left|\int{{{\bigl[1-\exp{{{\bigl(-\int_0^\infty l_{t - s } a{{{(\chi_s)}}}\,ds\bigr)}}}\bigr]}}}\nu(d\chi)\right.\\ & \qquad\ , -\left.\int{{{\bigl[1-\exp{{{\bigl(-q\int_0^\infty a{{{(\chi_s)}}}\,ds\bigr)}}}\bigr ] } } } \nu(d\chi)\right| .
\end { split}\ ] ] the first summand on the right - hand side of converges to zero as @xmath74 by assumption [ a : convergence_to_zero ] . by the first step , @xmath383 as @xmath74 .
hence , by the dominated convergence theorem and assumption [ a : general : finite_man_hours ] , the left - hand side of converges to zero as @xmath74 .
as @xmath185 is a probability measure by assumption , we conclude @xmath384 this implies theorem [ thm : general : extinction ] as the laplace transform is convergence determining , see e.g. lemma 2.1 in @xcite .
our proof of theorem [ thm : general : xlogx ] follows the proof of doney ( 1972 ) @xcite for supercritical crump - mode - jagers processes .
some changes are necessary because the recursive equation differs from the respective recursive equation in @xcite .
parts of our proof are analogous to the proof in @xcite which we nevertheless include here for the reason of completeness .
lemma [ l : lemma52 ] and lemma [ l : convergence_if_xlogx_fails ] below contain the essential part of the proof of theorem [ thm : general : xlogx ] .
for these two lemmas , we will need auxiliary lemmas which we now provide .
we assume throughout this section that a solution @xmath385 of equation exists .
note that this is implied by [ a : general : finite_man_hours ] and @xmath386 .
recall the definition of @xmath387 from .
for @xmath388 , define @xmath389}}}q(d\chi)\ ] ] for @xmath390 .
[ l : h_contracting ] the operator @xmath391 is contracting in the sense that @xmath392 for all @xmath393 .
the lemma follows immediately from @xmath394 and from the definition of @xmath387 .
[ l : h_nondecreasing ] the operator @xmath391 is nondecreasing in the sense that @xmath395 for all @xmath388 if @xmath396 for all @xmath388 .
the lemma follows from @xmath397 being increasing in @xmath120 for every @xmath398 . for every measurable function @xmath399 , define @xmath400 } } } q(d\chi ) . \end { split}\ ] ] for @xmath388 and @xmath401 where @xmath402 , @xmath16 .
if @xmath403 is a function of one variable , then we set @xmath404 where @xmath405 for @xmath388 , @xmath401 .
[ l : barh_nondecreasing ] the operator @xmath406 is nondecreasing in the sense that @xmath407 for all @xmath388 and @xmath401 if @xmath408 for all @xmath388 , @xmath401 .
the assertion follows from the basic fact that @xmath409 is nondecreasing .
[ l : eta_nondecreasing ] assume [ a : general : finite_man_hours ] . let @xmath410 be the identity map .
the function @xmath411 is nonnegative and nondecreasing .
furthermore , @xmath412 .
recall the definition of @xmath413 from . by equation
, we have @xmath414 .
thus , @xmath415 is nonnegative .
furthermore , @xmath412 follows from the dominated convergence theorem and assumption [ a : general : finite_man_hours ] .
let @xmath416 . then @xmath417 the inequality follows from @xmath418 for all @xmath419 .
the following lemma , due to athreya @xcite , translates the @xmath103-condition into an integrability condition on @xmath415 . for completeness
, we include its proof .
[ l : properties_of_eta ] assume [ a : general : finite_man_hours ] .
let @xmath415 be the function defined in .
then @xmath420 for some and then all @xmath398 , @xmath421 if and only if the @xmath103-condition holds .
by monotonicity of @xmath415 ( see lemma [ l : eta_nondecreasing ] ) , the two quantities in are finite or infinite at the same time .
fix @xmath398 . using fubini s theorem and the substitution @xmath422
, we obtain @xmath423}}}{{{\bigl(a_{{\ensuremath{\alpha}}}\bigr)}}}^2\,d{{\ensuremath{\lambda}}}\biggr]}}}dq\\ & = \int{{{\biggl[a_{{\ensuremath{\alpha}}}\int_0^{ca_{{\ensuremath{\alpha}}}}\frac{v-1+e^{-v}}{v^2 } \,dv\biggr]}}}dq . \end { split}\ ] ] it is a basic fact that @xmath424 as @xmath425
. in the following two lemmas , we consider uniqueness and existence of a function @xmath426 which satisfies : @xmath427 where @xmath428 is as in lemma [ l : fixed_point ] .
notice that the zero function does not satisfy ( c ) .
first , we prove uniqueness . [
l : uniqueness_psi ] assume [ a : general : finite_man_hours ] and @xmath96 .
if @xmath429 and @xmath430 satisfy , then @xmath431 .
notice that @xmath432 .
define @xmath433 for @xmath344 and note that @xmath434 by ( c ) . from lemma
[ l : h_contracting ] , we obtain for @xmath344 @xmath435 where @xmath436 is a probability measure because @xmath98 solves equation .
let @xmath437 , @xmath438 , be independent variables with distribution @xmath439 and note that @xmath440 .
we may assume that @xmath441 because @xmath442 implies @xmath443 for @xmath444 .
iterating inequality , we arrive at @xmath445 the convergence in follows from the weak law of large numbers . [ l : existence_psi ] assume [ a : general : finite_man_hours ] and @xmath96 .
there exists a solution @xmath426 of if and only if the @xmath103-condition holds .
assume that holds .
define @xmath446 , @xmath447 for @xmath388 and @xmath448 for @xmath344 and @xmath449 .
recall @xmath439 and @xmath450 from the proof of lemma [ l : uniqueness_psi ] .
note that @xmath441 because of @xmath96 .
arguments as in the proof of lemma [ l : uniqueness_psi ] imply @xmath451 where @xmath452 for @xmath68 . since @xmath453 by lemma [ l : eta_nondecreasing ] and @xmath454 we see that @xmath455 .
in addition , we conclude from @xmath453 that @xmath456 . by lemma [ l : h_nondecreasing ] , this implies inductively @xmath457 for @xmath68 , @xmath388 .
let @xmath458 .
we need to prove that @xmath459 . clearly @xmath460 , so we can choose @xmath140 with @xmath461 .
then @xmath462 define @xmath463 .
it follows from , , lemma [ l : eta_nondecreasing ] and lemma [
l : properties_of_eta ] that for all @xmath344 @xmath464 thus , @xmath465 is a cauchy sequence in @xmath466 $ ] .
hence , we conclude the existence of a function @xmath426 such that @xmath467 for every @xmath388 . by the dominated convergence theorem ,
@xmath426 satisfies ( b ) . to check ( a )
, we prove that @xmath468 is lipschitz continuous with constant one .
the induction step follows from lemma [
l : h_contracting ] @xmath469 in order to check ( c ) , note that since @xmath412 , it follows from that @xmath434 .
thus , @xmath470 as required
. finally , monotonicity of @xmath468 and @xmath471 for all @xmath472 imply monotonicity of @xmath426 and @xmath473 , respectively .
the last claim of ( d ) , namely @xmath474 , follows from letting @xmath338 in @xmath475 , monotonicity of @xmath426 and from lemma [ l : fixed_point ] together with @xmath476 . for the `` only if''-part of the lemma , suppose that there exists a solution @xmath426 of .
write @xmath477 for @xmath344 . since @xmath478 and @xmath479 , there exist constants @xmath480 such that @xmath481 for all @xmath482 $ ] . using ( b ) , @xmath483 for @xmath482 $ ] and lemma [ l : barh_nondecreasing ] , we obtain for @xmath482 $ ] @xmath484 let @xmath216 be such that @xmath485\bigr)}}}<1 $ ] and write @xmath486 .
then @xmath487 which we rewrite as @xmath488 where @xmath489 and @xmath490 . iterating this inequality results in @xmath491 for @xmath68 .
since @xmath492 is bounded on @xmath493 $ ] this implies that @xmath494 .
therefore , by lemma [ l : properties_of_eta ] , the @xmath103-condition holds .
recall @xmath495 , @xmath297 , @xmath302 and @xmath350 from , , and , respectively . as
before , let @xmath436 .
define @xmath496 @xmath497 and @xmath498 for @xmath344 and @xmath499 .
the following two lemmas follow lemma 5.1 and lemma 5.2 , respectively , in @xcite .
[ l : lemma51 ] assume [ a : general : finite_moments ] , [ a : general : finite_man_hours ] , [ a : for_skdist ] and @xmath96 .
if the @xmath103-condition holds , then @xmath500 as @xmath501 . inserting the definitions and of @xmath302 and @xmath350 , respectively , into
, we see that @xmath502}}}\geq 0 \qquad { { \ensuremath{\lambda}}},t>0\ ] ] is nonnegative where @xmath503 , @xmath16 .
insert the recursive equations and for @xmath302 and @xmath361 , respectively , into to obtain @xmath504}}}q(d\chi)\\ & = \int_0^t{{{\bigl [ \frac{m^q(t - s)}{e^{{{\ensuremath{\alpha}}}(t - s)}{\bar{m } } } -\frac{1}{{{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s } } l_{t - s}{{{\bigl(\frac{{{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s}}{e^{{{\ensuremath{\alpha}}}(t - s)}{\bar{m}}}\bigr)}}}\bigr ] } } } \mu^q_{{\ensuremath{\alpha}}}(ds)\\ & \quad + \frac{1}{{{\ensuremath{\lambda}}}}\int_0^t l_{t - s}{{{\bigl(\frac{{{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s}}{e^{{{\ensuremath{\alpha}}}(t - s)}{\bar{m}}}\bigr ) } } } \mu^q(ds)\\ & \qquad\quad-\frac{1}{{{\ensuremath{\lambda}}}}\int{{{\bigl[1- \exp{{{\bigl(-\int_0^t a{{{\bigl(\chi_s\bigr ) } } } l_{t - s}{{{\bigl(\frac{{{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s}}{e^{{{\ensuremath{\alpha}}}(t - s)}{\bar{m}}}\bigr)}}}\,ds\bigr)}}}\bigr]}}}q(d\chi)\\ & \quad + \int { { \ensuremath{\mathbf{e}}}}{{{\bigl[\frac{1-\exp{{{\bigl(-{{\ensuremath{\lambda}}}\frac{\phi_\chi(t)}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}\bigr)}}}}{{{\ensuremath{\lambda}}}}\bigr ] } } } { { { \bigl[1- \exp{{{\bigl(-\int_0^t a{{{\bigl(\chi_s\bigr ) } } } l_{t - s}{{{\bigl(\frac{{{\ensuremath{\lambda}}}}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}\bigr)}}}\,ds\bigr)}}}\bigr]}}}q(d\chi)\\ & \quad
+ \frac{1}{{{\ensuremath{\lambda}}}}\int { { \ensuremath{\mathbf{e}}}}{{{\biggl[\frac{{{\ensuremath{\lambda}}}\phi(t,\chi)}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}-1 + \exp{{{\bigl(-\frac{{{\ensuremath{\lambda}}}\phi(t,\chi)}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}\bigr)}}}\biggr ] } } } q(d\chi)\\ & = : \int_0^t d({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s},t - s)\mu^q_{{\ensuremath{\alpha}}}(ds ) + \frac{1}{{{\ensuremath{\lambda}}}}\bar{h}_{{\ensuremath{\alpha}}}{{{\bigl((t,{{\ensuremath{\lambda}}})\mapsto l_t{{{\bigl(\frac{{{\ensuremath{\lambda}}}}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}\bigr)}}}\bigr ) } } } + t_1+t_2 \end { split}\ ] ] where @xmath505 and @xmath506 are suitably defined . inequality implies @xmath507 where @xmath508 is a finite constant .
the last inequality is a consequence of theorem [ thm : general : expected_man_hours ] , equation , with @xmath185 replaced by @xmath183 .
lemma [ l : barh_nondecreasing ] together with implies @xmath509 using @xmath510 , inequality and @xmath511 , @xmath16 , we see that the expressions @xmath505 and @xmath506 are bounded above by @xmath512}}}q(d\chi ) \leq c_2{{\ensuremath{\lambda}}}\\ t_2&\leq\frac{{{\ensuremath{\lambda}}}}{2}\int{{\ensuremath{\mathbf{e}}}}{{{\bigl(\frac{\phi_\chi(t)}{e^{{{\ensuremath{\alpha}}}t}{\bar{m}}}\bigr)}}}^2 q(d\chi ) \leq c_3 { { \ensuremath{\lambda}}}\end { split}\ ] ] for all @xmath513 where @xmath514 are finite constants which are independent of @xmath515 and @xmath344 .
such constants exist by assumption [ a : for_skdist ] .
taking supremum over @xmath516 $ ] in and inserting and results in @xmath517 choose @xmath518 such that @xmath519)\in(0,1)$ ] .
then , by monotonicity of @xmath520 , @xmath521 for all @xmath344 . hence , @xmath520 is bounded by @xmath522 where @xmath523 and @xmath524 .
iterate this inequality to obtain @xmath525 now we need to prove @xmath526 .
looking at and using @xmath527 , we see that @xmath528 is bounded by @xmath529}}}$ ] .
this is finite because of inequality with @xmath277 together with [ a : for_skdist ] .
therefore @xmath526 . letting @xmath425 in
, we obtain @xmath530 the right - hand side is finite by lemma [ l : properties_of_eta ] . by lemma [ l : eta_nondecreasing ] , we know that @xmath412 and that @xmath415 is nondecreasing . letting @xmath501 in and using the dominated convergence theorem implies @xmath531 as @xmath501 .
[ l : lemma52 ] assume [ a : general : finite_moments ] , [ a : general : finite_man_hours ] , [ a : convergence_to_zero ] , [ a : for_skdist ] and @xmath96 if the @xmath103-condition holds , then @xmath532 for every @xmath388 where @xmath426 is the unique solution of .
the case @xmath533 is trivial . for @xmath534 , define @xmath535 furthermore , let @xmath536 and @xmath537 for @xmath344 . we will prove @xmath538 for @xmath344 . by theorem [ thm : general : expected_man_hours ] and ( c ) , @xmath539 hence , @xmath540 by lemma [ l : lemma51 ] . using and ( b )
, we estimate @xmath541 for a suitable constant @xmath358 .
the last inequality uses boundedness of @xmath361 , see the proof of theorem [ thm : general : extinction ] , and of @xmath426 . by assumption [ a : general : finite_man_hours ]
, the right - hand side of converges to zero as @xmath74 .
fix @xmath344 and let @xmath542 be such that @xmath543 . with this , we get @xmath544 the convergence in follows from [ a : convergence_to_zero ] and from the dominated convergence theorem together with assumption [ a : general : finite_man_hours ] .
recall @xmath545 from the proof of lemma [ l : uniqueness_psi ] . putting and together ,
we arrive at @xmath546 this finishes the proof .
if the @xmath103-condition fails to hold , then the rescaled supercritical virgin island model converges to zero .
the proof of this assertion follows kaplan @xcite . [
l : convergence_if_xlogx_fails ] assume [ a : general : finite_moments ] , [ a : general : finite_man_hours ] , [ a : convergence_to_zero ] and @xmath96 . if the @xmath103-condition fails to hold , then @xmath547 for every @xmath388 define @xmath548 for @xmath344 and @xmath549 .
it suffices to prove @xmath550 assume that @xmath551 for some @xmath552 .
we will prove that the @xmath103-condition holds .
an elementary calculation shows that @xmath553 is decreasing .
thus , both @xmath554 and @xmath555 are decreasing in @xmath556 .
furthermore , by theorem [ thm : general : expected_man_hours ] , @xmath557 fix @xmath518 , @xmath558 and let @xmath559 . inserting the recursive equation , @xmath560}}}q(d\chi)\\ & \leq \sup_{u\geq t_0}\int { { { \bigl(1-{{\ensuremath{\mathbf{e}}}}e^{-{{\ensuremath{\lambda}}}\phi_\chi(u)}\bigr ) } } } q(d\chi)\\ & \qquad\qquad\,+\int{{{\bigl(1-\exp{{{\bigl(-\int_0^{t - t_0 } a{{{\bigl(\chi_s\bigr)}}}l_{t - s}({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s } e^{-{{\ensuremath{\alpha}}}(t - s)})ds\bigr)}}}\bigr)}}}q(d\chi)\\ & \qquad\qquad\,+\int{{{\bigl(1-\exp{{{\bigl(-\int_{t - t_0}^{t } a{{{\bigl(\chi_s\bigr)}}}l_{t - s}({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s } e^{-{{\ensuremath{\alpha}}}(t - s)})ds\bigr)}}}\bigr)}}}q(d\chi)\\ & = : t_1+t_2+t_3 .
\end { split}\ ] ] by assumption [ a : convergence_to_zero ] , the first term converges to zero uniformly in @xmath561 as @xmath562 . for the third term
, we use inequality to obtain @xmath563 the right - hand side converges to zero uniformly in @xmath561 as @xmath562 by assumption [ a : general : finite_man_hours ] .
the second term is bounded above by @xmath564 recall @xmath545 from the proof of lemma [ l : uniqueness_psi ] .
define @xmath565 and @xmath566 , @xmath47 .
taking supremum over @xmath567 in and letting @xmath562 , we arrive at @xmath568 } } } -\frac{1}{{{\ensuremath{\lambda}}}}\bar{h}_{{\ensuremath{\alpha}}}{{{\bigl({{\ensuremath{\tilde{{{\ensuremath{\lambda}}}}}}}\mapsto{{\ensuremath{\delta}}}{{\ensuremath{\tilde{{{\ensuremath{\lambda}}}}}}}\bigr)}}}({{\ensuremath{\lambda } } } ) = { { \ensuremath{\mathbf{e}}}}{{{\bigl [ k_\infty { { { \bigl({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}r_1}\bigr)}}}\bigr ] } } } -{{\ensuremath{\delta}}}\eta({{\ensuremath{\delta}}}{{\ensuremath{\lambda}}})\\ & \leq\cdots\leq{{\ensuremath{\mathbf{e}}}}{{{\bigl [ k_\infty { { { \bigl({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s_n}\bigr)}}}\bigr ] } } } -{{\ensuremath{\delta}}}\sum_{k=0}^{n-1}{{\ensuremath{\mathbf{e}}}}\eta{{{\bigl({{\ensuremath{\delta}}}{{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s_k}\bigr ) } } } \end { split}\ ] ] for all @xmath68 .
the second inequality follows from @xmath569 for @xmath570 and lemma [ l : barh_nondecreasing ] .
boundedness of @xmath571 on @xmath572 $ ] , see , implies @xmath573 by the law of large numbers , we know that @xmath574 for large @xmath327 a.s . hence , @xmath575 where @xmath576 .
therefore , the @xmath103-condition holds by lemma [ l : properties_of_eta ] .
this finishes the proof .
assume that the @xmath103-condition holds .
insert into and use assumption [ a : convergence_to_zero ] to obtain @xmath577 } } } { \xrightarrow{t { \rightarrow}\infty}}\int{{{\biggl[\exp{{{\bigl(-\int_0^\infty \psi({{\ensuremath{\lambda}}}e^{-{{\ensuremath{\alpha}}}s } ) a(\chi_s)\,ds\bigr)}}}\biggr ] } } } \nu(d\chi)\ ] ] for @xmath388 . for this , we applied the dominated convergence theorem together with assumption [ a : general : finite_man_hours ] .
denote the right - hand side of by @xmath578 and note that @xmath579 is continuous and satisfies @xmath580 .
a standard result , e.g. lemma 2.1 in @xcite , provides us with the existence of a random variable @xmath100 such that @xmath581 for all @xmath388 .
this proves the weak convergence as the laplace transform is convergence determining .
note that @xmath582}}}\nu(d\chi)\ ] ] by the dominated convergence theorem .
furthermore , @xmath583}}}\,\nu(d\chi).\ ] ] if the @xmath103-condition fails to hold , then @xmath584}}}\to 0 $ ] as @xmath74 follows by inserting into together with [ a : convergence_to_zero ] .
[ sec : vor_excursions ]
recall the assumptions [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] , [ a : finite_man_hours ] and [ a : finite_man_hours_squared ] from section [ sec : main_results ] .
the process @xmath33 , the excursion set @xmath37 and the scale function @xmath126 have been defined in , in and in , respectively .
the stopping time @xmath585 has been introduced shortly after . in this section , we define the excursion measure @xmath124 and prove the convergence result of theorem [ thm : existence_excursion_measure ] .
we follow pitman and yor @xcite in the construction of the excursion measure . under assumptions
[ a : a1 ] and [ a : a2 ] , zero is an absorbing point for @xmath24 .
thus , we can not simply start in zero and wait until the process returns to zero .
informally speaking , we instead condition the process to converge to infinity . one way to achieve
this is by doob s h - transformation .
note that @xmath586 is a bounded martingale for every @xmath140 , see section v.28 in @xcite .
in particular , @xmath587}}}={\bar{s}}(y)\ ] ] for every @xmath588 . for @xmath140 , consider the diffusion @xmath589 on @xmath110 to be called the @xmath590-diffusion stopped at time @xmath585 defined by the semigroup @xmath591 where @xmath592}}},\quad y>0,t\geq0 , f\in{{\ensuremath{\mathbf{c}}}}_b{{{\bigl([0,\infty),{{\ensuremath{\mathbbm{r}}}}\bigr)}}}.\ ] ] the sequence of processes @xmath593 is consistent in the sense that @xmath594 for all @xmath595 and @xmath132 . therefore , we may define a process @xmath596 which coincides with @xmath597 until time @xmath585 for every @xmath140 .
note that the @xmath590-diffusion possibly explodes in finite time .
the following important observation of williams has been quoted by pitman and yor @xcite . because we assume that zero is an exit boundary for @xmath33 ,
zero is an entrance boundary but not an exit boundary for the @xmath590-diffusion .
more precisely , the @xmath590-diffusion started at its entrance boundary zero and run up to the last time it hits a level @xmath598 is described by theorem 2.5 of williams @xcite as the time reversal back from @xmath599 of the @xmath600-diffusion started at @xmath121 , where the @xmath600-diffusion is the process @xmath33 conditioned on @xmath601 .
hence , the process @xmath602 may be started in zero but takes strictly positive values at positive times .
pitman and yor @xcite define the excursion measure @xmath124 as follows . under @xmath603 that is , conditional on `` excursions reach level @xmath604 ''
, an excursion follows the @xmath590-diffusion until time @xmath585 and then follows the dynamics of @xmath33 .
in addition , @xmath605 . with this in mind , define a process @xmath606 which satisfies @xmath607 for @xmath608 .
in addition , @xmath609 and @xmath610 are independent .
define the excursion measure @xmath124 on @xmath37 by @xmath611 this is well - defined if @xmath612 holds for all @xmath613 .
the critical part here is the path between @xmath585 and @xmath614 .
therefore , follows from @xmath615 } } } = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta } } } ) } { { \ensuremath{\mathbf{e}}}}^{{\ensuremath{\varepsilon}}}{{{\bigl[f(y)|{t_{{{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta}}}}<t_0}\bigr]}}}}\\ & = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta } } } ) } { { \ensuremath{\mathbf{e}}}}^{{\ensuremath{\varepsilon}}}{{{\bigl[f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta}}}})\bigr ] } } } = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta } } } ) } { { \ensuremath{\mathbf{e}}}}^0{{{\bigl[f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon}}}+{{\ensuremath{\delta}}}}_{t_{{\ensuremath{\varepsilon}}}+\centerdot})\bigr]}}}. \end { split}\ ] ] the first equality is equation with @xmath616 , @xmath617 and @xmath618 .
the last equality is the strong markov property of @xmath619 .
the last but one equality is the following lemma .
[ l : conditioned_on_t_eps ] assume [ a : a1 ] and [ a : a2 ] . let @xmath620 .
then @xmath621 we begin with the proof of independence of @xmath609 and of @xmath610 .
let @xmath622 and @xmath623 be two bounded continuous functions on the path space .
denote by @xmath624 the @xmath43-algebra generated by @xmath625 .
then @xmath626}}}}\\ & = { { \ensuremath{\mathbf{e}}}}^y{{{\bigl[f{{{\bigl(y_{t_{{\ensuremath{\varepsilon}}}\wedge\centerdot}\bigr ) } } } { { \ensuremath{\mathbf{e}}}}^y{{{\bigl[g{{{\bigl(y_{t_{{\ensuremath{\varepsilon}}}+\centerdot}\bigr)}}}|{{\ensuremath{\mathcal f}}}_{t_{{\ensuremath{\varepsilon}}}}\bigr]}}}|t_{{\ensuremath{\varepsilon}}}<t_0\bigr]}}}\\ & = { { \ensuremath{\mathbf{e}}}}^y{{{\bigl[f{{{\bigl(y_{t_{{\ensuremath{\varepsilon}}}\wedge\centerdot}\bigr ) } } } |t_{{\ensuremath{\varepsilon}}}<t_0\bigr ] } } } { { \ensuremath{\mathbf{e}}}}^{{\ensuremath{\varepsilon}}}{{{\bigl[g{{{\bigl(y_{\centerdot}\bigr)}}}\bigr]}}}. \end { split}\ ] ] the last equality is the strong markov property of @xmath24 .
choosing @xmath627 in proves that the left - hand side of satisfies .
in addition , equation proves the desired independence . for the proof of @xmath628
we repeatedly apply the semigroup of @xmath597 to obtain @xmath629 } } } = \frac{1}{{\bar{s}}(y)}{{\ensuremath{\mathbf{e}}}}^y{{{\bigl [ { \bar{s}}(y_{t_n\wedge t_{{\ensuremath{\varepsilon } } } } ) \prod_{i=1}^nf_i{{{\bigl(y_{t_i\wedge t_{{\ensuremath{\varepsilon}}}}\bigr)}}}\bigr]}}}\ ] ] for bounded , continuous functions @xmath630 and time points @xmath631 . by equation with @xmath616 , @xmath632 } } } = { \bar{s}}({{\ensuremath{\varepsilon } } } ) { { \ensuremath{\mathbf{e}}}}^{y}{{{\bigl[{{\ensuremath{\mathbbm{1}}}}_{t_{{\ensuremath{\varepsilon}}}<t_0}|{{\ensuremath{\mathcal f}}}_{t_n\wedge t_{{\ensuremath{\varepsilon}}}}\bigr]}}}\ ] ] @xmath633almost surely where @xmath634 is the @xmath43-algebra generated by @xmath635 .
insert this identity in the right - hand side of to obtain @xmath636 } } } = \frac{1}{{{\ensuremath{\mathbf{p}}}}^y{{{\bigl(t_{{\ensuremath{\varepsilon}}}<t_0\bigr)}}}}{{\ensuremath{\mathbf{e}}}}^y{{{\bigl [ { { \ensuremath{\mathbbm{1}}}}_{t_{{\ensuremath{\varepsilon}}}<t_0 } \prod_{i=1}^nf_i{{{\bigl(y_{t_i\wedge t_{{\ensuremath{\varepsilon}}}}\bigr)}}}\bigr]}}}.\ ] ] this proves because finite - dimensional distributions determine the law of a process .
now we prove convergence to the excursion measure @xmath124 .
let @xmath139 be a bounded continuous function for which there exists an @xmath140 such that @xmath637 for every path @xmath324 .
let @xmath620 .
by lemma [ l : conditioned_on_t_eps ] , we obtain @xmath638}}}\\ & = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon } } } ) } { { \ensuremath{\mathbf{e}}}}^y f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon } } } } ) = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}})}{{\ensuremath{\mathbf{e}}}}^0 f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon}}}}_{t_y+\centerdot } ) .
\end { split}\ ] ] the last equality is the strong markov property of the @xmath590-diffusion .
the random time @xmath639 converges to zero almost surely as @xmath640 .
another observation we need is that every continuous path @xmath641 is uniformly continuous on any compact set @xmath642 $ ] .
hence , the sequence of paths @xmath643 converges locally uniformly to the path @xmath644 almost surely as @xmath640 .
therefore , the dominated convergence theorem implies @xmath645 putting and together , we arrive at @xmath646 which proves the theorem .
we will employ lemma [ l : conditioned_on_t_eps ] to calculate explicit expressions for some functionals of @xmath124 .
for example , we will prove in lemma [ l : q_explicit_man_hours ] together with lemma [ l : finite_man_hours ] that @xmath647 provided that assumptions [ a : a1 ] , [ a : a2 ] and [ a : finite_man_hours ] hold .
equation shows that condition and condition are equivalent .
the following lemmas prepare for the proof of .
[ l : mit_t_b ] assume [ a : a1 ] and [ a : a2 ] .
let @xmath648 have compact support in @xmath116 . furthermore , let the continuous function @xmath649 be nonnegative and nondecreasing .
then @xmath650 } } } { \xrightarrow{y{\rightarrow}0}}\int{{{\biggl[{{{\bigl ( \int_0^{t_b}\psi(s)f(\chi_s)\,ds\bigr)}}}^m\biggr]}}}{\bar{q}_y}(d\chi)\ ] ] for every @xmath651 and @xmath652 .
w.l.o.g .
assume @xmath653 .
let @xmath140 be such that @xmath654 and let @xmath655 .
using lemma [ l : conditioned_on_t_eps ] , we see that the left - hand side of is equal to @xmath656 } } } = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}})}{{\ensuremath{\mathbf{e}}}}^y{{{\biggl[{{{\bigl(\int_0^{t_b}\psi(s ) f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon}}}}_s)\,ds\bigr)}}}^m\biggr]}}}\\ & = \frac{1}{{\bar{s}}({{\ensuremath{\varepsilon}}})}{{\ensuremath{\mathbf{e}}}}^0{{{\biggl[{{{\bigl(\int_{t_y}^{t_b}\psi(s - t_y ) f({{\ensuremath{\hat{y}}}}^{{{\ensuremath{\varepsilon}}}}_{s})\,ds\bigr)}}}^m\biggr ] } } } { \xrightarrow{y{\rightarrow}0}}\int{{{\biggl(\int_0^{t_b}\psi(s ) f(\chi_s)\,ds\biggr)}}}^m{\bar{q}_y}(d\chi ) .
\end { split}\ ] ] the second equality is the strong markov property of @xmath657 and the change of variable @xmath658 . for the convergence
, we applied the monotone convergence theorem . the explicit formula on the right - hand side of originates in the explicit formula below , which we recall from the literature .
[ l : explicit_man_hours ] assume [ a : a1 ] and [ a : a2 ] . if @xmath659 or @xmath660 , then @xmath661 for all @xmath662 .
see e.g. section 15.3 of karlin and taylor @xcite .
let @xmath663 be a markov process with cdlg sample paths and state space @xmath664 which is a polish space .
for an open set @xmath665 , denote by @xmath666 the first exit time of @xmath663 from the set @xmath667 .
notice that @xmath666 is a stopping time . for @xmath668 , define @xmath669}}},\ y\in e , m\in{{\ensuremath{\mathbbm{n}}}}_0,\ ] ] for a given function @xmath670 . in the following lemma , we derive an expression for @xmath671 for which lemma [ l : explicit_man_hours ] is applicable .
[ l : conversion ] let @xmath663 be a time - homogeneous markov process with cdlg sample paths and state space @xmath664 which is a polish space .
let @xmath672 be as in with an open set @xmath665 and with a function @xmath670 .
then @xmath673 } } } \!\!\!\!&=&\!\!\!\!{{\ensuremath{\mathbf{e}}}}^y{{{\bigl(\int_0^{\tau}2 f({{\ensuremath{\tilde{y}}}}_s)w_1({{\ensuremath{\tilde{y}}}}_s)\,ds\bigr ) } } } \label{eq : wzz } \end{aligned}\ ] ] for all @xmath674 .
let @xmath674 be fixed .
for the proof of , we apply fubini to obtain @xmath675 the last equality follows from fubini and a change of variables .
the stopping time @xmath666 can be expressed as @xmath676 with a suitable path functional @xmath622 .
furthermore , @xmath666 satisfies @xmath677 for @xmath678 .
therefore , the right - hand side of is equal to @xmath679}}}\bigr)}}}\,dr = { { \ensuremath{\mathbf{e}}}}^y{{{\bigl(\int_0^{\tau}w_1({{\ensuremath{\tilde{y}}}}_r)\,dr\bigr)}}}. \end { split}\ ] ] the last but one equality is the markov property of @xmath663 .
this proves .
for the proof of , break the symmetry in the square of @xmath680 to see that @xmath680 is equal to @xmath681}}}dr\bigr)}}}}\\ & = 2\int_0^{\infty}{{\ensuremath{\mathbf{e}}}}^y{{{\bigl({{\ensuremath{\mathbbm{1}}}}_{r<\tau}f({{\ensuremath{\tilde{y}}}}_r ) { { \ensuremath{\mathbf{e}}}}^{{{\ensuremath{\tilde{y}}}}_r}{{{\bigl[\int_0^{\tau}f({{\ensuremath{\tilde{y}}}}_{s})\,ds\bigr]}}}\bigr)}}}\,dr = { { \ensuremath{\mathbf{e}}}}^y{{{\bigl(\int_0^{\tau}2f({{\ensuremath{\tilde{y}}}}_s)w_1({{\ensuremath{\tilde{y}}}}_s)\,ds\bigr)}}}. \end { split}\ ] ] this finishes the proof .
we will need that @xmath33 dies out in finite time .
the following lemma gives a condition for this .
recall @xmath682 .
[ l : finite_time_extinction ] assume [ a : a1 ] and [ a : a2 ] .
let @xmath598 .
then the solution @xmath33 of equation hits zero in finite time almost surely if and only if @xmath683 .
if @xmath684 , then @xmath33 converges to infinity as @xmath74 on the event @xmath685 almost surely . on the event @xmath686
, we have that @xmath687 almost surely .
the last inequality follows from lemma 15.6.2 of @xcite and assumption [ a : a2 ] .
therefore , theorem 2 of jagers @xcite implies that , with probability one , either @xmath33 hits zero in finite time or converges to infinity as @xmath74 . with equation ,
we obtain @xmath688 this proves the assertion .
the following lemma makes assumption [ a : finite_man_hours ] more transparent .
it proves that [ a : finite_man_hours ] holds if and only if the expected area under @xmath689 is finite .
l : finite_man_hours ] assume [ a : a1 ] and [ a : a2 ] .
assumption [ a : finite_man_hours ] holds if and only if @xmath690 if assumption [ a : finite_man_hours ] holds , then @xmath683 and @xmath691 for all @xmath608 and @xmath648 with @xmath692 .
let @xmath693 be the constants from [ a : a1 ] . in equation ,
let @xmath694 and apply monotone convergence to obtain @xmath695 } } } \frac{{\bar{s}}(y\wedge z)}{g(z){\bar { s}}(z)}\bigr ) } } } \,dz .
\end { split}\ ] ] hence , if assumption [ a : finite_man_hours ] holds , then assumption [ a : a2 ] implies that the right - hand side of is finite because @xmath696 , @xmath697 .
therefore , the left - hand side of with @xmath698 replaced by @xmath111 is finite .
together with @xmath699 , this implies that @xmath33 does not converge to infinity with positive probability as @xmath74 .
thus lemma [ l : finite_time_extinction ] implies @xmath683 and equation implies .
now we prove that assumption [ a : finite_man_hours ] holds if the left - hand side of with @xmath698 replaced by @xmath111 is finite .
again , @xmath699 and lemma [ l : finite_time_extinction ] imply @xmath683 . using monotonicity of @xmath700
, we obtain for @xmath130 @xmath701 the right - hand side is finite because with @xmath698 replaced by @xmath111 is finite .
therefore , assumption [ a : finite_man_hours ] holds .
[ l : bs_durch_y_bounded ] assume [ a : a1 ] , [ a : s_bar ] and let @xmath702 . if @xmath703 , then @xmath704 it suffices to prove @xmath705 because @xmath706 is locally bounded in @xmath116 and @xmath707 by assumption [ a : s_bar ] . by assumption [ a : a1 ] , @xmath708 for all @xmath709 and a constant @xmath710 .
let @xmath711 .
thus , @xmath712 the last inequality follows from @xmath713 , @xmath697 .
consequently , @xmath714 the proof of the second equation in is similar to the proof of the lemma of lhospital . from , we conclude @xmath715 which implies @xmath716 this finishes the proof .
now we prove equation .
recall @xmath682 .
define @xmath717 and @xmath718 for @xmath660 .
if @xmath683 , then @xmath719 is the monotone limit of the right - hand side of as @xmath694 .
[ l : q_explicit_man_hours ] assume [ a : a1 ] , [ a : a2 ] and @xmath683 .
let @xmath660 .
then @xmath720 for @xmath721 .
if [ a : finite_man_hours ] holds and if @xmath722 is bounded , then is finite for @xmath723 .
if [ a : finite_man_hours_squared ] holds and if @xmath722 is bounded , then is finite for @xmath724 .
choose @xmath725 with compact support in @xmath116 for every @xmath140 such that @xmath726 as @xmath727 .
fix @xmath140 and @xmath728 .
lemma [ l : mit_t_b ] proves that @xmath729 } } } = \int{{{\bigl(\int_0^{t_b } f_{{\ensuremath{\varepsilon}}}{{{(\chi_s)}}}\,ds\bigr)}}}^m { \bar{q}_y}{{{(d\chi)}}}. \end { split}\ ] ] let @xmath730 be as in with @xmath666 replaced by @xmath731 and @xmath409 replaced by @xmath732 . fix @xmath733 .
lemma [ l : conversion ] and lemma [
l : explicit_man_hours ] provide us with an expression for the left - hand side of equation .
hence , @xmath734 the last equation follows from dominated convergence and assumption [ a : a2 ] .
note that the hitting time @xmath735 as @xmath694 for every continuous path @xmath736 . by lemma [ l : explicit_man_hours ] and
the monotone convergence theorem , @xmath737 as @xmath738 .
let @xmath694 , @xmath727 and apply monotone convergence to arrive at equation .
similar arguments prove . instead of , consider @xmath739 which is implied by lemma [ l : mit_t_b ] .
furthermore , instead of applying lemma [
l : explicit_man_hours ] to equation , apply equation together with equation . for
the rest of the proof , assume that @xmath722 is bounded by @xmath740 .
let @xmath693 be the constants from [ a : a1 ] .
note that @xmath696 .
consider the right - hand side of .
if @xmath723 , then the integral over @xmath741 is finite by assumption [ a : finite_man_hours ] . if @xmath724 , then the integral over @xmath741 is finite by assumption [ a : finite_man_hours_squared ] .
the integral over @xmath742 is finite because of [ a : a2 ] and @xmath743,\ ] ] where @xmath744 is a finite constant .
the last inequality in follows from lemma [ l : bs_durch_y_bounded ] .
the convergence of theorem [ thm : existence_excursion_measure ] also holds for @xmath745 , @xmath93 fixed , if @xmath746 is a bounded function . for this
, we first estimate the moments of @xmath33 .
[ l : finite_moments_y ] assume [ a : a1 ] .
let @xmath33 be a solution of equation and let @xmath347 be finite .
then , for every @xmath747 , there exists a constant @xmath748 such that @xmath749}}}\leq c_t{y},\quad { { \ensuremath{\mathbf{e}}}}^y{{{\bigl[\sup_{t\leq t } { y_t}^n\bigr]}}}\leq c_t({y}+{y}^n ) \end { split}\ ] ] for all @xmath608 and every stopping time @xmath666 .
the proof is fairly standard and uses it s formula and doob s @xmath750-inequality .
[ l : functional_fixed_t ] assume [ a : a1 ] , [ a : a2 ] and [ a : s_bar ] . let @xmath751 be a continuous function such that @xmath752 for some @xmath753 , some constant @xmath754 and for all @xmath608 . if @xmath703 , then @xmath755}}}\ ] ] is bounded in @xmath515 .
choose @xmath725 with compact support in @xmath116 for every @xmath140 such that @xmath726 pointwise as @xmath727 . by theorem [ thm : existence_excursion_measure ]
, @xmath756 the left - hand side of converges to the left - hand side of as @xmath727 by the monotone convergence theorem .
hence , the first equality in follows from if the limits @xmath757 and @xmath758 can be interchanged . for this
, we prove the second equality in .
let @xmath728 .
the @xmath590-diffusion is a strong markov process .
thus , by , @xmath759 } } } = { { \ensuremath{{\displaystyle \lim_{y { \rightarrow}0}}}}}{{\ensuremath{\mathbf{e}}}}^y{{{\bigl[\frac{f(y_t^{\uparrow})}{{\bar{s}}(y_t^{\uparrow})}{{\ensuremath{\mathbbm{1}}}}_{t < t_b}\bigr]}}}}\\ & = { { \ensuremath{\mathbf{e}}}}^0{{{\bigl[{{\ensuremath{{\displaystyle \lim_{y { \rightarrow}0}}}}}\frac{f(y_{t+t_y}^{\uparrow})}{{\bar{s}}(y_{t+t_y}^{\uparrow } ) } { { \ensuremath{\mathbbm{1}}}}_{{t+t_y}<t_b}\bigr ] } } } = { { \ensuremath{\mathbf{e}}}}^0{{{\bigl[\frac{f(y_t^{\uparrow})}{{\bar{s}}(y_t^{\uparrow})}{{\ensuremath{\mathbbm{1}}}}_{t < t_b}\bigr]}}}. \end { split}\ ] ] the second equality follows from the dominated convergence theorem because of @xmath760 right - continuity of the function @xmath761 implies the last equality in .
now we let @xmath694 in and apply monotone convergence to obtain @xmath762 } } } = { { \ensuremath{\mathbf{e}}}}^0{{{\bigl[\frac{f(y_t^{\uparrow})}{{\bar{s}}(y_t^{\uparrow})}{{\ensuremath{\mathbbm{1}}}}_{t < t_\infty}\bigr]}}}.\ ] ] the following estimate justifies the interchange of the limits @xmath763 and @xmath758 @xmath764}}}\bigr|}}}}\\ & \leq c_f{{\ensuremath{{\displaystyle \lim_{b { \rightarrow}\infty}}}}}\sup_{y\leq 1}\frac{1}{{\bar{s}}(y ) } { { \ensuremath{\mathbf{e}}}}^y { { { \bigl[y_t\vee y_t^n { { \ensuremath{\mathbbm{1}}}}_{\sup_{s\leq t}y_s \geq
b}\bigr]}}}\\ & \leq c_f{{\ensuremath{{\displaystyle \lim_{b { \rightarrow}\infty}}}}}\frac{1}{b}\sup_{y\leq 1}\frac{y}{{\bar{s}}(y ) } \sup_{y\leq 1}\frac{1}{y } { { \ensuremath{\mathbf{e}}}}^y{{{\bigl [ \sup_{s\leq t}{{{\bigl(y_s^2+y_s^{n+1}\bigr)}}}\bigr]}}}=0 . \end { split}\ ] ] the last equality follows from @xmath707 and from lemma [ l : finite_moments_y ] . putting and together , we get @xmath765 } } } = { { \ensuremath{{\displaystyle \lim_{b { \rightarrow}\infty}}}}}\,{{\ensuremath{{\displaystyle \lim_{y { \rightarrow}0}}}}}\frac{1}{{\bar{s}}(y)}{{\ensuremath{\mathbf{e}}}}^y { { { \bigl[f(y_t){{\ensuremath{\mathbbm{1}}}}_{t < t_b}\bigr ] } } } = { { \ensuremath{\mathbf{e}}}}^0{{{\bigl[\frac{f(y_t^{\uparrow})}{{\bar{s}}(y_t^{\uparrow})}{{\ensuremath{\mathbbm{1}}}}_{t < t_\infty}\bigr]}}}. \end { split}\ ] ] note that is bounded in @xmath515 because of @xmath752 and lemma [ l : bs_durch_y_bounded ] .
we finish the proof of the first equality in by proving that the limits @xmath757 and @xmath758 on the right - hand side of interchange .
@xmath766 } } } = { { \ensuremath{{\displaystyle \lim_{{{\ensuremath{\varepsilon}}}{\rightarrow}0}}}}}\,{{\ensuremath{\mathbf{e}}}}^0{{{\bigl[\frac{f(y_t^{\uparrow})-f_{{{\ensuremath{\varepsilon}}}}(y_t^{\uparrow})}{{\bar{s}}(y_t^{\uparrow } ) } { { \ensuremath{\mathbbm{1}}}}_{t < t_\infty}\bigr ] } } } = 0 . \end { split}\ ] ] the first equality is with @xmath409 replaced by @xmath767 .
the last equality follows from the dominated convergence theorem .
the function @xmath768 converges to @xmath769 for every @xmath598 as @xmath727 .
note that @xmath770 almost surely for @xmath515 .
integrability of @xmath771 follows from finiteness of .
we have settled equation in lemma [ l : q_explicit_man_hours ] ( together with lemma [ l : finite_man_hours ] ) .
a consequence of the finiteness of this equation is that @xmath772 . in the proof of the extinction result for the virgin island model
, we will need that @xmath773 converges to zero as @xmath74 .
this convergence will follow from equation if @xmath774 is globally upward lipschitz continuous .
we already know that this function is bounded in @xmath93 by lemma [ l : functional_fixed_t ] .
[ l : integrand_ist_nullfolge ] assume [ a : a1 ] , [ a : a2 ] and [ a : s_bar ] .
let @xmath702 .
if @xmath703 , then @xmath775 we will prove that the function @xmath776 is globally upward lipschitz continuous .
the assertion then follows from the finiteness of with @xmath777 replaced by @xmath778 and with @xmath723 .
recall @xmath779 , @xmath780 and @xmath781 from the proof of lemma [ l : finite_moments_y ] . from and it s lemma , we obtain for @xmath598 and @xmath782 @xmath783 where @xmath784 . letting @xmath785 and then @xmath640 , we conclude from the dominated convergence theorem , lemma [ l : finite_moments_y ] and lemma [ l : functional_fixed_t ] that @xmath786}}}\,dr \leq { { \ensuremath{\tilde{c}}}}c_s{{\ensuremath{|t - s|}}}\ ] ] for some constant @xmath787 .
the last inequality follows from lemma [ l : bs_durch_y_bounded ] .
inequality implies upward lipschitz continuity which finishes the proof .
we will derive theorem [ thm : extinction_vim ] from theorem [ thm : general : extinction ] and theorem [ thm : expected_man_hours ] from theorem [ thm : general : expected_man_hours ] .
thus , we need to check that assumptions [ a : general : finite_moments ] , [ a : general : finite_man_hours ] and [ a : convergence_to_zero ] with @xmath788 , @xmath789 and @xmath790 hold under [ a : a1 ] , [ a : a2 ] , [ a : s_bar ] and [ a : finite_man_hours ] . recall that @xmath791 and @xmath792 .
assumption [ a : general : finite_moments ] follows from lemma [ l : finite_moments_y ] and lemma [ l : functional_fixed_t ] .
lemma [ l : finite_man_hours ] and lemma [ l : q_explicit_man_hours ] imply [ a : general : finite_man_hours ] .
lemma [ l : finite_time_extinction ] together with lemma [ l : finite_man_hours ] implies that @xmath33 hits zero in finite time almost surely . the second assumption in [ a : convergence_to_zero ]
is implied by lemma [ l : integrand_ist_nullfolge ] with @xmath793 and assumption [ a : finite_man_hours ] . by theorem [ thm : general : extinction ] , we now know that the total mass process @xmath75 dies out if and only if condition is satisfied . however , by lemma [ l : q_explicit_man_hours ] with @xmath723 and @xmath794 , condition is equivalent to condition .
this proves theorem [ thm : extinction_vim ] for an application of theorem [ thm : general : expected_man_hours ] , note that @xmath237 and @xmath236 are integrable by lemma [ l : finite_man_hours ] and lemma [ l : q_explicit_man_hours ] , respectively .
in addition , lemma [ l : finite_man_hours ] and lemma [ l : q_explicit_man_hours ] show that @xmath795 similar equations hold for @xmath796 and @xmath797 . moreover , the denominators in and are equal by lemma [ l : q_explicit_man_hours ] , equation , together with lemma [ l : finite_man_hours ] .
therefore , equations and follow from equations and , respectively . in the supercritical case , holds because of @xmath798 and lemma [ l : integrand_ist_nullfolge ] with @xmath793 together with assumption [ a : finite_man_hours ] .
furthermore , lemma [ l : functional_fixed_t ] together with lemma [ l : bs_durch_y_bounded ] and the dominated convergence theorem implies continuity of @xmath236 . therefore , theorem [ thm : general : expected_man_hours ] implies which together with reads as .
theorem [ thm : xlogx ] is a corollary of theorem [ thm : general : xlogx ] . for this , we need to check [ a : for_skdist ] .
the expression in is finite because of lemma [ l : functional_fixed_t ] with @xmath799 and assumption [ a : finite_man_hours_squared ] .
assumption [ a : a1 ] provides us with @xmath800 for all @xmath608 and some @xmath801 .
thus , @xmath802 which is finite by [ a : finite_man_hours_squared ] .
lemma [ l : integrand_ist_nullfolge ] with @xmath803 and lemma [ l : finite_moments_y ] show that @xmath804 is bounded in @xmath55 .
furthermore , hlder s inequality implies @xmath805}}}{\bar{q}_y}(d\chi)\bigr)}}}^2 \leq\int \chi_t^2\,{\bar{q}_y}(d\chi ) \int{{{\bigl(\int_0^\infty a(\chi_s)\,ds\bigr)}}}^2{\bar{q}_y}(d\chi)\ ] ] which is bounded in @xmath55 because of lemma [
l : q_explicit_man_hours ] with @xmath724 , @xmath794 and because of assumption [ a : finite_man_hours_squared ] .
therefore , we may apply theorem [ thm : general : xlogx ] . note that the limiting variable is not identically zero because of @xmath806 the right - hand side is finite because of lemma [ l : q_explicit_man_hours ] with @xmath724 , @xmath794 and because of assumption [ a : finite_man_hours_squared ] . | a continuous mass population model with local competition is constructed where every emigrant colonizes an unpopulated island . the population founded by an emigrant
is modeled as excursion from zero of an one - dimensional diffusion . with this excursion measure
, we construct a process which we call virgin island model . a necessary and sufficient condition for extinction of the total population
is obtained for finite initial total mass . |
for quite some time it was not known whether there exist seven points in the euclidean plane , no three on a line , no four on a circle , with pairwise integral distances .
kreisel and kurz @xcite found such a set of size @xmath3 , but it is unknown if there exists one of size @xmath4.-clusters . ] the hunt for those point sets was initiated by ulam in 1945 by asking for a dense point set in the plane with pairwise rational distances .
here we study a kind of opposite problem , recently considered by the authors for euclidean spaces , see @xcite : given a normed space @xmath1 , what is the maximum volume @xmath5 of an open set @xmath6 with @xmath7 connected components without a pair of points at integral distance .
we drop some technical assumptions for the normed spaces @xmath1 and mostly consider the euclidean spaces @xmath8 or @xmath9 equipped with a @xmath10-norm .
in theorem [ main_thm ] we state an explicit formula for the euclidean case @xmath11 .
we assume that our normed space @xmath1 admits a measure , which we denote by @xmath12 . by @xmath13
we denote the open ball with diameter one in @xmath1 , i.e. the set of points with distance smaller than @xmath14 from a given center . in the special case @xmath15 with @xmath16 , where @xmath17 , we have @xmath18 where @xmath19 denotes the famous gamma function , i.e. the extension of the factorial function . in the manhattan metric ,
i.e. @xmath20 , the volumes of the resulting cross - polytopes equal @xmath21 and in the maximum norm , i.e.@xmath22 , the volumes of the resulting hypercubes equal @xmath23 . at
first we observe that the diameter of a connected component @xmath24 of a set @xmath6 avoiding integral distances is at most @xmath23 . in @xmath1
is at most @xmath25 , see e.g. @xcite . ]
otherwise we can consider two points @xmath26
having a distance larger than @xmath23 and conclude the existence of a point @xmath27 on the curve connecting @xmath28 with @xmath29 in @xmath24 such that the distance between @xmath28 and @xmath29 is exactly @xmath23 . given two points @xmath30 , where @xmath31 , we may consider the line @xmath32 .
the restriction of @xmath1 to @xmath33 yields another , one - dimensional , normed space , where we can pose the same question .
we consider the map @xmath34 , @xmath35 where @xmath36 .
if @xmath37 is not injective on @xmath38 , the set @xmath0 contains a pair of point an integral distance apart .
since the map @xmath37 is length preserving modulo @xmath23 , that is , @xmath39 , its restriction to the connected components of @xmath40 is length preserving .
thus , we conclude the necessary condition for an open set avoiding integral distances that the length of each intersection with a line is at most @xmath23 .
having those two necessary conditions , i.e. the diameter of each connected component is at most @xmath23 and the length of each line intersection is at most @xmath23 , at hand we define @xmath41 as the maximum volume of open sets in @xmath1 with @xmath7 connected components , which satisfy the two necessary conditions .
we thus have @xmath42 for all @xmath43 .
in @xcite the authors provided an example of a connected open set @xmath44 such that the intersection of @xmath45 with each line has a total length of at most @xmath23 but the volume of @xmath45 is unbounded . based on a simple averaging argument ,
any given upper bound on one of the two introduced maximum volumes for @xmath7 components induces an upper bound for @xmath46 components in the same normed space @xmath1 : for each @xmath46 we have , [ lemma_averaging ] * @xmath47 whenever @xmath48 ; * @xmath49 whenever @xmath50 . for lower bounds we consider the following construction .
given a small constant @xmath51 , we arrange one open ball of diameter @xmath52 and @xmath53 open balls of diameter @xmath54 each , so that there centers are aligned on a line and that they are non - intersecting .
since everything fits into an open ball of diameter @xmath23 there can not be a pair of points at integral distance . as @xmath55 the volume of the constructed set approaches @xmath25 , so that we have @xmath56 we then have @xmath57 .
the map @xmath37 shows that equality is also attained for normed spaces @xmath1 of dimension @xmath23 .
so , in the following we consider sets consisting of at least two components and normed spaces of dimension at least two .
for one component the extremal example was the open ball of diameter @xmath23 . by choosing the line through the centers of two balls of diameter @xmath23
we obtain a line intersection of total length two , so that this can not happen in an integral distance avoiding set . the idea to circumvent
this fact is to truncate the open balls in direction of the line connecting the centers so that both components have a width of almost @xmath14 , see figure [ fig_two_component_construction_euclidean_plane ] for the euclidean plane @xmath58 .
components in @xmath58 without integral distances . ]
the volume @xmath59 of a convex body @xmath60 with diameter @xmath61 and minimal width @xmath62 is bounded above by @xmath63 see e.g. ( * ? ? ?
* theorem 1 ) .
equality holds iff @xmath64 is the @xmath2-dimensional spherical symmetric slice with diameter @xmath61 and minimal width @xmath62 .
we can easily check that the maximum volume of two @xmath2-dimensional spherical symmetric slices with diameter @xmath23 each and minimal widths @xmath65 and @xmath66 , respectively , so that @xmath67 is attained at @xmath68 , independently from the dimension .
motivated by this fact we generally define @xmath69 to be a spherical symmetric slice with diameter @xmath23 and width @xmath14 , i.e. a truncated open ball . in the @xmath2-dimensional euclidean case
we have @xmath70 the truncated disc in dimension @xmath71 has an area of @xmath72 . for the maximum norm , @xmath22 , and the manhattan metric , @xmath20 . ] on the left hand side of figure [ fig_spherical_slices ] we have drawn the spherical slice , i.e. a truncated ball , in dimension @xmath73 for the maximum norm , i.e. @xmath22 .
for general dimension @xmath2 we have @xmath74 . on the right hand side of figure [ fig_spherical_slices ] we have drawn the spherical slice in dimension @xmath73 for the manhattan metric , i.e. @xmath20 .
here we have @xmath75 .
since the line through the upper left _ corner _ of the left component and the lower right _ corner _ of the right component should have an intersection with the shaded region of total length at most @xmath23 , we consider truncated open balls of diameter @xmath76 and width @xmath77 , see figure [ fig_two_component_construction_euclidean_plane ] .
for some special normed spaces @xmath1 we can choose @xmath78 and move the centers of the two components sufficiently away from each other such that we can guarantee that no line intersection has a total length of more than @xmath23 .
[ lemma_two_components ] for arbitrary dimension @xmath79 and normed spaces @xmath15
with @xmath80 we have @xmath81 . for a given small @xmath78 place a truncated open ball of diameter @xmath76 and width @xmath77 with its center at the origin and a second copy
so that the two centers are at distance @xmath82 , see figure [ fig_two_component_construction_euclidean_plane ] for @xmath83 . since both components have diameters smaller than @xmath23 there can not be a pair of points at integral distance within the same component .
so let @xmath84 be a point of the left and @xmath85 a point of the right component , where we assume that the centers of the components are moved apart along the first coordinate axis . by construction
the distance between @xmath86 and @xmath87 is at least @xmath88 .
since @xmath89 for all @xmath90 and @xmath91 the distance between @xmath86 and @xmath87 is less than @xmath92 by choosing a sufficiently large integer @xmath88 we can guarantee that this term is at most @xmath93 , so that there is no pair of points at integral distance .
finally , we consider the limit as @xmath55 .
we conjecture that the lower bound from lemma [ lemma_two_components ] is sharp for all @xmath94 and remark that this is true for the euclidean case @xmath95 by inequality ( [ ie_isoperimetric_width ] ) .
in this section we restrict the connected components to open balls of diameter @xmath14 .
we remark that if an integral distance avoiding set contains an open ball of diameter @xmath96 , which fits into one of its components , then the other components can contain open balls of diameter at most @xmath97 .
one can easily check that the maximum volume of the entire set @xmath0 is attained at @xmath98 , at least for @xmath10-norms and dimensions @xmath79 .
if the set @xmath0 , i.e. the collection of @xmath7 open balls of diameter @xmath14 , does not contain a pair of points at integral distance , then the mutual distances between centers of different balls have to be elements of @xmath99 .
therefore dilating @xmath0 by a factor of @xmath73 yields the set @xmath100 of the centers of the @xmath7 balls with pairwise odd integral distances .
however , for euclidean spaces @xmath8 , it is known , see @xcite , that @xmath101 , where equality is possible if and only if @xmath102 .
it would be interesting to determine the maximum number of odd integral distances in other normed spaces .
assuming that the construction using truncated open balls from section [ sec_two_components ] is best possible or , at the very least , competitive , we can try to arrange @xmath7 copies of those @xmath69 .
since we have to control that each line meets at most two components we can not arrange the centers on a certain line . on the other hand , the cutting directions , i.e. the directions where we cut of the caps from the open balls , should be almost equal . to meet both requirements , we arrange the centers of the components on a parabola , where each component has diameter @xmath76 and width @xmath77 , for a small constant @xmath78 , see figure [ fig_n_component_construction ] for an example in @xmath58 .
[ lemma_n_components ] for arbitrary dimension @xmath79 , @xmath103 , and normed spaces @xmath15
with @xmath94 we have @xmath104 . for a given small @xmath78 consider @xmath7 truncated @xmath2-dimensional balls @xmath69 of width @xmath77 , where the truncation is oriented in the direction of the @xmath105-axis , with centers located at @xmath106 and diameter @xmath76 for @xmath107 , see figure [ fig_n_component_construction ] .
for @xmath88 large , there is no line intersecting three or more components .
it remains to check that each line meeting two @xmath2-dimensional balls centered at @xmath108 , @xmath109 has an intersection of length at most @xmath14 with each of the truncated balls .
this can be done by performing a similar calculation as in the proof of lemma [ lemma_two_components ] .
again , we consider the limiting configuration as @xmath55 .
we conjecture that the lower bound in lemma [ lemma_n_components ] is sharp .
in order to modify the construction from the previous section to obtain a lower bound for @xmath5 , we use results from diophantine approximation and : [ lemma_diagonals_p_gon ] given an odd prime @xmath10 , let @xmath110 for @xmath111 , where the @xmath112 are @xmath113th roots of unity
. then the @xmath114 are irrational and linearly independent over @xmath115 . a proof , based on a theorem of vanishing sums of roots of unity by mann @xcite , is given in @xcite .
this result can be applied to construct sets avoiding integral distances as follows .
we fix an odd prime @xmath10 with @xmath116 . for each integer @xmath117 and each @xmath118
we consider a regular @xmath10-gon @xmath119 with side lengths @xmath120 , i.e. with circumradius @xmath88 . at @xmath7
arbitrarily chosen vertices of the @xmath10-gon @xmath119 we place the centers of @xmath2-dimensional open balls with diameter @xmath76 .
since the diameter of each of the @xmath7 components is less than @xmath23 there is no pair of points at integral distance inside one of these @xmath7 components . for each pair of centers @xmath121 and @xmath122
, we cut off the corresponding two components such that each component has a width of @xmath77 in that direction , see figure [ fig_pentagon_construction ] for an example with @xmath123 .
-gon construction : integral distance avoiding point set for @xmath71 and @xmath124.,width=188 ] next we consider two points @xmath86 and @xmath87 from different components .
we denote by @xmath125 the distance of the centers of the corresponding components . from the triangle inequality
we conclude @xmath126 since the occurring distances @xmath125 are given by @xmath127 for @xmath111 we look for a solution of the following system of inequalities @xmath128 with @xmath129 , where @xmath130 denotes the positive fractional part of a real number @xmath131 , i.e. there exists an integer @xmath132 with @xmath133 and @xmath134 . by lemma [ lemma_diagonals_p_gon ] the factors @xmath135 are irrational and linearly independent over @xmath115 , so by weyl s theorem @xcite the systems admit a solution for all @xmath88 .
( actually we only use the denseness result , which weyl himself attributed to kronecker . )
we call the just described construction the @xmath10-gon construction . these ingredients we provided in more detail in @xcite
enable us to establish the conjectured exact values of the function @xmath136 : [ main_thm ] for all @xmath137 we have @xmath138 ( sketch ) + for a given number @xmath7 of components we choose an increasing sequence of primes @xmath139 .
for each @xmath140 we consider the @xmath10-gon construction with @xmath141 and @xmath7 neighbored vertices .
if the scaling factor @xmath88 tends to infinity the @xmath2-dimensional volume of the resulting sequence of integral distance avoiding sets tends to the volume of the following set : let @xmath142 arose as follows .
place an open ball of diameter @xmath23 at @xmath7 neighbored vertices of a regular @xmath143-gon with radius @xmath144 .
cut off caps in the direction of the lines connecting each pair of centers so that the components have a width of @xmath14 in that direction . in order to estimate the volume of @xmath142 we consider another set @xmath145 , which arises as follows .
place an open ball of diameter @xmath146 at @xmath7 neighbored vertices of a regular @xmath143-gon with radius @xmath144 .
cut off caps in the direction of the @xmath147-axis so that the components have a width of @xmath148 .
one can suitably choose @xmath149 so that @xmath145 is contained in @xmath142 .
since the centers of the components of @xmath142 tend to be aligned , as @xmath150 increases , @xmath149 tends to @xmath151 as @xmath150 approaches infinity .
we thus have @xmath152 .
we have proposed the question for the maximum volume of an open set @xmath0 consisting of @xmath7 components in an arbitrary normed space @xmath1 avoiding integral distances . for the euclidean plane
those sets need to have upper density @xmath151 , see @xcite .
theorem [ main_thm ] proves a conjecture stated in @xcite and some of the concepts have been transferred to more general spaces .
nevertheless , many problems remain unsolved and provoke further research .
h. furstenberg , y. katznelson , and b. weiss , _ ergodic theory and configurations in sets of positive density _ , mathematics of ramsey theory , coll .
graph theory , prague / czech .
, algorithms comb . 5 , 184198 , 1990 .
hernndez cifre , g. salinas , and s. segura gomis , _ two optimization problems for convex bodies in the @xmath7-dimensional space _ , beitrge algebra geom .
( contributions to algebra and geometry ) * 45 * ( 2004 ) , no . 2 , 549555 . | we study open sets @xmath0 in normed spaces @xmath1 attaining a large volume while avoiding pairs of points at integral distance . the proposed task is to find sharp inequalities for the maximum possible @xmath2-dimensional volume .
this problem can be viewed as an opposite to known problems on point sets with pairwise integral or rational distances . |
cooperative behavior is essential for the maintenance of public resources and for their preservation for future generations @xcite .
however , human cooperation is often threatened by the lure of short - term advantages that can be accrued only by means of freeriding and defecting .
bowing to such temptations leads to an unsustainable use of common resources , and ultimately such selfish behavior may lead to the `` tragedy of the commons '' @xcite .
there exist empirical and theoretical evidence in favor of the fact that our climate is subject to exactly such a social dilemma @xcite . and
recent research concerning the climate change has revealed that it is in fact the risk of a collective failure that acts as perhaps the strongest motivator for cooperative behavior @xcite .
the most competent theoretical framework for the study of such problems , inspired by empirical data and the fact that failure to reach a declared global target can have severe long - term consequences , is the so - called collective - risk social dilemma @xcite .
as the name suggests , this evolutionary game captures the fact discovered in the experiments that a sufficiently high risk of a collective failure can significantly elevate the chances for coordinating actions and for altogether avoiding the problem of vanishing public goods .
recent research concerning collective - risk social dilemmas has revealed that complex interaction networks , heterogeneity , wealth inequalities as well as migration can all support the evolution of cooperation @xcite ( for a comprehensive review see @xcite ) . moreover , sanctioning institutions can also promote public cooperation @xcite . more specifically , it has been shown that a decentralized , polycentric , bottom - up approach involving multiple institutions instead of a single , global one provides significantly better conditions both for cooperation to thrive as well as for the maintenance of such institutions in collective - risk social dilemmas @xcite .
voluntary rewards have also been shown to be effective means to overcome the coordination problem and to ensure cooperation , even at small risks of collective failure @xcite .
the study of collective - risk social dilemmas can thus inform relevantly on the mitigation of global challenges , such as the climate change @xcite , but it is also important to make further steps towards more realistic and sophisticated models , as outlined in the recent review by pacheco et al .
@xcite and several enlightening commentaries that appeared in response . here
we consider the collective - risk social dilemma , where in addition to the standard personal endowments , players own additional assets that are prone to being lost if the collective target is not reached . indeed , individual asset has been considered in the behavioral experiments regarding the collective - risk social dilemma @xcite .
however , different from the experimental study that investigates the interaction between wealth heterogeneity and meeting intermediate climate targets @xcite , we here explore in detail whether and how the so - called risky assets provide additional incentives for individuals to cooperate in well - mixed and structured populations .
it is important to emphasize that , within our setup , individuals might lose more from a failed collective action than they can gain if the same action is successful .
naturally , this constitutes an important feedback for the selection of the most appropriate strategy . a simple example from real life to illustrate the relevance of our approach is as follows : imagine farmers living around a river that often floods .
the farmers needs to invest into a dam to prevent the floods from causing damage .
if the farmers cooperate and successfully build the dam , they will be able to enjoy the harvest .
however , if the farmers fail to build the dam , they will lose not only the harvest , but they will also incur property damage to their fields , houses and stock .
further to the motivation of our research , it is also often the case that individuals have limited investment capabilities , which they have to carefully distribute among many groups . in other words
, individuals may participate in several collective - risk social dilemmas , for example in each with a constant contribution @xcite .
rationally , however , individuals tend to allocate their asset into groups so as to avoid , or at least minimize , potential losses based on the information concerning risk in the different groups . to account for these realistic scenarios , we consider the collective - risk social dilemma with risky assets in finite and infinite well - mixed populations , as well as in structured populations .
we first explore how the introduction of risky assets affects the evolutionary dynamics in well - mixed populations , where we observe new stable and unstable mixed steady states , whereby the stable mixed steady state converges to full cooperation in dependence on the risk .
subsequently , we turn to structured populations , where the distribution of assets amongst the groups where players are members becomes crucial . in general
, we will show that the introduction of risky assets can promote the evolution of cooperation even at low risks , both in well - mixed and in structured populations , and by doing so thus contributes to the resolution of collective - risk social dilemmas .
we first consider the simplest collective - risk social dilemma game with constant individual assets . from a well - mixed population ,
@xmath0 players are chosen randomly to form a group for playing the game . in the group ,
each player @xmath1 with the amount of asset @xmath2 can choose to cooperate with strategy @xmath3 or defect with strategy @xmath4 .
cooperators contribute a cost @xmath5 to the collective target while defectors contribute nothing . if all the contributions within the group either reach or exceed the collective target @xmath6 , each player @xmath1 within the group obtains the benefit @xmath7 @xmath8 , such that the net payoff is @xmath9 .
however , if the collective target is not reached , all the players within the group lose their investment and asset with probability @xmath10 , such that the net payoff is then @xmath11 , while with probability @xmath12 the payoff remains the same if the collective target @xmath6 is reached .
based on these definitions , the payoff of player @xmath1 with strategy @xmath13 in a group with @xmath14 cooperators is @xmath15 \nonumber \\&-&ra[1-\theta(j - t)]-cs_y,\end{aligned}\ ] ] where @xmath16 if @xmath17 and @xmath18 otherwise .
we emphasize that an individual will suffer from a risk to lose everything ( the investment and the asset ) it has , if the collective target is not reached in the minimal model .
this is in line with the original definition of the collective - risk social dilemma @xcite .
different from the original model , however , in our case the asset together with the investment can be more than the expected benefit of mutual cooperation . as argued in the introduction , such scenarios do exist in reality , and as we will show in the results section , the risky assets influence significantly the evolutionary dynamics in both well - mixed and structured populations .
we also refer to the appendix for details with regards to the performed analysis .
we here extend the collective - risk social dilemma game with risky assets to be played on the square lattice with periodic boundary conditions , where @xmath19 players are arranged into overlapping groups of size @xmath20 such that everyone is connected to its four nearest neighbors .
accordingly , each individual belongs to five different groups and it participates in five collective - risk games .
concerned for the loss of its assets , each individual @xmath1 aims to transfer these assets into those groups that have a lower probability to fail to reach the collective target . with the information at hand from the previous round of the game , player @xmath1 at time @xmath21 transfers the asset @xmath22 into the group @xmath23 centered on player @xmath24 according to @xmath25^{\alpha}}{\sum_{n\in g_y}[1-r_n(t-1)]^{\alpha}},\ ] ]
where @xmath26 if at time @xmath27 the number of cooperators @xmath28 and @xmath29 otherwise , and @xmath30 is the allocation strength of the asset . here , we mainly consider @xmath31 , given that players generally prefer to allocate their asset into a relatively safe environment .
we note that @xmath32 means allocating the asset equally into all the groups without taking into account the information about risk .
accordingly , we will refer to this allocation scheme as uniform or equal .
conversely , @xmath33 means that individuals allocate their assets only into the most successful groups
. we will refer to this as the fully rational allocation scheme .
lastly , for @xmath34 , we have the so - called bounded rational allocation of assets . in agreement with the above definitions ,
the payoff of player @xmath1 at time @xmath21 with strategy @xmath35 and being member of the group that is centered on player @xmath24 is @xmath36+b(1-r)\{1-\theta[n_c^m(t)-t]\ } \nonumber \\&-&ra_y^m(t)\{1-\theta[n_c^m(t)-t]\}-cs_y(t).\end{aligned}\ ] ] the total payoff at time @xmath21 is then simply the accumulation of payoffs from each of the five individual groups where player @xmath1 is member , given as @xmath37 .
after the accumulation of payoffs as described above , each player @xmath1 is allowed to learn from one randomly chosen neighbor @xmath38 with a probability given by the fermi function @xcite @xmath39^{-1},\ ] ] where in agreement with the settings in finite well - mixed populations ( see appendix for details ) , we use the intensity of selection @xmath40 .
further with regards to the simulations details , we note that initially each player is designated either as a cooperator or defector with equal probability , and it equally allocates its asset into all the groups in which it is involved .
monte carlo simulations are carried out for the population on the square lattice .
we emphasize there exist ample evidence , especially for games that are governed by group interactions @xcite , in favor of the fact that using the square lattice suffices to reveal all the relevant evolutionary outcomes . because the system may reach a stationary state where cooperators and defectors coexist in the finite structured population in the absence of mutation @xcite , we determine the fraction of cooperators when it becomes time - independent .
the final results were obtained over @xmath41 independent initial conditions to further increase accuracy , and their robustness has been tested on populations ranging from @xmath42 to @xmath43 in size .
we begin by showing the results obtained in well - mixed populations , when individual risky assets are incorporated into the collective - risk social dilemma as described by the minimal model . in infinite well - mixed populations , according to the replicator equation , it can be observed that only the presence of a high risk leads to two additional , interior equilibria beside the two boundary equilibria @xmath44 ( stable ) and @xmath45 ( unstable ) in the standard collective - risk social dilemma ( appendix a ) @xcite . the unstable interior equilibrium , if it exists , divides the range @xmath46 $ ] of @xmath47 into two basins of attraction . in other words , when there is no risk or a low risk , the population has no interior equilibria , and the only stable equilibrium is at @xmath44 , corresponding to the emergence of full defection [ fig .
[ fig1](a ) ] .
however , in this unfavorable situation , we find that the introduction of risky assets can lead to the emergence of two mixed internal equilibria , which renders cooperation viable and in fact yields similar effects as a high risk environment .
furthermore , we find that as the asset increases , the stable interior equilibrium rapidly increases closely to one , while the unstable interior equilibrium rapidly decreases closely to zero [ fig .
[ fig1](a ) ] ( see appendix a for more analytical results ) . in finite well - mixed population ,
we present the stationary distribution of cooperation for different values of asset @xmath2 for a population of size @xmath48 , where the group size @xmath20 and the threshold @xmath49 have been used , as shown in fig .
[ fig1](b ) .
it is worth pointing out that the stationary distribution characterizes the pervasiveness in time of a given configuration of the population in the presence of behavioral mutations ( see appendix for details ) .
we see that in the absence of individual assets , the population spends most of the time in configurations where defectors prevail , especially at low risks of collective failure .
however , when risky assets are introduced , the population begins to spend more time in configurations where cooperative behavior thrives .
in particular , as the asset increases to a high value , the population spends most of the time in configurations where cooperators prevail , while it spends little time in configurations where defectors prevail .
these results highlight that the introduction of risky assets into the theoretical model of the collective - risk social dilemma @xcite is found to raise the chances of coordinating actions and escaping the tragedy of the commons . accordingly , we conclude that individual assets significantly enhance the positive impact of risk on the evolution of cooperation .
we continue with presenting the results obtained in structured populations , where at each time step an individual participates in five collective - risk dilemma games and allocates its assets as described by the extended model .
figure [ fig2 ] shows the fraction of cooperation in the stationary state for two different values of the risk @xmath10 .
it can be observed that the fraction of cooperators increases steadily with increasing the value of asset @xmath2 , which is in agreement with the results obtained in well - mixed populations .
moreover , the fraction of cooperators increases with increasing @xmath2 irrespective of the value of risk @xmath10 and the allocation strength @xmath30 .
more precisely , when the risk is small [ fig .
[ fig2](a ) and ( c ) ] , defectors always dominate if @xmath50 , and this regardless of the value of @xmath30 . on the other hand , if @xmath51 , with increasing @xmath30 the fraction of cooperators first increases , reaches a maximum , but then decreases slowly .
this indicates that there exists an optimal allocation strength which can maximize the level of cooperation .
in other words , neither uniform allocation nor completely rational allocation is optimal if the risk @xmath10 is small .
instead , only bounded rational allocation ensures the highest cooperation levels . when the risk is large [ fig .
[ fig2](b ) and ( d ) ] the fraction of cooperators still increases with increasing the asset value @xmath2 , but decreases with increasing the allocation strength @xmath30 .
specifically , for @xmath52 the fraction of cooperators monotonously decreases with increasing @xmath30 . for larger assets ,
however , full cooperation is achieved by uniform allocation , but the same can also be attained with @xmath30 values that are somewhat larger than zero . but as the value of @xmath30 increases further , the fraction of cooperators slowly decreases .
this can be counteracted by increasing the value of @xmath2 , since then the range of @xmath30 values where full cooperation is attained increases . taken together ,
these results indicate that completely rational allocation of assets is not optimal for the successful evolution of cooperation . instead , in low - risk situations bounded rational allocation of assets works best , while in high - risk situations the simplest uniform distribution of assets among all the groups is optimal . in order to study how the reported dependence of the fraction of cooperators on allocation strength relies on the value of risk , we first show the stationary fraction of cooperators as a function of risk and allocation strength together at a certain asset value @xmath53 in fig .
[ fig3](a ) .
it can be observed that for risk @xmath54 defectors always dominate the population , regardless of the value of @xmath30 . for @xmath55 ,
there exists an intermediate value of @xmath30 that maximizes the fraction of cooperators . for @xmath56 ,
the fraction of cooperators decreases with increasing @xmath30 . finally , for even larger @xmath10 values
, cooperators always dominate the population .
results presented in fig . [ fig3](b ) show two typical behaviors depicting the dependence of the cooperation level on the allocation strength at intermediate values of the risk , as reported in fig .
we stress that the range of the risk values for the typical outcomes depends on other parameters , but the existence of these results is robust against the variations of the parameters ( see fig .
[ fig5 ] for details ) .
moreover , we compute the optimal value of the allocation strength @xmath57 for several intermediate risk values , as shown in fig .
[ fig3](c ) .
we see that @xmath57 decreases with increasing @xmath10 , and finally reaches zero .
this result highlights that with increasing the risk the optimal allocation scheme of risky assets gradually translates from bounded rational to the completely uniform allocation . to explain these results , we continue by showing a series of snapshots depicting the spatial distribution of strategies over time . when producing the snapshots we use different colors not just for cooperators and defectors , but also for distinguishing whether an individuals central group is successful or not .
more precisely , blue ( yellow ) color denotes cooperators ( defectors ) whose central group succeeds to reach the collective target . on the other hand , green ( pink )
color denotes cooperators ( defectors ) whose central group fails to reach the collective target . in the top three rows of fig .
[ fig4 ] , we show the representative sequences for three different values of @xmath30 , all obtained at a relatively low value of risk .
when the risk is small , defectors have an evolutionary advantage over cooperators @xcite , and they utilize this advantage by gathering the benefit and ultimately becoming a successful strategy . for low @xmath30 ( top row of fig . [ fig4 ] ) , both cooperators and defectors allocate the asset equally to all the groups .
the uniform allocation can not reverse the direction of invasion of defectors since both cooperators and defectors may lose a similar amount of asset at a low risk probability if they put the asset into a predominantly defective group .
gradually , the success of defectors easily drives the community into the tragedy of the commons , which is indicated by the emergence of pink defectors .
note that isolated islands of cooperators are in the sea of pink defectors , and finally they disappear completely .
for intermediate values of @xmath30 ( second row of fig .
[ fig4 ] ) , individuals tends to put most of their asset into the successful groups .
thus , islands of cooperators can more easily preserve their assets than islands of defectors , even if the risk is low .
hence , cooperators do not lose their assets as much as defectors . due to the introduction of risky assets , individual net payoffs , especially the payoffs from a failed group , depends strongly on the risk of collective failure .
thus , grouped blue cooperators are likely to have a higher payoff , at least locally . at the same time , yellow defectors remain successful if they are in the vicinity of cooperators , because then they can also preserve most of their assets and also enjoy most of the benefits from the surrounding successful groups .
but once they invade their neighboring cooperators , they fast become pink defectors . although pink defectors who are around blue cooperators also put most of their assets into the blue cooperators central group , and thus preserve the part of asset , they still put some assets into the neighboring unsuccessful groups where the centered individual is a green cooperator or a pink defector .
groups around green cooperators can partake the assets of pink defectors , which essentially protects the neighboring blue cooperators . although the loss of assets happens with a low probability , it is still sufficiently probable for cooperators being able to resist the invasion of defectors and thus to form a mixed dynamical equilibrium in the stationary state .
when @xmath30 is further increased ( third row of fig .
[ fig4 ] ) , defectors will lose a lower amount of their asset at the low risk probability since they put almost all their assets into the surrounding successful groups , if only such groups are present .
the increment of @xmath30 thus sustains the evolutionary advantage of defectors and the number of blue cooperators consequently decreases .
a few isolated islands of cooperators can withstand the invasion of defectors because they locally get enough support from the grouped companions , but if @xmath30 increases further ( not shown ) , defectors raise to complete dominance .
the two bottom rows of fig .
[ fig4 ] show representative snapshot sequences for two different values of @xmath30 , as obtained at a relatively high risk . for low values of @xmath30 ( fourth
row of fig .
[ fig4 ] ) , blue cooperators do not lose their assets much and thus manage to have a higher net payoff then defectors .
hence , they can form compact clusters and expand across the whole population @xcite . on the contrary , pink defectors have more neighbors who are either also pink defectors or green cooperators .
thus , they lose some of their assets by placing them in the unsuccessful groups when @xmath30 is zero or somewhat larger than zero .
even if @xmath30 is increased , blue cooperators still lose less of their asset than other individuals , and can thus still have the highest net payoff
. therefore they have an evolutionary advantage , and can eventually dominate the whole population . when @xmath30 is sufficiently large ( bottom row of fig . [ fig4 ] ) , both cooperators and defectors tend to allocate their assets predominantly into the successful surrounding groups . but grouped
defectors still lose some of their assets because they simply do not have enough neighboring cooperators that would sustain successful groups .
hence , blue cooperators can still expand across the whole population , although some tiny specks of defectors manage to survive by holding on to the blue cooperators . as a result ,
a few defectors survive in the sea of cooperators .
to conclude , we report on the robustness of our findings by investigating changes in dependence on the cost - to - benefit ratio @xmath58 , the collective target @xmath6 , and the interaction structure , as shown in fig .
[ fig5 ] . when the threshold value changes , as shown in fig .
[ fig5](a ) , we find that there exists an intermediate value of the allocation strength @xmath30 that maximizes the fraction of cooperators at relatively small values of @xmath10 , while the fraction of cooperators decreases with increasing allocation strengths for relatively large values of @xmath10 .
although the range of @xmath10 values for the observation of the two typical behaviors varies as the threshold changes , as indicated by the results presented in fig .
[ fig5](a ) , our main conclusion remain unchanged .
moreover , in fig .
[ fig5](b ) , we find that our results are robust also against the variations of cost - to - benefit ratio @xmath58 .
in particular , as the @xmath58 ratio increases , the evolution of cooperation is impaired , which is in agreement with previous research @xcite . but as we increase the value of the individual assets this trend is again reversed and cooperative behavior is as pervasive as by low @xmath58 ratios .
lastly , in fig . [ fig5](c ) , we change the interaction structure by replacing the von neumann neighborhood with the moore neighborhood .
it can be observed that there still exists an intermediate @xmath30 value that maximizes the fraction of cooperators at relatively small values of @xmath10 , and that the fraction of cooperators decrease with increasing @xmath30 values at relatively large @xmath10 values .
this indicates that our main conclusions are robust also against the changes in the structure of the interaction network .
we have introduced and studied collective - risk social dilemma games with risky assets , and we have demonstrated how this can lead to elevated levels of cooperation in well - mixed and structured populations .
the introduction of risky assets increases the stakes for each individual player , since insufficient contributions to the common pool result not only in the loss of personal endowments , but also in the loss of the assets .
thus , players are more prone to cooperating , and this regardless of their interaction range in the population .
more precisely , we have shown that in infinite well - mixed populations new stable and unstable mixed steady states emerge , whereby the stable mixed steady state converges to full cooperation as either the risk of collective failure or the amount of risky assets increases . in finite well - mixed populations , we have shown that the introduction of risky assets drives the population towards configurations where cooperative behavior abounds .
for comparison , in the absence of risky assets , finite well - mixed populations are prone to spend the majority of time in configurations where defectors prevail . in structured populations , where players have a limited interaction range ,
we have studied an extended collective - risk social dilemma games with risky assets , where the distribution of assets could be tuned by means of an allocation strength parameter . among fully rational , bounded rational and uniform allocation , we have identified the latter as being optimal for the evolution of cooperation in high - risk situations .
conversely , in low - risk situations bounded rational allocation of assets works best .
most surprisingly , the fully rational allocation of assets only to the most successful groups , where in principle the assets would be least prone to being lost , is never optimal .
we have explained these results with characteristic snapshot sequences of strategy distributions in the population , and we have identified pattern formation as being crucial for the observed evolutionary outcomes .
we have also tested the robustness of our results with regards to variations of the cost - to - benefit ratio , the collective target to be reached with the contributions of players , and with regards to the variation of the interaction structure , always observing that at least qualitatively our main conclusions do not change and are fully robust . as we have emphasized in the introduction ,
the consideration of risky assets in the realm of the collective - risk social dilemma game is well aligned with reality , in which it is relatively straightforward to come up with examples where our model could apply .
our research shows that , at least in theory , such risky or unsecured , i.e. , not immune to loss if the collective target is not reached , assets significantly promote cooperation and thus contribute to solving the collective - risk social dilemma .
research based on behavioral experiments has already considered individual assets @xcite , although the focus was on the interaction between wealth heterogeneity and intermediate climate targets .
the experimental results show that , if players collectively face intermediate climate targets , then rich players are willing to substitute for missing contributions by poor players @xcite . following this experimental study , a theoretical work by abou chakra and traulsen @xcite further showed that rich players contribute on behalf of poor players only when their own external assets are worth protecting , and moreover , that rich players maintain cooperation by assisting poor players under a certain degree of uncertainty .
although the motivation behind our work and the setup are different , we hope that , collectively , the demonstrated importance of individual assets will inspire more research along this line , perhaps in the realm of other evolutionary games or in coevolutionary settings .
for studying the evolutionary dynamics in infinite well - mixed populations , we use the replicator equation @xcite . to begin , we assume a large population , a fraction @xmath47 of which is composed of cooperators , the remaining fraction @xmath59 being defectors . accordingly , the replicator equation is @xmath60 where @xmath61 and @xmath62 are the average payoffs of cooperators and defectors , respectively .
next , let groups of @xmath0 individuals be sampled randomly from the population .
the average payoff of cooperators is @xmath63 while the average payoff of defectors is @xmath64 with these definitions , the replicator equation has two boundary equilibria , namely @xmath44 and @xmath45 , whereby full defection is stable while full cooperation is not .
interior equilibria , on the other hand , can be determined by equating @xmath61 and @xmath62 , thus obtaining @xmath65 furthermore , to determine the interior equilibria , we study the slope and the curvature of the function @xmath66 , which we define as @xmath67 note that @xmath68 is equivalent to @xmath69 $ ] . we thus compute @xmath70,\ ] ] from where it follows that , since @xmath71 , @xmath72 has a unique internal root at @xmath73 when @xmath74 .
moreover , @xmath75 for @xmath76 and @xmath77 for @xmath78 .
accordingly , @xmath79 is a unique interior maximum of @xmath66 .
solving the equation @xmath69 $ ] thus yields the following conclusions : 1 . when @xmath80 $ ] , eq .
( [ eq1 ] ) has two interior equilibria , denoted by @xmath81 and @xmath82 with @xmath83 . since @xmath75 for @xmath76 and @xmath77 for @xmath78
, @xmath81 is an unstable equilibrium and @xmath82 is a stable equilibrium .
2 . when @xmath84 $ ] , eq .
( [ eq1 ] ) has only one interior equilibrium @xmath85 , which is a tangent point , and is thus unstable .
3 . when @xmath86 $ ] , eq .
( [ eq1 ] ) has no interior equilibria . when @xmath87 or @xmath88 , however , for @xmath89\geq 1 $ ] eq . ( [ eq1 ] ) has no interior equilibria .
while for @xmath89<1 $ ] , eq .
( [ eq1 ] ) has only one interior equilibrium @xmath90 .
note that @xmath91\}^{1/(n-1)}$ ] is stable for @xmath87 since @xmath92 , while @xmath93\}^{1/(n-1)}$ ] is unstable for @xmath88 since @xmath94 .
for studying the evolutionary dynamics in finite well - mixed populations , we consider a population of finite size @xmath95 . here
the average payoffs of cooperators and defectors in the population with @xmath96 cooperators are respectively given by @xmath97 and @xmath98 next , we adopt the pair - wise comparison rule to study the evolutionary dynamics , based on which we assume that player @xmath1 adopts the strategy of player @xmath38 with a probability given by the fermi function @xmath99^{-1},\ ] ] where @xmath100 is the intensity of selection that determines the level of uncertainty in the strategy adoption process @xcite . without loosing generality
, we use @xmath40 throughout this work . with these definitions ,
the probability that the number of cooperators in the population increases or decreases by one is @xmath101}]^{-1}.\ ] ] following previous research @xcite , we further introduce the mutation - selection process into the update rule , and compute the stationary distribution as the key quantity that determines the evolutionary dynamics in finite well - mixed populations . we note that , in the presence of mutations , the population will never fixate in any of the two possible absorbing states . thus , the transition matrix of the complete markov chain is @xmath102_{(z+1)\times(z+1)}^{t},\ ] ] where @xmath103 if @xmath104 , otherwise @xmath105 , @xmath106 and @xmath107 .
accordingly , the stationary distribution of the population , that is the average fraction of the time the population spends in each of the @xmath108 states , is given by the eigenvector of the eigenvalue @xmath109 of the transition matrix @xmath110 @xcite .
in the results section , fig .
[ fig1](b ) is obtained by using @xmath111 .
this research was supported by the national natural science foundation of china ( grant no .
@xmath112 and no .
@xmath113 ) , the national 973 program of china ( grant no .
2013cb329404 ) , and the slovenian research agency ( grants j1 - 4055 and p5 - 0027 ) . m.p . also acknowledges funding by the deanship of scientific research ( dsr ) , king abdulaziz university , under grant no .
( 76 - 130 - 35-hici ) . | in the collective - risk social dilemma , players lose their personal endowments if contributions to the common pool are too small .
this fact alone , however , does not always deter selfish individuals from defecting .
the temptations to free - ride on the prosocial efforts of others are strong because we are hardwired to maximize our own fitness regardless of the consequences this might have for the public good . here
we show that the addition of risky assets to the personal endowments , both of which are lost if the collective target is not reached , can contribute to solving the collective - risk social dilemma . in infinite well - mixed populations
risky assets introduce new stable and unstable mixed steady states , whereby the stable mixed steady state converges to full cooperation as either the risk of collective failure or the amount of risky assets increases .
similarly , in finite well - mixed populations the introduction of risky assets enforces configurations where cooperative behavior thrives . in structured populations cooperation
is promoted as well , but the distribution of assets amongst the groups is crucial .
surprisingly , we find that the completely rational allocation of assets only to the most successful groups is not optimal , and this regardless of whether the risk of collective failure is high or low . instead , in low - risk situations bounded rational allocation of assets works best , while in high - risk situations the simplest uniform distribution of assets among all the groups is optimal .
these results indicate that prosocial behavior depends sensitively on the potential losses individuals are likely to endure if they fail to cooperate . |
in this article , we consider a family of surface tessellations with interesting discrete spectral gap properties . more specifically , our family of graphs , denoted by @xmath0 ( @xmath1 ) , are trivalent expander graphs tessellating hyperbolic surfaces with large isometry groups growing linear on genus . as we show below , from @xmath2 onwards
, there is no direct relation between our family of graphs and other graphs associated to modular congruence subgroups .
let us first give a brief overview over the construction of @xmath0 , and then go into some details .
we start with a sequence of @xmath3-groups @xmath4 , following the construction in ( * ? ? ?
* section 2 ) .
then we consider @xmath5-valent cayley graphs @xmath6 of these groups and apply @xmath7-transformations in all triangles of @xmath6 , to finally obtain the trivalent graphs @xmath0 .
the @xmath8-transformations are standard operations to simplify electrical circuits , and were also used in @xcite in connection with colin de verdire s graph parameter . the finite groups @xmath4 are constructed as follows .
we start with the infinite group @xmath9 of seven generators and seven relations : @xmath10 where the indices are taken modulo @xmath11 .
as explained in @xcite , this group acts on a thick euclidean building of type @xmath12 .
let @xmath13 , and consider the index two subgroup @xmath14 , generated by @xmath15 .
( note that @xmath16 . ) as explained in ( * ? ? ?
* section 2 ) , we use a representation of the group @xmath17 by infinite ( finite band ) upper triangular toeplitz matrices .
the entries of these toeplitz matrices are elements of the ring @xmath18 ( i.e. , @xmath19-matrices over @xmath20 ) with special periodicity properties .
we denote the group of all these toeplitz matrices by @xmath21 , and by @xmath22 the normal subgroup of matrices whose first @xmath23 upper diagonals are zero .
the groups @xmath4 are then the quotients @xmath24 . the _ finite width conjecture _ in @xcite claims that the groups @xmath4 have another purely abstract group theoretical description via the _ lower exponent-@xmath3 series _
@xmath25 with @xmath26p_{k-1}(g)^2 $ ] for @xmath27 : namely , @xmath28 for @xmath27 ( see ( * ? ? ? * conj .
magma computations confirm this conjecture for all indices up to @xmath29 . for simplicity
, we use the same notation for the elements @xmath30 in @xmath17 and their images in the quotients @xmath4 .
then @xmath31 , and @xmath0 are their @xmath8-transformations .
the graphs @xmath0 can be naturally embedded as tessellations into both compact hyperbolic surfaces @xmath32 and non - compact finite area hyperbolic surfaces @xmath33 .
the edges of the tessellation are geodesics and the vertices are their end points .
our results are given in the following theorem : [ thm : main ] let @xmath1 .
then every eigenvalue @xmath34 of @xmath6 gives rise to a pair @xmath35 of eigenvalues of the bipartite graph @xmath0 .
in particular , there exists a positive constant @xmath36 such that * the graphs @xmath6 are @xmath5-valent expanders with spectrum in @xmath37 \cup \{6\}$ ] , * the bipartite graphs @xmath0 are trivalent expanders with spectrum in @xmath38 \cup \{\pm 3 \}$ ] . moreover , the isometry group of the compact hyperbolic surface @xmath32 has order @xmath39 , where @xmath40 denotes the set of vertices of @xmath0 .
let @xmath41 then we have @xmath42 , @xmath43 and @xmath44 for the vertices , edges and faces of @xmath0 , and all faces of @xmath0 are regular @xmath45-gons .
the genus of @xmath32 can be estimated by @xmath46 the above - mentioned finite width conjecture would imply that the inequalities for the vertices , edges , faces of @xmath0 and the genus of @xmath32 in theorem [ thm : main ] hold with equality .
theorem [ thm : main ] is proved in section [ subsec : main ] .
it is instructive to compare our tessellations @xmath0 to the well studied tessellations of hyperbolic surfaces by _ platonic graphs _
@xmath47 , which are defined as follows .
let @xmath48 be a positive integer @xmath49 .
the vertices of @xmath47 are equivalence classes @xmath50 = \ { \pm ( \lambda,\mu ) \}$ ] with @xmath51 two vertices @xmath50 $ ] and @xmath52 $ ] are connected by an edge if and only if @xmath53 note that every vertex of @xmath47 has degree @xmath48 .
these graphs can also be viewed as triangular tessellations of finite area hyperbolic surfaces @xmath54 , where @xmath55 denotes the hyperbolic upper half plane and @xmath56 is a principal congruence subgroup of the modular group @xmath57 . these and related graphs have been thoroughly investigated by several different communities .
for example , in the general framework of _ regular maps _ , they were studied by d. singerman and co - authors ( see @xcite ) . for odd prime numbers @xmath58 , the graphs @xmath59 have maximal vertex connectivity @xmath60 , diameter @xmath61 , and are ramanujan graphs .
analogous properties hold for the induced subgraphs @xmath62 , where @xmath62 is obtained from @xmath59 by removing the set of vertices @xmath63 $ ] with vanishing second coordinate and all their adjacent edges .
note that @xmath62 is a @xmath64-valent graph tessellating the same surface @xmath65 .
as in the case of our family @xmath0 , the graphs @xmath47 and @xmath66 can also be embedded into smooth compact hyperbolic surfaces , denoted by @xmath67 .
it turns out that the graphs @xmath68 and @xmath69 are isomorphic .
since the valence of the dual graph @xmath70 is a power of @xmath3 , any isomorphism of @xmath70 with a platonic graph @xmath47 would imply @xmath71 with @xmath72 .
however , this leads to a contradiction _ for all _ @xmath2 .
the next proposition summarizes the comparison between our graphs and platonic graphs showing that , generally , _ these two families are of very different nature_. the proof is given in section [ subsec : isomt2pi8 ] . [ prop : isomt2s8 ] the graph @xmath73 is the dual of the platonic graph @xmath69 in the _ unique _ genus 5 hyperbolic surface @xmath74 with maximal automorphism group of order 192 . for @xmath2 , there is no graph isomorphism between @xmath70 and @xmath47 , for any @xmath48 .
let us say a few more words about the platonic graphs @xmath59 and their modifications @xmath62 .
the modified graphs @xmath62 have an alternative description as cayley graphs of the quotients @xmath75 of congruence subgroups ( see end of section [ subsec : vertconnect ] ) .
however , we did not find these modified graphs explicitly in the literature ( for example , they do not appear explicitly in ( * ? ? ?
8.2.3 ) or in @xcite ) .
_ cheeger constant estimates _ for @xmath59 have been obtained , e.g. , in @xcite , but we do not know of any reference for the _ maximal vertex connectivity _ , and present a proof of this fact for both graph families @xmath59 and @xmath62 in section [ subsec : vertconnect ] . to our knowledge , all proofs for the _ ramanujan property _ of the graphs @xmath59 in the literature ( see , e.g. , @xcite ) are based on some amount of number theory ( characters of representations ) .
we think it is remarkable that there is also an easy proof for the ramanujan properties of the graphs @xmath59 and @xmath62 with no reference to number theory other than the irrationality of @xmath76 ( see section [ subsec : ramanujan ] ) .
these facts are summarized in the following theorem .
[ thm : platonic ] let @xmath60 be an odd prime .
then the graphs @xmath59 and @xmath62 have diameter @xmath61 and maximal vertex connectivity @xmath60 and @xmath77 , respectively . moreover , the spectrum of the graph @xmath62 consists of * @xmath77 with multiplicity one , * @xmath78 with multiplicity @xmath77 , * @xmath79 with multiplicity @xmath80 , and * @xmath81 with multiplicity @xmath82 , each . in particular , the graphs @xmath62 are ramanujan .
as mentioned earlier , our family @xmath0 is based on powers of the prime number 2 .
the question whether the ramanujan property of @xmath47 for primes @xmath58 still holds for composite numbers @xmath48 or , at least , for prime powers @xmath83 , was answered in the negative in ( * ? ? ?
4.7 ) . but the ramanujan property for prime powers is preserved if one considers platonic graphs over finite fields @xmath84 instead of the rings @xmath85 ( see @xcite ) .
however , we do not know how these ramanujan graphs associated to prime powers could be naturally embedded into appropriate surfaces .
it is easily checked that any triangular tessellation @xmath86 of a compact oriented surface @xmath87 satisfies @xmath88 , i.e. , the number of edges of every triangulation with at least two vertices is @xmath89 .
therefore , the ratio @xmath90 measures the non - flatness of such a triangulation , i.e. , how effectively the edges of @xmath86 are chosen to generate a surface of high genus .
the following asymptotic results hold for the triangulations @xmath47 and @xmath70 . [
prop : asymptnonflat ] we have @xmath91 and @xmath92 in the second case , note that the dual graph @xmath70 is a triangulation of @xmath32 and that the number of edges of @xmath0 and @xmath70 coincide .
the two formulas in this proposition are proved in sections [ subsec : prop12a ] and [ subsec : prop12b ] .
our trivalent expander graphs @xmath0 can also be used to construct another family of compact hyperbolic surfaces @xmath93 by glueing together regular @xmath94-pieces , as explained in buser @xcite .
the surfaces @xmath93 can be viewed as tubes around the graphs @xmath0 with a hyperbolic metric . using the results in @xcite , the expander properties of @xmath0 translate directly into a uniform lower bound of the first non - trivial eigenvalue @xmath95 of the laplacian on these surfaces .
[ cor : lowerlambda1 ] the compact hyperbolic surfaces @xmath93 ( @xmath1 ) have genus @xmath96 and isometry groups of order @xmath97 .
they form a tower of coverings @xmath98 where all the covering indices are powers of @xmath3 .
there is a positive constant @xmath99 such that we have , for all @xmath23 , @xmath100 corollary [ cor : lowerlambda1 ] is proved in section [ subsec : corlambda1 ] .
there is a well - known classical result by randol @xcite which is in some sense complementary to this corollary .
namely , there exist finite coverings @xmath101 of every compact hyperbolic surface @xmath87 with arbitrarily small first eigenvalues .
it would be interesting to find out whether there are also uniform positive lower bounds for @xmath95 of our other compact hyperbolic surfaces @xmath32 .
it seems that the methods in @xcite are not applicable in this case , since the shapes of the hyperbolic triangles tessellating these surfaces are changing with @xmath23 .
* acknowledgement : * we like to thank hugo parlier for useful discussions .
the explicit construction of the cayley graphs @xmath31 and of the trivalent graphs @xmath0 was explained in the introduction . for the underlying @xmath3-groups @xmath4
, we refer the reader to ( * ? ? ?
* section 2 ) .
let us now construct the hyperbolic surface @xmath33 : we start with a 3-punctured sphere @xmath103 , by glueing together two ideal hyperbolic triangles along their corresponding edges .
note that @xmath103 carries a hyperbolic metric .
it is useful to think of the two ideal triangles of @xmath103 to be coloured black and white .
let @xmath104 be the center of the black triangle .
choose a geometric basis @xmath105 such that @xmath106 is a simple counterclockwise look around the @xmath107-th cusp of @xmath103 and @xmath108 .
the surjective homomorphism @xmath109 given by @xmath110 , @xmath111 and @xmath112 , induces a riemannian covering map @xmath113 .
the surface @xmath114 is a hyperbolic surface , tessellated by @xmath115 ideal hyperbolic triangles , half of them black and the others white .
hurwitz s formula yields @xmath116 where @xmath117 in the case @xmath118 we have @xmath119 and @xmath120 , which leads to @xmath121 @xmath4 acts simply transitive on the black triangles of @xmath114 .
let @xmath122 and @xmath123 be the sets of centers of black and white triangles , respectively .
choose a reference point @xmath124 , and identify the vertices of the cayley graph @xmath6 with the points in @xmath125 by @xmath126 . then
two adjacent vertices in @xmath6 are the centers of two black triangles which share a white triangle as their common neighbour .
the corresponding edge is then the minimal geodesic passing through these three ideal triangles and connecting these two vertices
. we could instead start the process by glueing together two _ compact _ hyperbolic triangles with angles @xmath127 , @xmath128 and @xmath129 , and obtain an orbifold @xmath103 .
the same arguments then lead to an embedding of @xmath6 into a smooth compact hyperbolic surface @xmath87 with the same genus as @xmath114 .
the surface @xmath87 is triangulated by compact black and white triangles , and every black triangle contains a vertex of @xmath6 .
in fact , @xmath4 acts on the surface @xmath87 by isometries .
the abstract @xmath8-transformation of a graph adds a new vertex @xmath130 for every triangle , removes the three edges of this triangle and replaces them by three edges connecting @xmath130 with the vertices of this triangle .
we apply this rule to our graph @xmath6 and obtain a graph @xmath0 , which we can view as an embedding in @xmath87 with the following properties : the vertex set of @xmath0 coincides with @xmath131 , and there is an edge ( geodesic segment ) connecting every black / white vertex in @xmath131 with the vertices in the three neighbouring white / black triangles .
the best way to illustrate this transformation is to present it in the universal covering of the surface @xmath87 , i.e. , the poincar unit disc @xmath132 ( see figure [ fig : dytrafo ] , the new vertices replacing every triangle are green ) . note that @xmath0 has twice as many vertices as @xmath6 , which shows that the isometry group of the above compact surface @xmath133 has order @xmath134 .
moreover , @xmath0 is indeed the dual of the triangulation of @xmath87 by the abovementioned compact black and white triangles .
( left ) and of the @xmath8-transformation @xmath73 ( right ) to the poincar unit disc @xmath132,title="fig:",scaledwidth=40.0% ] ( left ) and of the @xmath8-transformation @xmath73 ( right ) to the poincar unit disc @xmath132,title="fig:",scaledwidth=40.0% ] we first establish the expander properties of @xmath6 and @xmath0 and the relations between their eigenvalues , stated in theorem [ thm : main ] .
it was proved in ( * ? ? ?
* section 2 ) that the group @xmath17 generated by @xmath135 is an index two subgroup of the group @xmath9 in .
@xmath17 is explicitly given by @xmath136 with @xmath137 ( note that our group @xmath9 is denoted in @xcite by @xmath138 , which is reserved for @xmath139 in this paper .
) moreover , both groups @xmath9 and @xmath17 have kazhdan property ( t ) ( see ( * ? ? ?
* section 3 ) ) . using ( * ? ? ?
3.3 ) , we conclude that the cayley graphs @xmath6 are expanders .
the adjacency operator @xmath140 , acting on functions on the vertices of a graph , is defined as @xmath141 note that @xmath142 is a subset of @xmath40 .
we have the following relations between the eigenfunctions of the adjacency operators on @xmath6 and @xmath0 .
[ thm : releigfunc ] * every eigenfunction @xmath143 on @xmath0 to an eigenvalue @xmath144 $ ] gives rise to an eigenfunction @xmath145 to the eigenvalue @xmath146 $ ] on @xmath6 ( with @xmath147 for all @xmath148 ) .
* every eigenfunction @xmath145 on @xmath6 to an eigenvalue @xmath149 - \{3\}$ ] gives rise to two eigenfunctions @xmath150 to the eigenvalues @xmath151 on @xmath0 with @xmath152 * an eigenfunction @xmath145 on @xmath6 to the eigenvalue @xmath153 gives rise to an eigenfunction @xmath143 to the eigenvalue @xmath79 of @xmath0 with @xmath154 if and only if we have , for all triangles @xmath155 in @xmath6 , @xmath156 .
\(a ) let @xmath145 and @xmath143 be two functions on @xmath6 and @xmath0 , related by @xmath147 for all @xmath148 . then @xmath157 which can also be written as @xmath158 .
( note that @xmath159 denotes adjacency in @xmath6 , and @xmath160 is the combinatorial distance in @xmath0 . )
this implies immediately the connection between the eigenfunctions and eigenvalues .
\(b ) let @xmath161 and @xmath150 be defined as in the theorem .
let @xmath162 .
then we have for @xmath148 : @xmath163 and for @xmath164 : @xmath165 note that @xmath166 is well defined since @xmath167 .
\(c ) in the case of @xmath168 we have @xmath169 , and the above calculation for @xmath148 goes through without changes . for @xmath164 ,
the condition @xmath170 translates into the condition that the summation of @xmath145 over the vertices of all the triangles must vanish .
theorem [ thm : releigfunc ] implies that the expander property of the family @xmath6 carries over to the graphs @xmath0 .
moreover , the spectrum of @xmath6 can not contain eigenvalues in the interval @xmath171 , since this would lead to non - real eigenvalues of @xmath0 .
this finishes the proof of the spectral statements in theorem [ thm : main ] . since @xmath6 are cayley graphs of quotients of the group @xmath17 with property ( t ) , not all of these graphs can be ramanujan ( see ( * ? ? ? * prop . 4.5.7 ) ) .
but what can we say about their @xmath8-transformations @xmath0 ?
theorem [ thm : releigfunc ] implies that @xmath0 is ramanujan if and only if the largest non - trivial eigenvalue of @xmath6 is @xmath172 .
magma computations provide the following numerical results : l|l|l graph & number of vertices & largest non - trivial eigenvalue + ' '' '' @xmath173 & 32 & 2.828427124746190 + ' '' '' @xmath174 & 128 & 4.340172973252067 + ' '' '' @xmath175 & 1024 & 4.475244292138809 + ' '' '' @xmath176 & 8192 & 5.160252515773351
this implies that only @xmath173 and @xmath174 are ramanujan ; their largest non - trivial eigenvalue needs to be @xmath177 , which is no longer true for @xmath178 .
moreover , we have @xmath179 , since the graphs @xmath6 are a tower of coverings .
similarly , only @xmath180 are ramanujan , since the covering properties of @xmath6 carry over to their @xmath8-transformations @xmath0 .
finally , we obtain from ( * ? ? ?
2.3 ) that @xmath181 ) holds for all @xmath182 , holds for these indices with equality .
the faces of @xmath0 are determined by the orders of the generators @xmath30 in the group @xmath4 , which we determine next .
let @xmath183 & \alpha_1 = \begin{pmatrix } 0 & 0 & 0 & 0 & 1 & 1 & 0 & 1 & 0\\ 0 & 1 & 0 & 1 & 0 & 0 & 0 & 0 & 1\\ 1 & 1 & 1 & 0 & 0 & 0 & 0 & 1 & 0 \end{pmatrix } , & \beta_1 = \begin{pmatrix } 0 & 0 & 0 & 0 & 1 & 1 & 0 & 1 & 0\\ 0 & 1 & 0 & 1 & 0 & 0 & 0 & 0 & 1\\ 1 & 1 & 1 & 0 & 0 & 0 & 0 & 1 & 0 \end{pmatrix},\\[.2 cm ] & \alpha_3 = \begin{pmatrix } 0 & 0 & 0 & 0 & 1 & 1 & 0 & 1 & 0\\ 0 & 1 & 1 & 1 & 0 & 1 & 0 & 0 & 0\\ 1 & 0 & 0 & 0 & 1 & 1 & 0 & 0 & 1 \end{pmatrix } , & \beta_3 = \begin{pmatrix } 0 & 0 & 0 & 0 & 0 & 1 & 0 & 1 & 1\\ 1 & 1 & 0 & 0 & 1 & 1 & 0 & 0 & 0\\ 0 & 1 & 1 & 0 & 0 & 1 & 1 & 0 & 0 \end{pmatrix}.\end{aligned}\ ] ] [ lm : ordersxi ] we have , in the notation of @xcite , for @xmath184 : @xmath185 this implies , in particular , that @xmath186 with @xmath187 given in .
since @xmath4 is a @xmath3-group , @xmath188 has to be a power of @xmath3 .
the formulas follow from a straightforward calculation from @xmath189 and proposition 2.5 in @xcite .
this implies that @xmath186 if and only if @xmath190 , i.e. , @xmath191 .
lemma [ lm : ordersxi ] implies that the faces of @xmath0 are regular @xmath45-gons . the estimates for @xmath192 and @xmath193 in theorem [ thm : main ] follow immediately from this fact , the inequality , and the trivalence of the graphs @xmath0 .
the genus estimate for @xmath32 can then be deduced from and .
this finishes the proof of theorem [ thm : main ] .
we conclude from the trivalence of @xmath0 and that @xmath194 note that @xmath195 , which implies that @xmath196 recall from and lemma [ lm : ordersxi ] that @xmath197 as @xmath198 , finishing the proof of .
we first recall a few important facts about the platonic graphs @xmath47 and the surfaces @xmath199 and @xmath67 . for more details , see @xcite .
let @xmath200 be the farey tessellation of the hyperbolic upper half plane @xmath55 , and let @xmath201 be the set of oriented geodesics in @xmath200 . recall that the _ farey tessellation _ is a triangulation of @xmath55 with vertices on the line at infinity @xmath202 , namely , the subset of extended rationals @xmath203 .
two rational vertices with reduced forms @xmath204 and @xmath205 are joined by an edge , a geodesic of @xmath55 , if and only if @xmath206 ( see ( * ? ? ?
1 ) for an illustration of the farey tessellation ) . the group of conformal transformations of @xmath55 that leave @xmath200 invariant is the modular group @xmath57 , which acts transitively on @xmath201 .
the _ principal congruence subgroups _ of @xmath138 are normal subgroups defined by @xmath207 it is well known ( see , e.g , @xcite ) that @xmath208 and @xmath47 are isomorphic , and @xmath208 is a triangulation of the surface @xmath209 by ideal triangles ( the vertices are , in fact , the cusps of @xmath199 ) . the tessellation @xmath210 can be interpreted as a _ map _ @xmath211 in the sense of jones / singerman @xcite .
the group @xmath212 of automorphisms of @xmath211 is the group of orientation preserving isometries of @xmath199 preserving the triangulation . as @xmath56 is normal in @xmath138
, we have that the map @xmath211 is _ regular _ , meaning that @xmath212 acts transitively on the set of directed edges of @xmath47 ( see ( * ? ? ?
* thm 6.3 ) ) .
moreover , by ( * ? ? ?
* thm 3.8 ) , @xmath213 ( note that in the case of a prime power @xmath83 , @xmath214 is the group defined over the ring @xmath215 and not over the field with @xmath216 elements . ) let @xmath217 .
noticing that all vertices of @xmath47 have degree @xmath48 , we obtain a smooth compact surface @xmath67 by substituting every ideal triangle in @xmath210 by a compact hyperbolic @xmath218-triangle , and glueing them along their edges in the same way as the ideal triangles of @xmath199 .
the group of orientation preserving isometries of @xmath67 preserving this triangulation is , again , isomorphic to @xmath214 .
hence , the automorphism group of the triangulation @xmath219 is @xmath220 of order 192 .
this implies that @xmath221 is the unique compact hyperbolic surface of genus 5 with maximal automorphism group ( see @xcite ) .
: each triangle corresponds to a hyperbolic @xmath222-triangle of the tessellation of @xmath223 .
the edges along the boundary path are pairwise glued to obtain @xmath221 . ]
the @xmath69-triangulation of @xmath221 is shown in figure [ fig : s8 ] ; the black - white pattern on the triangles is a first test whether this triangulation can be isomorphic to the @xmath68-triangulation of @xmath224 .
( the @xmath47-triangulations for @xmath225 can be found in figs . 3 and 4 of @xcite . )
@xmath220 acts simply transitively on the directed edges of this triangulation .
consider now a refinement of this triangulation by subdividing each @xmath222-triangle into six @xmath226-triangles .
it is easily checked that the smaller @xmath226-triangles admit also a black - white colouring such that the neighbours of all smaller black triangles are white triangles and vice versa .
each black @xmath226-triangle is in 1 - 1 correspondence to a half - edge of @xmath69 which , in turn , can be identified with a directed edge of @xmath69 . consequently , the orientation preserving isometries of the surface @xmath221 corresponding to the elements in @xmath220 act simply transitively on the black @xmath226-triangles .
in fact , @xmath220 can be interpreted as a quotient of the triangle group @xmath227 , namely , @xmath228 where @xmath229 correspond to rotations by @xmath230 about the three vertices of a given @xmath226-triangle .
magma computations show that @xmath220 has a unique normal subgroup @xmath48 of index 6 , generated by the elements @xmath231 , @xmath232 and @xmath233 , which is isomorphic to the triangle group quotient @xmath234 via the explicit isomorphism @xmath235 note that the matrices in , viewed as elements in @xmath139 , generate a group acting simply transitively on the black triangles of the farey tessellation in @xmath55 , as illustrated in figure [ fig : fargen ] .
the images of a black triangle @xmath236 with vertices @xmath237 under @xmath238 are the six black triangles each sharing a common white triangle with @xmath236 . on a triangle @xmath236 with vertices @xmath237 of the farey tessellation.,scaledwidth=80.0% ] magma computations also show that we have the explicit isomorphism @xmath239 given by @xmath240 .
the normal group @xmath241 is of order 32 and the quotient @xmath242 is an orbifold consisting of two hyperbolic @xmath222-triangles ( one of them black and the other white ) .
we conclude from the explicit isomorphism @xmath243 that the covering procedure discussed in section [ subsec : surfs ] leads to isometric surfaces @xmath221 and @xmath224 , and that @xmath244 is dual to the tessellation @xmath245 . on the spectral side , the adjacency operators on the graphs @xmath173 and @xmath69 compare as follows : l|c|c|c|c|c|c|c|c|l eigenvalue & @xmath246 & @xmath5 & @xmath247 & @xmath3 & @xmath79 & @xmath248 & @xmath249 & @xmath250 & total + multiplicity in @xmath173 & @xmath79 & @xmath251 & @xmath5 & @xmath5 & @xmath252 & @xmath253 & @xmath5 & @xmath79 & @xmath254 ' '' '' + multiplicity in @xmath69 & @xmath251 & @xmath79 & @xmath5 & @xmath79 & @xmath253 & @xmath79 & @xmath5 & @xmath3 & @xmath255 ' '' '' we also like to mention that , for @xmath118 , the @xmath8-transformation @xmath256 has a group theoretical interpretation .
there exists a group extension @xmath257 of @xmath258 by @xmath259 , generated by involutions @xmath260 satisfying @xmath261 , @xmath262 and @xmath263 , and @xmath73 is the cayley graph of @xmath257 with respect to the generators @xmath260 .
this group theoretic interpretation of the @xmath8-transformation _ fails _ for @xmath264 . in fact , the group @xmath265 with @xmath266 given in is finite and of order 6144 . if the introduction of the above involutions @xmath260 would lead to a group extension @xmath267 , then @xmath267 would have to be of order @xmath268 and a quotient of @xmath269 and , therefore , of order @xmath270 .
however , we have @xmath271 in contradiction to the second condition .
thus we do not obtain a cayley graph representation of the graphs @xmath0 for @xmath264 via this procedure .
let us finally explain why we can no longer have an isomorphism @xmath272 for @xmath2 , with @xmath273 appropriately chosen .
let us assume that @xmath274 .
comparison of the vertex degrees of @xmath275 and @xmath70 leads to @xmath276 , with @xmath187 given in .
moreover , we conclude from below that @xmath277 . we know from theorem [ thm : main ] that @xmath278 the condition @xmath279
together with @xmath276 implies that @xmath280 with @xmath281 in , i.e. , @xmath282 but one easily checks that this inequality holds only for @xmath283 .
( in the case @xmath284 , we have @xmath285 , since @xmath286 is combinatorially the cube and @xmath287 is the octagon . )
this shows that the graph family @xmath47 can not contain any of the dual graphs @xmath70 , for indices @xmath2 .
the identity @xmath288 between the genus of the surface @xmath93 and the number of vertices of the trivalent graph @xmath0 is easily checked .
moreover , every automorphism of the graph @xmath0 induces an isometry on @xmath93 . since the graphs @xmath0 form a power of coverings with powers of @xmath3 as covering indices , the same holds true for the associated surfaces @xmath289 .
now , @xcite shows that @xmath290 can be estimated from below by a fixed multiple of the isoperimetric cheeger constant of @xmath0 .
the expander property implies that the cheeger constants of @xmath0 have a uniform lower positive bound .
this finishes the proof of the corollary [ cor : lowerlambda1 ] .
let us briefly recall some algebraic facts from @xcite . both groups @xmath291 and @xmath214 act on @xmath292 via @xmath293
= [ a \lambda+b \mu , c \lambda+ d \mu],\ ] ] and there is a 1 - 1 correspondence between the vertex set @xmath292 and the cosets @xmath294 . here , @xmath295 is the congruence subgroup given by @xmath296 in @xcite , the set of vertices was partitioned into _ axes_. two vertices belong to the same axis if they have the same stabilizer in @xmath214 .
since @xmath214 acts transitively on @xmath292 , all axes have the same number of vertices .
an interesting observation is that if an element of @xmath214 leaves a vertex @xmath50 $ ] invariant , then any vertex @xmath52 $ ] with @xmath297 is also invariant under the same element .
thus the axis containing @xmath298 $ ] is given by @xmath299 \mid \gcd(\lambda , n ) = 1 \},\ ] ] and we call this axis the _ principal axis _ of @xmath47 .
the set of all axes of @xmath47 is denoted by @xmath300 .
there is a 1 - 1 correspondence between the axes of @xmath47 and the cosets @xmath301 , where @xmath302 is the congruence subgroup @xmath303 from the 1 - 1 correspondences with the cosets of @xmath304 , we can immediately obtain the numbers of directed edges , vertices and axes of @xmath47 as the indices of these subgroups in @xmath138 : @xmath305 where the products run over the distinct prime divisors of @xmath48 , see for example @xcite .
in particular , for a prime @xmath60 , @xmath59 has @xmath306 axes , @xmath307 vertices and @xmath308 undirected edges .
let @xmath60 be a fixed odd prime and @xmath314 .
the _ wheel structure _ of @xmath59 was already discussed in ( * ? ? ?
* thm 2.1 ) .
let us present this and other geometric facts in our terminology .
the principal axis of @xmath59 is given by @xmath315 \in v(\pi_p ) \mid 1 \le i \le n \}.\ ] ] the vertices of @xmath316 and their @xmath251-ring neighbours form a partition of @xmath317 into @xmath318 components with @xmath306 vertices each .
we call these components the _ wheels _ of @xmath59 , see figure [ fig : wheels ] . the wheel with center @xmath319 $ ] ( @xmath320 )
is denoted by @xmath321 and is a subgraph of @xmath59 with @xmath306 vertices and @xmath322 edges .
we also use the notation @xmath323 for the induced subgraph with vertex set @xmath324 \}$ ] .
we call @xmath323 the boundary of the @xmath107-th wheel .
note that @xmath323 is isomorphic to the cyclic graph of @xmath60 vertices .
every vertex that is not in @xmath316 is adjacent to exactly two vertices of the boundary of any given wheel @xmath321 , @xmath325 . indeed , because @xmath326 acts transitively on @xmath317 , we may consider , w.l.o.g . , the vertex @xmath327
\in \partial w_1 $ ] .
the @xmath77 vertices adjacent to @xmath327 $ ] that are not in @xmath316 are @xmath328 $ ] with @xmath329 . to find the vertices
@xmath328 $ ] in @xmath323 , we need to solve @xmath330 which has exactly two solution @xmath331 ( where we think of @xmath332 ) which correspond to two distinct vertices of @xmath59 .
consists of @xmath333 wheels .
each vertex at the boundary of a wheel is connected with the center of its wheel and exactly two points on the boundary of any wheel ( including itself).,scaledwidth=80.0% ] * let @xmath335 be two different vertices in @xmath323 and also @xmath336 be two different vertices in the same set @xmath323 ( @xmath337 is allowed ) .
then there exists a permutation @xmath338 and two vertex distinct paths @xmath339 in @xmath323 , such that @xmath340 connects @xmath341 with @xmath342 and @xmath343 connects @xmath344 with @xmath345 . *
every @xmath346 has precisely two neighbours in @xmath347 . *
assume additionally that @xmath348 .
then there exists a bijective map @xmath349 such that @xmath350 for all vertices @xmath351 .
note that @xmath323 is isomorphic to the cyclic graph of @xmath60 vertices .
( a ) is then a straighforward inspection of all possible cases .
( b ) is already proved by our previous arguments .
it remains to prove ( c ) : think of @xmath352 .
then the vertices in @xmath323 are of the form @xmath353 $ ] and the vertices in @xmath347 of the form @xmath354 $ ] with @xmath355 .
the map @xmath356 , defined by @xmath357 , is obviously a bijection , and we have @xmath353 \sim [ \phi(\mu),j^{-1}]$ ] , finishing the proof . note that the wheel structure is not confined to the choice of the principal axis .
since the group @xmath326 maps axes to axes and acts transitively on them , we can choose any axis @xmath358 as the centers of the @xmath318 wheels , and lemma [ lm : help ] is still valid in this setting .
ad ( i ) : assume that the end vertex is @xmath363 $ ]
. then we already have three vertex disjoint paths given by @xmath364 \to [ \nu,1 ] , \quad [ 1,0 ] \to [ \nu\pm1,1 ] \to [ \nu,1].\ ] ] we need to find vertex disjoint paths starting with @xmath298 \to [ \nu\pm i,1]$ ] and ending at @xmath363 $ ] , for @xmath365 . by lemma [ lm : help](c )
, we can find two different vertices @xmath366 such that @xmath367 \sim x_1 $ ] and @xmath368 \sim x_2 $ ] . by lemma [ lm : help](b ) ,
@xmath363 $ ] has two different neighbours @xmath369 in @xmath323 .
we now use lemma [ lm : help](a ) to complete the paths . ad ( ii ) : we assume @xmath370 , for otherwise there is nothing to prove .
let us assume that the end vertex is in @xmath323 with @xmath365 , and let us denote this vertex by @xmath371 .
let @xmath372 be the two neighbours of @xmath373 in the first wheel .
choose three different vertices @xmath374 , and use lemma [ lm : help](c ) to find three different vertices @xmath375 such that @xmath376 for @xmath377 . w.l.o.g .
, we can assume that the pair @xmath378 separates @xmath379 and @xmath373 within @xmath323 .
let @xmath380 be the two vertex disjoint paths in @xmath381 connecting @xmath373 with @xmath382 and @xmath383 , respectively .
then we already have five vertex disjoint paths given by @xmath364 \to v_\pm \to v , \quad [ 1,0 ] \to v_2 \to w_2 \to [ i,0 ] \to w,\ ] ] and @xmath364 \to v_1 \to w_1 \stackrel{q_1}{\longrightarrow } w , \quad [ 1,0 ] \to v_3 \to w_3 \stackrel{q_3}{\longrightarrow } w.\ ] ] notice that for any wheel @xmath384 with @xmath385 , we have not yet used any edges with one vertex in @xmath347 .
we will see that every such wheel allows us to create two more vertex disjoint paths from @xmath298 $ ] to @xmath373 , finishing this case .
let @xmath336 be the two different vertices in wheel @xmath347 adjacent to @xmath373 .
choose two different vertices @xmath386 which have not been used yet and associate to them two different vertices @xmath387 such that @xmath388 and @xmath389 , using lemma [ lm : help](c ) .
then we can use lemma [ lm : help](a ) to complete the paths within @xmath390 .
ad ( iii ) : this is the easiest case . assume that the end vertex is @xmath319 \in w_i$ ] with @xmath365 .
we use the bijection @xmath391 in lemma [ lm : help](c ) to create the @xmath60 vertex disjoint paths @xmath364 \to [ 0,\mu ] \to \phi([0,\mu ] ) \to [ i,0]\ ] ] with @xmath392 .
case @xmath398 : obviously , we can choose two vertex disjoint paths within @xmath323 to connect @xmath130 and @xmath373 .
next , we show that every wheel @xmath347 with @xmath399 gives rise to two additional vertex disjoint paths .
let @xmath387 be the two distinct neighbours of @xmath130 , and @xmath400 be the two distinct neighbours of @xmath373 .
then we can use lemma [ lm : help](a ) to complete the paths within @xmath347 .
case @xmath348 : let @xmath401 be the neighbours of @xmath130 and @xmath402 be the neighbours of @xmath373 .
then , using only additional edges in @xmath403 , we can find four vertex disjoint paths @xmath404 , @xmath405 ( for @xmath283 ) .
again , every wheel @xmath406 with @xmath407 will give rise to two more vertex disjoint paths .
let @xmath408 be the neighbours of @xmath130 , and @xmath409 be the neighbours of @xmath373 .
use lemma [ lm : help](c ) to complete the paths within @xmath410 .
let us first confirm that any two different vertices in the same wheel can be connected by a path of length @xmath3 : let @xmath320 and @xmath353,[\nu , i^{-1 } ] \in \partial w_i$ ] ( thinking of @xmath412 ) be the two vertices .
the required path is then given by @xmath413 \to [ 2(\mu-\nu)^{-1}\mu i -i,2(\mu-\nu)^{-1 } ] \to [ \nu , i^{-1}].\ ] ] now choose two vertices @xmath351 and @xmath414 on different wheels .
let @xmath415 be one of the two neighbours of @xmath373 in the @xmath107-th wheel .
connecting @xmath130 and @xmath416 by a path of length 2 ( as shown before ) implies that @xmath417 .
note that the graph @xmath62 was initially defined as the induced subgraph of @xmath59 with vertex sets @xmath418 .
alternatively , @xmath62 can also described as the cayley graph @xmath419 with @xmath420 the vertices @xmath50 \in v(\pi_p')$ ] ( with non - vanishing second coordinate @xmath421 ) are then identified with the matrices @xmath422 . as before , we assume that @xmath60 is a fixed odd prime and @xmath423 .
the considerations of the previous section show also that @xmath59 is a @xmath318-fold covering @xmath424 of the complete graph @xmath425 , where the preimages @xmath426 correspond to the axes of @xmath59 .
it is useful to think of the vertices in @xmath425 as the points in the finite projective line over the field @xmath427 , i.e. , @xmath428 and the covering map is then given , algebraically , by @xmath429 ) = \lambda\mu^{-1},\ ] ] with the usual convention @xmath430 and @xmath431 .
in particular , we have @xmath432 .
note that @xmath214 acts also on the vertices of @xmath425 via @xmath433 one easily checks that @xmath434 for all @xmath435 and @xmath436 .
let us now explicitely derive the spectra of the graphs @xmath59 and @xmath62 .
we will use the following notation : for a linear operator @xmath269 on a finite dimensional vector space , we denote the eigenspace of @xmath269 to the eigenvalue @xmath437 by @xmath438 .
every eigenfunction @xmath145 of @xmath425 gives rise to an eigenfunction @xmath439 of the same eigenvalue via @xmath440 .
the spectrum of the adjacency operator on @xmath425 is given by ( see , e.g. , @xcite ) @xmath441 this implies that @xmath442 contains the eigenvalue @xmath60 with multiplicity one and the eigenvalue @xmath78 with multiplicity @xmath443 .
our next aim is to prove that the eigenspace @xmath444 of the square of the adjacency operator on @xmath59 has dimension @xmath445 .
let @xmath446 be a function satisfying @xmath447 note that can be viewed as a homogenous system of @xmath307 linear equations .
the key observation is that all linear equations corresponding to vertices of the same axis coincide , i.e. , we end up with only @xmath306 linear independent homogeneous equations ( since @xmath306 equals the number of axes ) , showing that the eigenspace has dimension at least @xmath448 indeed , since @xmath326 acts transitively on the vertices , we only need to show that the linear equations of corresponding to the vertices in the principal axis @xmath362 coincide .
recall that @xmath449 has the wheel - structure given in figure [ fig : wheels ] .
let @xmath450 .
then we have @xmath451 since there are exactly @xmath60 paths of length 2 from @xmath130 to itself , no paths of length 2 from the centers of all the other wheels to @xmath130 , and for every @xmath452 there are exactly @xmath3 paths from @xmath373 to @xmath130 of length 2 , because of lemma [ lm : help](b ) .
note that the combination of and simplifies to @xmath453 independently of the choice of @xmath450 .
this shows that @xmath454 . adding up the multiplicities of all eigenvalues
, we see that @xmath455 . if @xmath456 span the space @xmath444 , then the @xmath457 functions @xmath458 are eigenfunctions of @xmath140 to the eigenvalues @xmath81 , and they also span @xmath444 .
this shows that we have @xmath459 finally , the equality @xmath460 follows from lemma [ lm : eqdim ] below . the proof is based on the fact that @xmath462 is irrational .
let @xmath463 $ ] be the characteristic polynomial of @xmath269 .
we split @xmath464 into its even and odd part , i.e. , @xmath465 with even polynomials @xmath466 .
note that we have @xmath467 and @xmath468 .
therefore , if @xmath462 is a root of @xmath464 , then @xmath462 is also a root of both polynomials @xmath469 and @xmath470 , separately .
this implies that @xmath471 is also a root of @xmath464 .
we can then split off the factor @xmath472 from @xmath464 , and repeat the procedure with the remaining polynomial .
next we derive the spectrum of the modified graph @xmath62 , using the @xmath318-fold covering map @xmath473 and the wheel - structure , which partitions the vertex set @xmath474 into @xmath318 disjoint sets @xmath323 of @xmath60 vertices , each .
this will finish the proof of theorem [ thm : platonic ] .
\(i ) let @xmath475 be the vector space of all functions which are constant on the wheels .
we first introduce a basis of eigenfunctions of this vector space .
let @xmath476 and , for @xmath477 , define @xmath478 note that @xmath479 is the constant function to the eigenvalue @xmath77 .
it is easily checked that @xmath480 for @xmath481 .
since these functions are linearly independent , they form a basis of @xmath482 .
moreover , we have @xmath483 .
\(ii ) let @xmath484 be the vector space of all functions which are constant along all axes .
every such function is a lift @xmath440 of a function @xmath145 on @xmath485 .
note that eigenfunctions of @xmath485 are lifted to eigenfunctions to the same eigenvalue , so @xmath484 can be viewed as the span of a constant function and @xmath77 linear independent eigenfunctions to the eigenvalue @xmath78 . in particular , we have @xmath486 .
let @xmath490 be a basis of this orthogonal complement by eigenfunctions with @xmath491 .
we now extend each @xmath492 trivially to a function @xmath493 on @xmath449 by setting @xmath494 for all @xmath450 .
note that these extensions are eigenfunctions of the platonic graph @xmath59 to the same eigenvalue , i.e. , @xmath495 .
therefore , we must have @xmath496 . as discussed in the previous proof , the span of the eigenfunctions of @xmath59 to the eigenvalues @xmath78 and @xmath306
is obtained via lifting the eigenfunctions of @xmath425 , and the restriction of these functions to @xmath62 must therefore lie in @xmath497 .
this shows that we must have @xmath498 . | we introduce a family of trivalent expanders which tessellate compact hyperbolic surfaces with large isometry groups .
we compare this family with platonic graphs and modifications of them and prove topological and spectral properties of these families . |
an @xmath0-curve on the boundary of a solid torus refers to the curve @xmath1+r[\mu]$ ] , where @xmath2 is the longitude - meridian basis for the homology of the torus , and we denote this by the fraction @xmath3 the @xmath0-cable of a knot type @xmath4 , denoted @xmath5 , is the knot type obtained by taking the @xmath0-curve on the boundary of a tubular neighborhood of a representative of @xmath6 .
let @xmath4 be a positive @xmath7-torus knot , where we may assume @xmath8 and @xmath9 and let @xmath5 be its @xmath0-cable , also with @xmath10
. this paper concerns the classification of legendrian and transverse knots representing @xmath5 and solid tori representing @xmath4 . though the proofs of our classification results are heavily dependent on the ambient contact manifold being @xmath11 , all the legendrian and transversal classification results hold in any tight contact manifold , as can be seen by consulting @xcite . studying legendrian and transverse knots in cabled knot types has been very fruitful .
for example , in @xcite cabling was used to better understand open book decompositions of contact structures ; in particular , leading to non - positive monodromy maps supporting stein fillable contact structures , monoids in the mapping class group associated to contact geometry and procedures to construct open books on manifolds after allowable transverse surgery ( from an open book for the original contact manifold ) . moreover , the first classification of a non - transversely simple knot type was done in @xcite for the @xmath12-cable of the @xmath12-torus knot . in that paper
it was also shown that studying solid tori with convex boundary that represent a given knot type ( that is , their core curves are in a given knot type ) is key to understanding cables ; such an analysis for solid tori representing negative torus knots yielded simple legendrian and transverse classifications for cables of negative torus knots .
tori representing iterated cables of torus knots were further studied in @xcite as well as @xcite .
building on these works we completely classify embeddings of solid tori representing positive torus knots and use this to give a complete classification of legendrian and transverse knots in the knot types of cables of positive torus knots . before discussing the technical classification results we state qualitative versions that demonstrate new phenomena in the geography of legendrian knots .
we begin with some notation . given a topological knot type @xmath13 and integers @xmath14 and @xmath15 we denote by @xmath16 the set of legendrian knots ( up to legendrian isotopy ) topologically isotopic to @xmath13 and by @xmath17 we similarly denote the set of transverse knots isotopic to @xmath13 by @xmath18 and the ones having self - linking number @xmath19 by @xmath20 .
we first consider cables of the right handed trefoil , that is , the @xmath12-torus knot .
[ qual1 ] let @xmath13 be the positive trefoil knot in @xmath21 the knot @xmath22 formed by @xmath0-cabling @xmath13 is legendrian simple if and only if @xmath23 furthermore , given positive integers @xmath24 , @xmath25 , and @xmath26 , where @xmath27 and @xmath28 , there exists a slope @xmath29 such that @xmath30 contains @xmath26 legendrian knots for some pair of integers @xmath31 with @xmath32 ; moreover , one of these does not destabilize , and they remain distinct when stabilized fewer than @xmath24 times ( and there are @xmath24 stabilizations that will make them isotopic ) .
this theorem gives the first example of a knot type with non - destabilizable legendrian knots with thurston - bennequin invariant arbitrarily far from the maximal thurston - bennequin invariant .
we note that in @xcite it was shown there are knot types that have arbitrarily many legendrian knots with fixed classical invariants , so the above theorem gives only the second family of knots known to have this property .
we also observe that this theorem gives the first set of legendrian knots with the same invariants that requires arbitrarily many stabilizations before becoming legendrian isotopic .
[ qual2 ] let @xmath13 be the positive trefoil knot in @xmath21 the knot @xmath22 formed by @xmath0-cabling @xmath13 is transversely simple if and only if @xmath23 furthermore , given positive integers @xmath24 , @xmath25 , and @xmath26 , where @xmath33 and @xmath28 , let @xmath34 .
then there is some @xmath35 such that @xmath36 contains @xmath37 distinct transverse knots with @xmath38 , of which @xmath39 are non - destabilizable , and such that there is another non - destabilizable knot with @xmath40 .
moreover , these non - destabilizable knots must be stabilized until their self - linking number is @xmath41 before they become transversely isotopic . in @xcite
it was shown that there are knot types , specifically certain twist knots , that have arbitrarily many transverse knots with the same self - linking number .
the above theorem also gives such examples but , in addition , demonstrates three new phenomena concerning transverse knots that were not previously known .
specifically it gives the first example of knot types that have transverse knots with the same self - linking number that require arbitrarily many stabilizations before they become transversely isotopic , and it also gives the first examples where there are non - destabilizable transverse knots whose self - linking number is arbitrarily far from maximal . finally , the theorem also gives the first knot type where there are non - destabilizable knots with distinct self - linking numbers .
with all the interesting and complicated behavior exhibited by cables of the right handed trefoil knot , one would expect to see behavior at least as complicated for cables of other positive torus knots .
surprisingly , cables of such knots turn out to be relatively simple .
[ qual3 ] let @xmath13 be a positive @xmath7-torus knot with @xmath42 .
then for any rational number @xmath43 and any @xmath31 with @xmath44 odd , there are at most 3 legendrian knots in @xmath30 and at most 2 for all but one pair @xmath31 .
[ qual4 ] let @xmath13 be a positive @xmath7-torus knot with @xmath42 .
then for any rational number @xmath43 there are at most two transverse knots isotopic to the @xmath0-cable of @xmath13 with the same self - linking number . however , for any positive integers
@xmath26 and @xmath25 with @xmath45 , there is a rational number @xmath46 for which there is a non - destabilizable transverse knot with self - linking number at most @xmath47 and it must be stabilized exactly @xmath25 times to become isotopic to the destabilizable transverse knot with the same self - linking number . as indicated above the key to proving these classification results
is classifying solid tori with convex boundary realizing positive torus knots .
this classification , discussed below , is the first complete such classification and exhibits features not seen before , such as the existence of partially thickenable tori ( see subsection [ solid tori ] ) . in the next two subsections we state the precise classification theorems that lead to the above qualitative results .
in subsection [ knotclassifications ] we state knot classification theorems for cables ; in subsection [ solid tori ] we state classification theorems for embeddings of solid tori .
we begin with cables of the right handed trefoil knot .
[ main23 ] let @xmath13 be the @xmath12-torus knot .
then the @xmath0-cable of @xmath13 , @xmath22 , is legendrian simple if and only if @xmath48 , and the classification of legendrian knots in the knot type @xmath22 is given as follows . 1 .
if @xmath49 $ ] then there is a unique legendrian knot @xmath50 with thurston - bennequin invariant @xmath51 and rotation number @xmath52 all others are stabilizations of @xmath53 2 .
if @xmath54 , then the maximal thurston - bennequin invariant for a legendrian knot in @xmath55 is @xmath56 and the rotation numbers realized by legendrian knots with this thurston - bennequin invariant are + @xmath57 where @xmath26 is the integer that satisfies @xmath58 all other legendrian knots @xmath50 are stabilizations of these .
two legendrian knots with the same @xmath59 and @xmath60 are legendrian isotopic .
3 . suppose @xmath61 for a positive integer @xmath26 ; then @xmath22 is not legendrian simple and has the following classification ( see also figure [ fig : range ] ) .
1 . the maximal thurston - bennequin number is @xmath62 .
2 . there are @xmath26 legendrian knots @xmath63 , with @xmath64 3 . if @xmath65 then there are two legendrian knots @xmath66 that do not destabilize but have @xmath67 4 .
all legendrian knots in @xmath55 destabilize to one of the @xmath68 or @xmath69 .
[ item511a ] let @xmath70 . for
any @xmath71 and @xmath72 the legendrian @xmath73 is not isotopic to a stabilization of any of the other @xmath74 s the @xmath75 , @xmath69 or @xmath76 . 6 . [ item512a ] let @xmath77 . for
any @xmath78 and @xmath79 the legendrian @xmath80 is not isotopic to a stabilization of any of the @xmath74 s or @xmath76 .
any two stabilizations of the @xmath81 or @xmath69 , except those mentioned in item ( [ item511a ] ) and ( [ item512a ] ) , are legendrian isotopic if they have the same @xmath59 and @xmath60 .
the transverse classification is now an immediate corollary .
[ main23 t ] let @xmath13 be the @xmath12-torus knot . if @xmath82 then @xmath22 is transversely simple and all transverse knots are stabilizations of the one with maximal self - linking number @xmath83 . if @xmath84 for a positive integer @xmath26 then @xmath22 is not transversely simple and has the following classification . 1 . the maximal self - linking number is @xmath85 and there is a unique transverse knot in @xmath36 with this self - linking number .
2 . there are @xmath86 distinct transverse knots in @xmath36 that do not destabilize and have self - linking number @xmath87 3 . if @xmath88 then there is a unique transverse knot in @xmath36 that does not destabilize and has self - linking number @xmath89 .
all other transverse knots in @xmath36 destabilize to one of the ones listed above .
none of the transverse knots listed above become transversely isotopic until they have been stabilized to have self - linking number @xmath90 there is a unique transverse knot in @xmath36 with self - linking number less than or equal to @xmath90 for the classification of cables of other positive torus knots we need some notation .
given a rational number @xmath91 let @xmath92 be the largest rational number with an edge in the farey tessellation to @xmath93 see figure [ fig : nonsimple ] .
( the @xmath94 superscript stands for `` anti - clockwise '' , as @xmath92 is anti - clockwise of @xmath95 in the farey tessellation . ) similarly the smallest rational number with an edge in the farey tessellation to @xmath95 will be denoted by @xmath96 a formula for computing these numbers will be given in subsection [ sec : continuedfrac ] .
we will refer to the interval @xmath97 as the _ interval of influence _ for @xmath95 .
given a positive @xmath7-torus knot and @xmath24 a positive integer , define @xmath98 we will see in subsection [ solid tori ] that such @xmath99 represent boundary slopes of _ non - thickenable _ solid tori , and that the half - intervals of influence @xmath100 will represent boundary slopes of _ partially thickenable _ solid tori when @xmath101 .
we will refer to the @xmath99 as _ exceptional slopes_. if we think of the fractions @xmath102 as representing curves on a torus , we denote the homological intersection of @xmath0 curves with the @xmath102 curves by @xmath103 we can now state the precise classification theorems for cables of general positive @xmath7-torus knots .
[ main ] let @xmath13 be a @xmath7-torus knot with @xmath42 .
let @xmath104 and @xmath105 where @xmath106 is the interval of influence for the exceptional slope @xmath107 defined above .
the @xmath108 are all disjoint .
the classification of legendrian knots in the knot type @xmath22 is then given as follows . 1 .
if @xmath109 then @xmath22 is legendrian simple . moreover , in this case we have the following classification . 1 .
if @xmath110 $ ] then there is a unique legendrian knot @xmath50 with thurston - bennequin invariant @xmath111 and rotation number @xmath52 all others are stabilizations of @xmath53 2 . if @xmath112 or @xmath54 , then the maximal thurston - bennequin invariant for a legendrian knot in @xmath55 is @xmath56 and the rotation numbers realized by legendrian knots with this thurston - bennequin invariant are @xmath113 where @xmath26 is the least integer bigger than @xmath114 .
all other legendrian knots @xmath50 are stabilizations of these . two legendrian knots with the same @xmath59 and @xmath60 are legendrian isotopic .
2 . if @xmath115 then there is some @xmath116 such that @xmath117 and @xmath22 is not legendrian simple .
the classification of legendrian knots in @xmath22 is as follows . 1 .
the maximal thurston - bennequin invariant of @xmath118 is @xmath56 .
2 . for each integer
@xmath119 in the set @xmath120 there is a legendrian @xmath121 with @xmath122 3 .
there are two legendrian knots @xmath123 satisfying @xmath124 if @xmath125 ; however , if @xmath126 then @xmath127 and @xmath69 is not destabilizable .
all legendrian knots in @xmath55 destabilize to one of the @xmath128 or @xmath129 5 .
] let @xmath130 for any @xmath78 and @xmath72 the legendrian @xmath80 is not isotopic to a stabilization of any of the @xmath128 or @xmath76 .
any two stabilizations of the non - destabilizable thurston - bennequin invariant legendrian knots in @xmath55 , except those mentioned in item ( [ item ? ?
] ) , are legendrian isotopic if they have the same @xmath59 and @xmath60 . from this theorem
we can easily derive the transverse classification .
[ maint ] let @xmath13 be a @xmath7-torus knot with @xmath42 . using notation from theorem [ main ] we have the following classification of transverse knots in @xmath36 . 1 .
if @xmath131 for any @xmath132 then @xmath22 is transversely simple and all transverse knots in this knot type are stabilizations of the one with self - linking number @xmath133 .
2 . if @xmath115 for some @xmath132 then @xmath22 is not transversely simple .
there is a unique transverse knot @xmath134 in this knot type with maximal self - linking number , which is @xmath133 .
there is also a unique non - destabilizable knot @xmath135 in this knot type and it has self - linking number @xmath136 .
all other transverse knots in @xmath36 destabilize to either @xmath134 or @xmath135 and the stabilizations of @xmath134 and @xmath135 stay non - isotopic until they are stabilized to the point that their self - linking numbers are @xmath137 in the case of @xmath138 , and @xmath139 in the case of @xmath140 .
we now turn from classification results for cables of positive torus knots , to classification results for embeddings of solid tori representing the positive torus knots themselves .
let @xmath141 be a solid torus in a manifold @xmath142 we say @xmath141 @xmath13 , or @xmath13 , if the core curve of @xmath141 is in the knot type @xmath143 we say a solid torus @xmath141 with convex boundary in a contact manifold @xmath144 if there is a solid torus @xmath145 that contains @xmath141 such that @xmath145 has convex boundary with dividing slope different from @xmath146 the existence of non - thickenable tori was first observed in @xcite ; the following theorem shows that non - thickenable tori exist for all positive @xmath7-torus knots .
[ thm : nonthick ] let @xmath141 be a solid torus in the knot type of a positive @xmath7-torus knot . in the standard tight contact structure @xmath147 on @xmath148
suppose that @xmath149 is convex with two dividing curves of slope @xmath3 then @xmath141 thickens unless @xmath43 is an exceptional slope @xmath150 for some positive integer @xmath151 in which case it might or might not thicken .
moreover for each positive integer @xmath152 there are , up to contact isotopy , exactly two solid tori @xmath153 with convex boundary having @xmath154 dividing curves of slope @xmath99 that do not thicken , where @xmath155 .
for @xmath156 there is exactly one solid torus @xmath157 with convex boundary having two dividing curves of slope @xmath158 this solid torus is a standard neighborhood of a legendrian @xmath7-torus knots with maximal thurston - bennequin invariant and it does not thicken .
a key feature in the knot classification results above in subsection [ knotclassifications ] is a complete understanding of not only non - thickenable tori but also tori , that is tori with convex boundary that thicken , but not to a maximally thick torus in the given knot type .
the existence of such tori has not been observed before , but it is clear that such tori will be key to future legendrian classification results .
in addition it is likely they will be important in understanding contact surgeries .
the following theorem shows that partially thickenable tori exist for all positive @xmath7-torus knots .
[ thm : partially ] let @xmath13 be a positive @xmath7-torus knot and let @xmath159 be the exceptional slopes .
let @xmath160 and @xmath161 .
all solid tori below will represent the knot type @xmath13 . 1 .
if @xmath162 then @xmath163 and we have the following . 1 .
the intervals @xmath164 so @xmath165 2 . any solid torus @xmath141 with convex boundary thickens to @xmath153 or to @xmath157 ( that is a neighborhood of the maximal thurston - bennequin invariant @xmath12-torus knot ) .
any solid torus inside @xmath153 with convex boundary having dividing slope greater than @xmath24 ( that is in @xmath166 ) does not thicken past the slope @xmath167 4 .
any solid torus inside @xmath153 with convex boundary having negative ( or infinite ) dividing slope will thicken to a neighborhood of the maximal thurston - bennequin invariant @xmath12-torus knot .
if @xmath42 then we have the following . 1 . for any @xmath168 , any solid
torus @xmath141 inside @xmath153 with either boundary slope different from @xmath99 , or less than @xmath154 dividing curves , thickens to @xmath157 .
all the @xmath166 with @xmath169 are disjoint .
3 . any solid torus @xmath141 with convex boundary having dividing slope in @xmath166 thickens to @xmath153 or to @xmath157 ( that is a neighborhood of the maximal thurston - bennequin invariant @xmath7-torus knot ) .
any solid torus inside @xmath153 for some @xmath169 , and with convex boundary having dividing slope in @xmath166 , does not thicken past the slope @xmath167 5 .
any solid torus inside @xmath153 with convex boundary having dividing slope outside of @xmath166 ( that is greater than or equal to @xmath170 or negative ) will thicken to a neighborhood of the maximal thurston - bennequin invariant @xmath7-torus knot .
from this theorem we can classify solid tori in the knot types of positive torus knots .
[ cor : tori ] let @xmath13 be a positive @xmath7-torus knot and let @xmath159 be the exceptional slopes .
let @xmath160 and @xmath161 . 1 .
if @xmath162 , then given a slope @xmath171 there is some integer @xmath26 such that @xmath172 and there are exactly @xmath173 solid tori representing the knot type @xmath13 with convex boundary having dividing slope @xmath19 and two dividing curves , only two of which thicken to a standard neighborhood of a legendrian knot .
2 . if @xmath42 , then given any slope @xmath174 we have the following . 1 . if there is some integer @xmath175 such that @xmath176 and @xmath177 for any @xmath169 , then there are exactly @xmath178 solid tori representing the knot type @xmath13 with convex boundary having dividing slope @xmath19 and two dividing curves each of which thickens to a standard neighborhood of a legendrian knot with @xmath179 .
2 . if there is some integer @xmath175 such that @xmath176 and @xmath180 for any @xmath169 , then there are exactly @xmath181 solid tori representing the knot type @xmath13 with convex boundary having dividing slope @xmath19 and two dividing curves , all but two of which thicken to a standard neighborhood of a legendrian knot with @xmath179 .
3 . if there is some @xmath175 such that @xmath182 , then there are exactly @xmath183 solid tori representing the knot type @xmath13 with convex boundary having dividing slope @xmath19 and two dividing curves and they each represent a standard neighborhood of a legendrian knot with @xmath179 .
3 . given any negative slope @xmath19 there is some negative integer @xmath184 such that @xmath185 . a solid torus with convex boundary having dividing slope @xmath19 and two dividing curves will thicken to a solid torus that is a standard neighborhood of a @xmath186 legendrian knot .
we conclude this introduction with an outline of what follows . in section [ sec : pre ] we collect needed preliminaries , including facts about continued fractions and convex surfaces , and we outline a strategy for classifying legendrian knots . in section [ classification of solid tori ]
we classify embeddings of solid tori representing positive torus knots . in section [ simple cables ]
we provide classifications for all simple cables of positive torus knots , and in sections [ non - simple cables non - trefoil ] and [ non - simple cables trefoil ] we establish classifications for all non - simple cables of positive torus knots .
* acknowledgments . *
the first and third authors were partially supported by nsf grant dms-0804820 .
the second author was partially supported by qgm ( centre for quantum geometry of moduli spaces ) funded by the danish national research foundation .
some of the work presented in this paper was carried out in the spring of 2010 while the first and third author were at msri , we gratefully acknowledge their support for this work .
in this section we first prove some important facts about continued fractions in subsection [ sec : continuedfrac ] .
the remaining sections recall various facts concerning the classification of legendrian and transverse knots from @xcite .
the reader is assumed to be familiar with the basic notions associated to convex surfaces and legendrian and transverse knots , but these sections are included for the convenience of the reader and to make the paper as self - contained as possible .
all this information can be found in @xcite . in this section
we collect various facts about continued fractions and the farey tessellation ( see figure [ fig : ft ] ) that will be needed throughout our work . given a rational number @xmath187 we may represent it as a continued fraction @xmath188 with @xmath189 and the other @xmath190 . we will denote this as @xmath191 $ ] .
if we know that @xmath192 $ ] then we define @xmath193,\ ] ] with the convention that if @xmath194 then @xmath195 ; we also define @xmath196.\ ] ] [ lem : eaec ] the number @xmath92 is the largest rational number bigger than @xmath95 with an edge to @xmath95 in the farey tessellation and @xmath197 is the smallest rational number less than @xmath95 with an edge to @xmath95 in the farey tessellation .
moreover there is an edge in the farey tessellation between @xmath92 and @xmath197 and @xmath95 is the mediant of @xmath92 and @xmath197 , that is if @xmath198 and @xmath199 then @xmath200 define @xmath201 $ ] and @xmath202 one may easily verify using induction that @xmath203 from this one can inductively deduce that @xmath204 thus there is an edge in the farey tessellation between @xmath205 and @xmath206 similarly , let @xmath207 $ ] and @xmath208 $ ] and notice that @xmath209 now we see that @xmath210 and a similar expression for @xmath211 and induction yield @xmath212 in particular , there is an edge in the farey tessellation between @xmath213 and @xmath214 finally by setting @xmath215 $ ] and noticing that @xmath216 , we can use the above formulas , and analogous ones , to inductively prove that @xmath217 .
this establishes an edge in the farey tessellation between @xmath218 and @xmath219 .
since there is an edge in the farey tessellation between each pair of numbers in the set @xmath220 the lemma is established by noticing that the numerators ( and denominators ) of @xmath92 and @xmath197 are both smaller than the numerator ( and denominator ) of @xmath93 we recall that if we choose a basis for @xmath221 then there is a one - to - one correspondence between embedded essential oriented curves on @xmath222 and rational numbers @xmath223 , written in lowest common terms . moreover given two rational numbers
@xmath223 and @xmath114 we denote their homological intersection ( which also happens to be the signed minimal intersection number ) between the corresponding curves on @xmath222 by @xmath224 and it can be computed by @xmath225 notice that this number is only well defined up to sign ( since the orientation on the curve corresponding to a fraction is not determined ) . throughout this work
we will only be concerned with the absolute value of this number ( if the exact number is ever needed we will specify the orientations on the homology class corresponding to a fraction ) .
[ lem : basicrange ] fix some positive integer @xmath26 and set @xmath226 for @xmath227 and @xmath228 .
if @xmath229 then the intervals @xmath230 for @xmath169 are all disjoint . if @xmath231 then the intervals are nested @xmath232
. if @xmath15 is a positive rational number less than @xmath233 or greater than @xmath170 then for any @xmath234 we have @xmath235 with equality only if @xmath236 or @xmath233 . if @xmath237 and @xmath238 then @xmath239 if @xmath231 then it is clear that @xmath240 and one easily checks that @xmath241 and @xmath242 so @xmath243 .
if @xmath229 then we notice that any number in @xmath244 is a mediant of @xmath99 and @xmath233 and hence has denominator strictly bigger than @xmath26 ( since the denominator of @xmath99 is @xmath26 ) , thus @xmath245 can not be in this interval for any @xmath246
. similarly @xmath245 can not be in the interval @xmath247 .
thus the intervals @xmath248 are disjoint . for the second statement notice that @xmath249 and @xmath250 have the same sign and @xmath251 will be some non - negative integral linear combination of @xmath249 and @xmath250 . for the last statement note that @xmath251 will be some positive integral linear combination of @xmath252 and @xmath249 .
in this subsection we discuss the main tools we will be using throughout the paper convex surfaces .
we assume the reader is familiar with convex surfaces as used in @xcite ; but , for the convenience of the reader , we recall the fundamental facts from the theory that we will use in this paper .
recall a surface @xmath253 in a contact manifold @xmath144 is if it has a neighborhood @xmath254 , where @xmath255 is some interval , and @xmath256 is @xmath257-invariant in this neighborhood .
any closed surface can be @xmath258-perturbed to be convex .
moreover if @xmath259 is a legendrian knot on @xmath253 for which the contact framing is non - positive with respect to the framing given by @xmath253 , then @xmath253 may be perturbed in a @xmath260 fashion near @xmath259 , but fixing @xmath259 , and then again in a @xmath258 fashion away from @xmath259 so that @xmath253 is convex . given a convex surface @xmath253 with @xmath257-invariant neighborhood let @xmath261 be the multicurve where @xmath256 is tangent to the @xmath257 factor .
this is called the of @xmath262 if @xmath253 is oriented it is easy to see that @xmath263 where @xmath256 is positively transverse to the @xmath257 factor along @xmath264 and negatively transverse along @xmath265 .
if @xmath259 is a legendrian curve on a @xmath253 then the framing of @xmath259 given by the contact planes , relative to the framing coming from @xmath253 , is given by @xmath266 .
moreover if @xmath267 then the rotation number of @xmath259 is given by @xmath268 .
a convex torus @xmath134 is said to be in if @xmath134 can be identified with @xmath269 so that @xmath270 consists of @xmath173 vertical curves ( note @xmath270 will always have an even number of curves and we can choose a parameterization to make them vertical ) and the characteristic foliations consists of @xmath173 vertical lines of singularities ( @xmath26 lines of sources and @xmath26 lines of sinks ) and the rest of the foliation is by non - singular lines of slope @xmath19 .
see figure [ fig : ct ] .
the lines of singularities are called and the other curves are called .
we notice that the giroux flexibility theorem allows us to isotope any convex torus into standard form , @xcite .
let @xmath253 be a convex surface and @xmath271 a legendrian arc in @xmath253 that intersects the dividing curves @xmath272 in 3 points @xmath273 ( where @xmath274 are the end points of the arc ) .
then a , see figure [ fig : ct ] , is a convex disk @xmath275 with legendrian boundary such that 1 .
@xmath276 2 .
@xmath277 3 .
@xmath278 4 .
@xmath279 are corners of @xmath275 and elliptic singularities of @xmath280 a surface @xmath253 locally separates the ambient manifold . if a bypass is contained in the ( local ) piece of @xmath281 that has @xmath253 as its oriented boundary then we say the bypass will be attached to the front of @xmath253 otherwise we say it is attached to the back of @xmath253 .
when a bypass is attached to a torus @xmath134 then either the dividing curves do not change , their number increases by two , or decreases by two , or the slope of the dividing curves changes .
the slope of the dividing curves can change only when there are two dividing curves .
( see @xcite for more details . )
if the bypass is attached to @xmath134 along a ruling curve then either the number of dividing curves decreases by two or the slope of the dividing curves changes . to understand the change in slope
we need the following .
let @xmath282 be the unit disk in @xmath283 recall the of @xmath282 is constructed as follows .
label the point @xmath284 on @xmath285 by @xmath286 and the point @xmath287 with @xmath288 now join them by a geodesic .
if two points @xmath289 @xmath290 on @xmath285 with non - negative @xmath291-coordinate have been labeled then label the point on @xmath292 half way between them ( with non - negative @xmath291-coordinate ) by @xmath293 then connect this point to @xmath294 by a geodesic and to @xmath290 by a geodesic .
continue this until all positive fractions have been assigned to points on @xmath285 with non - negative @xmath291-coordinates . now repeat this process for the points on @xmath295 with non - positive @xmath291-coordinate except start with @xmath296 see figure [ fig : ft ] .
the key result we need to know about the farey tessellation is given in the following theorem .
[ bpaferry ] let @xmath134 be a convex torus in standard form with @xmath297 dividing slope @xmath19 and ruling slope @xmath298 let @xmath275 be a bypass for @xmath134 attached to the front of @xmath134 along a ruling curve . let @xmath135 be the torus obtained from @xmath134 by attaching the bypass @xmath299 then @xmath300 and the dividing slope @xmath301 of @xmath302 is determined as follows : let @xmath303 $ ] be the arc on @xmath292 running from @xmath15 counterclockwise to @xmath304 then @xmath301 is the point in @xmath303 $ ] closest to @xmath15 with an edge to @xmath305 if the bypass is attached to the back of @xmath134 then the same algorithm works except one uses the interval @xmath306 $ ] on @xmath292 . as we see that bypasses are useful in changing dividing curves on a surface we mention a standard way to try to find them called the imbalance principle .
suppose that @xmath253 and @xmath307 are two disjoint convex surfaces and @xmath308 is a convex annulus whose interior is disjoint from @xmath253 and @xmath307 but its boundary is legendrian with one component on each surface .
if @xmath309 then there will be a dividing curve on @xmath308 that cuts a disk off of @xmath308 that has part of its boundary on @xmath253 .
it is now easy to use the giroux flexibility theorem to show that there is a bypass for @xmath310 on @xmath308 .
we will frequently need to analyze what happens to the contact geometry when we have a topological isotopy between two convex surfaces @xmath253 and @xmath307 .
this can be done by the technique of @xcite ( see also @xcite for its use in studying legendrian knots ) .
given an isotopy between @xmath253 and @xmath307 one can find a sequence of convex surfaces @xmath311 such that 1 .
all the @xmath312 are convex and 2 .
@xmath312 and @xmath313 are disjoint and @xmath313 is obtained from @xmath312 by a bypass attachment .
thus if one is trying to understand how the contact geometry of @xmath281 and @xmath314 relate , one just needs to analyze how the contact geometry of the pieces of @xmath315 changes under bypass attachment .
in particular , many arguments can be reduced from understanding a general isotopy to understanding an isotopy between two surfaces that cobound a product region .
there is also a relative version of isotopy discretization where @xmath253 and @xmath307 are convex surfaces with legendrian boundary consisting of ruling curves on a convex torus .
if @xmath316 and there is a topological isotopy of @xmath253 to @xmath307 relative to the boundary then we can find a discrete isotopy as described above . in this section
we briefly recall the standard strategy for classifying legendrian knots in a given knot type @xmath13 as laid out in @xcite .
we begin by recalling the `` normal form '' for a neighborhood of a legendrian or transverse knot and the relation between them . given a legendrian knot @xmath259 , a of @xmath259 is a solid torus @xmath317 that has convex boundary with two dividing curves of slope @xmath318 ( and of course we will usually take @xmath319 to be a convex torus in standard form ) .
conversely given any such solid torus it is a standard neighborhood of a unique legendrian knot .
up to contactomorphism one can model a standard neighborhood as a neighborhood @xmath320 of the @xmath321-axis in @xmath322 with contact structure @xmath323 using this model we can see that @xmath324 is a @xmath325-transverse curve .
the image of @xmath326 in @xmath317 is called the of @xmath259 and @xmath327 is called the .
one may easily check that @xmath328 is well - defined and compute that @xmath329 one may understand stabilizations and destabilizations of a legendrian knot @xmath330 in terms of the standard neighborhood . specifically , inside the standard neighborhood @xmath317 of @xmath259 , @xmath259 can be positively stabilized to @xmath331 , or negatively stabilized to @xmath332 .
let @xmath333 be a neighborhood of the stabilization of @xmath259 inside @xmath334 as above we can assume that @xmath333 has convex boundary in standard form .
it will have dividing slope @xmath335 thus the region @xmath336 is diffeomorphic to @xmath337 $ ] and the contact structure on it is easily seen to be a , see @xcite .
there are exactly two basic slices with given dividing curves on their boundary and as there are two types of stabilization of @xmath259 we see that the basic slice @xmath338 is determined by the type of stabilization done , and vice versa .
moreover if @xmath317 is a standard neighborhood of @xmath259 then @xmath259 destabilizes if the solid torus @xmath317 can be thickened to a solid torus @xmath339 with convex boundary in standard form with dividing slope @xmath340 moreover the sign of the destabilization will be determined by the basic slice @xmath341 .
finally , we notice that using theorem [ bpaferry ] we can destabilize @xmath259 by finding a bypass for @xmath317 attached along a ruling curve whose slope is clockwise of @xmath342 ( and anti - clockwise of @xmath343 ) .
a neighborhood of a transverse knot @xmath134 can be modeled by the solid torus @xmath344 for sufficiently small @xmath94 , where @xmath345 are polar coordinates on @xmath346 and @xmath347 is the angular coordinate on @xmath348 , with the contact structure @xmath349 .
notice that the tori @xmath350 inside of @xmath351 have linear characteristic foliations of slope @xmath352 thus for all integers @xmath26 with @xmath353 we have tori @xmath354 with linear characteristic foliation of slope @xmath355 let @xmath356 be a leaf of the characteristic foliation of @xmath357 any legendrian @xmath259 legendrian isotopic to one of the @xmath356 so constructed will be called a of @xmath358 [ legapprox ] if @xmath356 is a legendrian approximation of the transverse knot @xmath134 then @xmath359 is transversely isotopic to @xmath358 moreover , @xmath360 is legendrian isotopic to the negative stabilization of @xmath361 this lemma is a key ingredient in the following result from which our transverse classification results will follow from our legendrian classification results .
[ tequivaletsl ] the classification of transverse knots up to transverse isotopy is equivalent to the classification of legendrian knots up to negative stabilization and legendrian isotopy
. the classification of legendrian knots in a given knot type can be done in a ( roughly ) three step process .
* step i * _ identify the maximal thurston - bennequin invariant of @xmath13 and classify legendrian knots realizing this .
_ * step ii * _ identify and classify the non - maximal thurston - bennequin legendrian knots in @xmath13 that do not destabilize and prove that all other knots destabilize to one of these identified knots . _
* step iii * _ determine which stabilizations of the maximal thurston - bennequin invariant knots and non - destabilizable knots are legendrian isotopic .
_ as stabilization of a legendrian knot is well defined and positive and negative stabilizations commute , it is clear that these steps will yield a classification of legendrian knots in the knot type @xmath13 .
step ii is facilitated by the observation above that bypasses attached to appropriate ruling curves of a standard neighborhood of a legendrian knot yield destabilizations .
similarly , if @xmath259 is a legendrian knot contained in a convex surface @xmath253 ( and the framing given to @xmath259 by @xmath253 is less than or equal to the framing given by a seifert surface ) and there is a bypass for @xmath259 on @xmath253 then this leads to a destabilization of @xmath259 . moreover one can find such a bypass in some cases by the imbalance principle discussed above .
we begin by recalling a result of eliashberg concerning the contactomorphism group of the standard contact structure @xmath147 on @xmath148 .
fix a point @xmath362 in @xmath148 and let @xmath363 be the group of orientation - preserving diffeomorphisms of @xmath148 that fix the plane @xmath364 and let @xmath365 be the group of diffeomorphisms of @xmath148 that preserve @xmath147 . [ thm : htpyequiv ] the natural inclusion of @xmath366 is a weak homotopy equivalence . using this fact it is clear that if one has a contactomorphism @xmath347 of @xmath11 that takes a set @xmath367 to @xmath368 , then there is a contact isotopy of @xmath11 that takes @xmath141 to @xmath145 . in particular ,
if one is trying to show that two embeddings of a contact structure on a torus are contact isotopic then one merely needs to construct a contactomorphism that takes one torus to the other . similarly to show two legendrian knots are legendrian isotopic one only needs to construct a contactomorphism that takes one knot to the other ( or takes a standard neighborhood of one of the knots to the other , that is understand the contactomorphism type of the complement of the standard neighborhood ) . in this subsection
we collect various facts that are useful in computing the classical invariants of legendrian knots on tori .
let @xmath134 be a convex torus in a contact manifold @xmath371 where @xmath256 has euler class 0 .
now we define an invariant of homology classes of curves on @xmath358 let @xmath372 be any globally non - zero section of @xmath256 and @xmath373 a section of @xmath374 that is transverse to and twists ( with @xmath256 ) along the legendrian ruling curves and is tangent to the legendrian divides . if @xmath375 is a closed oriented curve on @xmath134 then set @xmath376 equal to the rotation of @xmath372 relative @xmath373 along @xmath377 one may check the following properties ( cf .
@xcite ) . 1 .
the function @xmath378 is well - defined on homology classes . 2 .
the function @xmath378 is linear .
3 . the function @xmath378 is unchanged if we isotope @xmath134 through convex tori in standard form .
if @xmath375 is a @xmath0-ruling curve or legendrian divide then @xmath379 we recall two simple lemmas from @xcite .
the first concerns the computation of the thurston - bennequin invariant for cables .
[ tbcomp ] let @xmath13 be a knot type and @xmath317 a solid torus representing @xmath13 whose boundary is a standard convex torus .
suppose that @xmath380 is contained in @xmath319 . 1 .
suppose @xmath381 is a legendrian divide and @xmath382
. then @xmath383 2 .
suppose @xmath381 is a legendrian ruling curve and @xmath384 . then @xmath385 a simple consequence of the discussion in subsection [ sec : rotationfunction ] yields the following computation of the rotation number for cables . [ rcomp ]
let @xmath13 be a knot type and @xmath317 a solid torus representing @xmath13 whose boundary is a standard convex torus .
suppose that @xmath380 is contained in @xmath319 .
then @xmath386 where @xmath275 is a convex meridional disk of @xmath317 with legendrian boundary on a contact - isotopic copy of the convex surface @xmath319 , and @xmath253 is a convex seifert surface with legendrian boundary in @xmath387 which is contained in a contact - isotopic copy of @xmath388 .
we end with a lemma that was established as claim 4.2 in @xcite .
recall that the _ contact width _ of a knot type @xmath13 is given by @xmath389 where here @xmath141 ranges over all solid tori with convex boundary representing @xmath13 .
[ lem : maxtb ] given a knot type @xmath13 , suppose @xmath0 is a pair of relatively prime integers such that @xmath390 .
then the maximal thurston - bennequin invariant of @xmath22 is @xmath391
in subsection [ ss : nont ] we classify non - thickenable tori in the knot types of the positive torus knots , and in subsection [ subsec : partiallythick ] we classify the partially thickenable tori .
subsection [ knotsontori ] discusses legendrian knots sitting on these tori as ruling curves and legendrian dividing curves .
when considering tori @xmath317 that realize the knot type of @xmath7-torus knot @xmath392 there are two different natural " coordinates to use .
the first is the longitude - meridian coordinates where the longitude comes from the intersection of a seifert surface with @xmath393 this longitude will be called the @xmath394-longitude , and these coordinates will be called the @xmath395 coordinates .
the other coordinate system has the longitude given by the framing coming from the heegaard torus that @xmath13 sits on in @xmath21 this longitude will be called the @xmath396-longitude and these coordinates will be called the @xmath397 coordinates . except where stated otherwise we will always use the more standard @xmath395 coordinates . [
basecasethickening ] suppose that the solid torus @xmath317 represents the knot type of a positive @xmath7-torus knot @xmath13 .
if @xmath317 has convex boundary then @xmath317 will thicken unless it has dividing slope @xmath98 for some @xmath398 , and @xmath154 dividing curves where @xmath399 we begin by ignoring the contact structure and building a topological model for the complement of @xmath334 see figure [ fig : setup ] .
the knot @xmath13 can be thought to sit on a torus @xmath134 that separates @xmath148 into two solid tori @xmath400 and @xmath401 , each of which can be thought of as a neighborhood of an unknot @xmath402 and @xmath403 as @xmath317 is a neighborhood of @xmath13 , we can isotope @xmath134 so that it intersects @xmath317 in an annulus and thus @xmath404 is an annulus in the complement of @xmath317 with boundary on @xmath393 moreover , there is a small neighborhood of @xmath405 which we denote @xmath406 such that @xmath407 consists of two solid tori , which we may think of as @xmath400 and @xmath408 turning this construction around @xmath409 is the complement of @xmath334 we can identify @xmath410 as a neighborhood of an annulus @xmath308 that has one boundary component a @xmath7 curve on @xmath411 and the other boundary component a @xmath412 curve on @xmath413 thus , topologically , the complement of @xmath317 can be built as the neighborhood of two unknots ( that form a hopf link ) union the neighborhood of an annulus @xmath414 bringing the contact structure back into the picture we can assume that @xmath128 , @xmath415 , is a legendrian representative of @xmath416 in the complement of @xmath417 which maximize @xmath418 in the complement of @xmath317 , subject to the condition that @xmath419 is isotopic to @xmath420 in the complement of @xmath317 . let @xmath421 , where @xmath422 if @xmath423 is a regular neighborhood of @xmath128 , then @xmath424 with respect to @xmath425 .
notice that @xmath426 is diffeomorphic to @xmath427 $ ] and contains @xmath334 we wish to change coordinates on @xmath222 so that @xmath317 is a vertical solid torus in @xmath146 specifically , @xmath222 inherits coordinates as the boundary of @xmath428 that is using the coordinate system coming from the framing @xmath429 we change coordinates so that the @xmath7 curve on @xmath222 becomes the @xmath430 curve ( which can be thought of as the longitude in the @xmath397 framing ) .
this can be done by sending the oriented basis @xmath431 for @xmath432 where @xmath433 , to the basis @xmath434 this corresponds to the map @xmath435 . then @xmath436 maps @xmath437 .
since we are only interested in slopes , we write this as @xmath438 .
similarly , we change from @xmath439 to @xmath397 .
the only thing we need to know here is that @xmath440 maps to @xmath441 .
thus @xmath141 is a thickened torus @xmath337 $ ] with dividing slope @xmath442 on @xmath443 and @xmath444 on @xmath445 now suppose @xmath446 .
this would mean that the twisting of legendrian ruling representatives of @xmath330 on @xmath447 and @xmath448 would be unequal .
then we could apply the imbalance principle to a convex annulus @xmath449 in @xmath450 between @xmath447 and @xmath448 to find a bypass along one of the @xmath451 .
this bypass in turn gives rise to a thickening of @xmath423 , allowing , by the twist number lemma @xcite , the increase of @xmath418 by one .
hence , eventually we arrive at @xmath452 and a standard convex annulus @xmath449 ; that is , the dividing curves on @xmath308 run from one boundary component of @xmath308 to the other .
since @xmath453 , the smallest solution to @xmath454 is @xmath455 .
all the other positive integer solutions are therefore obtained by taking @xmath456 and @xmath457 with @xmath458 a non - negative integer .
we can then compute the boundary slope of the dividing curves on @xmath459 where @xmath460 this will be the boundary slope for the solid torus @xmath461 containing @xmath317 .
we have @xmath462 after changing from @xmath463 to @xmath464 coordinates , and setting @xmath465 , these slopes become @xmath466 as desired .
we also notice that @xmath467 has @xmath468 dividing curves .
thus any solid torus @xmath317 will thicken unless it satisfies the conditions stated in the lemma .
we have not yet proved that tori as described in the above lemma actually exist . to rectify this problem we explicitly construct such tori in the tight contact structure on @xmath148 by gluing together tight contact structures on the pieces used in the proof of lemma [ basecasethickening ] .
more specifically we have the following . [ constructnk ]
let @xmath317 be a solid torus in the knot type of @xmath13 and set @xmath469 as noted in the proof above we can think of @xmath470 as the union of two solid tori @xmath471 ( which we think of as a neighborhood of a hopf link @xmath472 ) , together with a product neighborhood @xmath473 of an annulus @xmath308 that has one boundary component a @xmath7-curve on @xmath411 and the other boundary component a @xmath412-curve on @xmath474
. also recall that @xmath473 can be thought of as a neighborhood of an annulus @xmath475 that has boundary on @xmath317 and that the union of @xmath317 and @xmath473 is a thickened torus @xmath337 $ ] whose complement is @xmath476 now let @xmath153 denote @xmath317 with one of the two universally tight contact structures on @xmath317 with convex boundary having boundary slope @xmath477 with respect to @xmath395 , and with @xmath154 dividing curves . from the classification of tight contact structures on solid tori
this is equivalent to the condition that the convex meridional disks all have bypasses all of the same sign and thus the two contact structures on @xmath153 differ by @xmath478 .
( note that when @xmath156 there is only one contact structure . to avoid unnecessary notation
we will frequently write @xmath479 realizing that @xmath480 is the same as @xmath481 . )
let @xmath482 $ ] denote a product neighborhood of @xmath475 and put a @xmath483$]-invariant contact structure on it , where the dividing curves on @xmath484 are in standard form .
the set @xmath485 is diffeomorphic to @xmath337 $ ] and we can think of it as fibering over the annulus with fiber circles representing the knot type @xmath143 for either choice of contact structure on @xmath486 , the contact structure on @xmath487 can be isotoped to be transverse to the fibers of @xmath487 , while preserving the dividing set on @xmath488 .
it is well known , see for example @xcite , that such a horizontal contact structure is universally tight .
moreover , we see the boundary conditions on @xmath487 are @xmath489 and ( with appropriately chosen dividing curves on @xmath475 ) @xmath490 , @xmath491 when using the coordinates on @xmath222 coming from the framing @xmath492 .
we know that there are exactly two universally tight contact structures on @xmath337 $ ] with these dividing curves , differing by @xmath478 , and their horizontal annuli contain bypasses all of the same sign ; one can easily see they correspond to the two choices of universally tight contact structures on @xmath493 we know that each of these universally tight contact structures on @xmath487 embeds in the standard tight contact structure as the region between a legendrian realization of the hopf link @xmath494 thus the standard tight contact structure on @xmath148 minus @xmath487 give standard neighborhoods of a legendrian realization @xmath495 of @xmath402 , and @xmath496 of @xmath403 moreover , we know that if @xmath402 and @xmath497 are oriented so that their linking is @xmath498 then for one choice of universally tight contact structure on @xmath487 we have that @xmath495 and @xmath496 are both obtained from maximal thurston - bennequin unknots by only positive stabilizations and for the other choice of universally tight contact structure on @xmath487 we have only negative stabilizations .
we first notice that these @xmath153 just constructed in @xmath148 are non - thickenable solid tori .
[ lem : nonthicken ] the tori @xmath153 from construction [ constructnk ] are non - thickenable . by lemma [ basecasethickening ]
, it suffices to show that @xmath486 does not thicken to any @xmath499 for @xmath500 .
( we drop the @xmath501 from the notation for @xmath486 for the remainder of this proof and just assume one choice of sign is fixed throughout . ) to this end , observe that the @xmath7-torus knot is a fibered knot over @xmath348 with fiber a seifert surface @xmath253 of genus @xmath502 ( see @xcite ) .
moreover , the monodromy map @xmath347 of the fibration is periodic with period @xmath503 .
thus , @xmath504 has a @xmath503-fold cover @xmath505 .
if one thinks of @xmath504 as @xmath506 $ ] modulo the relation @xmath507 , then one can view @xmath508 as @xmath503 copies of @xmath506 $ ] cyclically identified via the same monodromy .
now note that in @xmath504 , the @xmath396-longitude intersects any given seifert surface @xmath503 times efficiently .
it is therefore evident that we can view @xmath504 as a seifert fibered space with two singular fibers ( the components of the hopf link ) .
the regular fibers are topological copies of the @xmath396-longitude , which itself is a legendrian ruling curve on @xmath509 with twisting @xmath510 .
we claim the pullback of the tight contact structure to @xmath508 admits an isotopy where the @xmath348 fibers are all legendrian and have twisting number @xmath510 with respect to the product framing . to see this we consider the contact structure on @xmath511 the neighborhood of the legendrian unknot @xmath128 ( we will use notation form construction [ constructnk ] ) . in the @xmath503-cover of @xmath504 the torus @xmath400 will lift to @xmath362 copies of the @xmath512-fold cover @xmath513 of @xmath400 and similarly @xmath401 will lift to @xmath512 copies of the @xmath362-fold cover @xmath514 of @xmath408 we can assume that @xmath411 has ruling slope @xmath515 ( that is the ruling curves are legendrian isotopic to a legendrian @xmath396-curve on @xmath516 ) and similarly for @xmath413 the ruling curves lift to curves of slope @xmath517 in @xmath518 in particular they are longitudes and have twisting @xmath519 moreover the dividing curves on @xmath513 are also longitudinal ( a different longitude of course ) .
thus we see that the contact structure on @xmath513 is just a standard neighborhood of one of the ruling curves ( pushed into the interior of the solid torus )
. similarly for @xmath520 thus each of these tori is foliated by legendrian curves isotopic to the ruling curves .
as @xmath508 is made from copies of the @xmath521 and copies of covers of the convex neighborhoods of the annuli @xmath308 we see the claimed isotopy of @xmath508 so that the @xmath348 fibers are all legendrian . if @xmath486 can be thickened to @xmath499 , then there exists a legendrian curve topologically isotopic to the regular fiber of the seifert fibered space @xmath504 with twisting number greater than @xmath510 , measured with respect to the seifert fibration . pulling back to the @xmath503-fold cover @xmath508
, we have a legendrian knot which is topologically isotopic to a fiber but has twisting greater than @xmath510 . call this legendrian knot with greater twisting @xmath375 .
we will obtain a contradiction , thus proving that @xmath486 can not be thickened to @xmath499 .
since @xmath253 is a punctured surface of genus @xmath522 , we can cut @xmath253 along @xmath523 disjoint arcs @xmath524 , all with endpoints on @xmath525 , that yield a polygon @xmath526 .
thus we have a solid torus @xmath527 embedded in @xmath508 .
we first calculate @xmath528 as measured in the product framing .
to do so , note that a longitude for this torus intersects @xmath529 @xmath530 times , and a meridian for this torus is composed of @xmath531 copies each of the @xmath523 arcs @xmath524 , as well as @xmath532 arcs @xmath533 from @xmath525 .
now since @xmath525 is a preferred longitude downstairs in @xmath504 , we know that @xmath310 intersects these @xmath533 , @xmath534 times positively .
but then the edge - rounding that results at each intersection of an @xmath535 with an @xmath536 yields @xmath532 negative intersections with @xmath310 .
thus we obtain after edge - rounding that @xmath537 .
now as in lemma 3.2 in @xcite , we take @xmath538 and pass to a ( new ) finite cover of the base by tiling enough copies of @xmath526 together so that @xmath375 is contained in a solid torus @xmath539 .
we notice that @xmath540 is foliated by legendrian knots with twisting @xmath510 that are isotopic to the @xmath348 fibers in the product structure and that the dividing curves on the boundary of the solid torus are longitudinal .
thus @xmath539 is a standard neighborhood of a legendrian curve with twisting @xmath510 with respect to the product structure . we know that inside any such solid torus any legendrian isotopic to the core of the torus
has twisting less than or equal to @xmath510 ( or else one could violate the bennequin bound ) .
thus @xmath375 can not exist .
we now observe that the @xmath153 are the only candidates for non - thickenable tori in the knot type of a positive @xmath7-torus knot .
in addition , we compute what the rotation numbers of legendrian curves on @xmath541 are .
[ candidates][mustbenk ] let @xmath317 be a solid torus with convex boundary representing the positive @xmath7-torus knot .
if @xmath317 does not thicken then @xmath317 must be isotopic to one of the @xmath153 from construction [ constructnk ] .
moreover , if @xmath541 is isotoped so that the ruling curves are meridional then the meridional curves will have rotation number @xmath542 , and if @xmath541 is isotoped so that the ruling curves are @xmath394-longitudes then the @xmath394-longitudes have rotation number 0 .
let @xmath317 be a solid torus with convex boundary as in the lemma .
if @xmath317 does not thicken then from the discussion in the proof of lemma [ basecasethickening ] we see that @xmath543 can be thought of as the union of two solid tori @xmath471 ( which we think of as a standard neighborhood of a legendrian realization @xmath544 of the hopf link @xmath472 ) together with a product neighborhood @xmath473 of an annulus @xmath308 that has one boundary component a @xmath7-curve @xmath545 on @xmath411 and the other boundary component a @xmath412-curve @xmath546 on @xmath474 . from the proof of lemma [ basecasethickening ]
we see that @xmath547 and @xmath548 for some positive integer @xmath549 we can assume that @xmath550 are ruling curves on the tori @xmath411 and @xmath413 ruling curves on @xmath308 provide a legendrian isotopy form @xmath545 to @xmath551 thus @xmath545 and @xmath546 have the same rotation numbers . from this and the discussion in construction [ constructnk ] we see that the signs of the stabilizations must be the same , thus @xmath552 and @xmath553 hence @xmath543 is contactomorphic to @xmath554 thinking of the neighborhood @xmath473 as a product neighborhood @xmath410 of the annulus @xmath475 ( using the notation from lemma [ basecasethickening ] and construction [ constructnk ] ) we see that @xmath555 must be a universally tight contact structure on @xmath337 $ ] ( or else we could find a bypass for one of the @xmath128 and hence thicken @xmath317 ) .
we will only get a universally tight contact structure on @xmath555 if @xmath317 has convex meridian discs with bypasses all of the same sign , as one may easily check by computing the relative euler class of @xmath555 .
the statement about meridional ruling curves is obvious . to verify the statement for the @xmath394-longitudes we need to use the function @xmath378 that measures the rotation numbers of curves on convex tori @xmath134 that was discussed in subsection [ sec : rotationfunction ] .
we fix our attention on @xmath556 ( leaving the analogous case for @xmath557 to the reader ) .
recall @xmath495 is a legendrian unknot obtained from the maximal thurston - bennequin unknot by @xmath558 positive stabilizations .
thus if @xmath400 is a standard neighborhood of @xmath495 and @xmath330 is a @xmath7-ruling curve on @xmath411 then we see @xmath559 where @xmath560 is a meridional curve on @xmath411 and @xmath561 is a longitude . if we isotope @xmath562 so that the ruling curves are @xmath396-curves then there is a convex annulus @xmath563 in @xmath148 from the curve @xmath330 on @xmath411 to an @xmath396-longitude @xmath564 on @xmath562 that has dividing curves that run from one boundary component to the other .
thus we can rule @xmath563 by curves parallel to @xmath330 and @xmath564 and see that @xmath330 and @xmath564 are legendrian isotopic .
in particular @xmath565 let @xmath566 denote a @xmath394-longitude on @xmath567 since we know that @xmath568 where @xmath569 is a meridian on @xmath570 we see that @xmath571 the theorem merely collects the statements of lemmas [ basecasethickening ] , [ lem : nonthicken ] , and [ mustbenk ] , together with construction [ constructnk ] .
in this section we use the notation established in construction [ constructnk ] and the subsequent lemmas of the previous section .
we notice that @xmath572 can always be constructed so that it is contained in any arbitrarily small neighborhood of the annulus @xmath308 union @xmath573 from construction [ constructnk ] and any two such constructed @xmath572 are isotopic ( and hence the corresponding @xmath153 are isotopic too ) . throughout this subsection
we will always be talking about tori in the knot type of a positive @xmath7-torus knot .
[ lem1 ] let @xmath317 be a solid torus in @xmath153 with standard convex boundary having dividing slope @xmath574 if @xmath575 , then there can be no bypass @xmath275 inside @xmath576 for @xmath577 attached along a ruling curve of slope @xmath396 .
notice that @xmath576 is diffeomorphic to @xmath578.$ ] moreover the slope on @xmath579 is in @xmath580 and on @xmath581 is @xmath167 if such a bypass existed then there would be a torus @xmath134 in @xmath578 $ ] with dividing slope @xmath170 . thus the contact structure on @xmath337 $ ]
is not minimally twisting , but this is impossible as the contact structure on @xmath148 we are considering is tight .
( notice that if @xmath582 then a bypass can be attached that merely reduces the number of dividing curves . )
[ movea ] assume that @xmath575 .
let @xmath495 and @xmath496 be the two unknots used in the construction of @xmath572 and @xmath308 the annulus , see construction [ constructnk ] .
let @xmath573 be the standard neighborhood of @xmath544 used in this construction .
suppose that @xmath583 is any convex annulus in the complement of @xmath584 , which has boundary legendrian ruling curves parallel to @xmath550 on @xmath585 , and such that @xmath583 is isotopic to @xmath308 in the complement of @xmath334 then the dividing curves on @xmath583 run from one boundary component to the other and there is a contact isotopy of @xmath148 taking @xmath586 to @xmath587 first notice that if @xmath583 is disjoint from @xmath308 then the first statement is clear since if the dividing curves were not as stated there would be a bypass for @xmath588 on a @xmath396 ruling curve contradicting lemma [ lem1 ] .
( to see this recall that @xmath589 . ) for the second statement notice that there will be a diffeomorphism of @xmath148 fixing ( set - wise ) @xmath573 and sending @xmath308 to @xmath590 moreover we can assume this diffeomorphism preserves the dividing sets on @xmath591 and sends @xmath592 to @xmath593 thus we may isotope the diffeomorphism so that it is a contactomorphism in a neighborhood of @xmath594 as the contact structure on the complementary solid torus is unique ( as indicated in the proof of lemma [ mustbenk ] ) we can further isotope this map to a contactomorphism of @xmath148 . as the space of contactomorphisms of the standard contact structure on @xmath148 ( that fix a point ) is contractible
it is standard to find a contact isotopy as desired .
if @xmath583 and @xmath308 are not disjoint then we can use isotopy discretization as discussed in subsection [ sssec : discritize ] to find a sequence of annuli @xmath595 such that @xmath596 each @xmath597 is a convex annulus with boundary legendrian ruling curves parallel to @xmath550 and for each @xmath598 and @xmath599 are disjoint .
the result now follows .
[ prop : pt ] let @xmath317 be a solid torus in @xmath153 with standard convex boundary having dividing slope @xmath574 if @xmath575 , then @xmath317 will thicken to a solid torus @xmath320 of slope @xmath99 but not beyond .
moreover , @xmath320 is isotopic to @xmath600 [ rem : pt ] notice that if @xmath601 then @xmath153 can be thinned to a torus @xmath320 that has the same dividing slope as @xmath153 but fewer dividing curves .
this will allow for the destabilization of the legendrian knots @xmath495 and @xmath496 used in lemma [ basecasethickening ] , which in turn , allow for the thickening of @xmath320 past @xmath153 .
thus we see when @xmath602 that there are no partially thickenable tori in @xmath153 . for the right handed trefoil knot there is another ,
arguably simpler , proof of this result that is more in the spirit of the previous subsection .
we present a unified proof for all @xmath7 here and refer to @xcite for the alternate argument .
suppose that @xmath317 can be thickened past the slope @xmath99 .
then it can be thickened to @xmath499 for some @xmath603 we can arrange @xmath317 to have ruling curves isotopic to @xmath396-longitudes .
taking an annulus @xmath604 from a ruling curve on @xmath319 to a ruling curve on @xmath605 ( of slope @xmath396 ) we see that there are enough disjoint bypasses @xmath606 along @xmath604 for @xmath319 to thicken @xmath317 to a solid torus with dividing slope outside the interval @xmath607 if the bypasses were contained in @xmath486 this would of course be a contradiction , as we could attach them to @xmath319 to obtain a convex torus in @xmath608 with slope @xmath245 .
we now argue that we can isotope @xmath486 so that it contains all the bypasses .
this contradiction will imply that @xmath317 can not be thickened to @xmath499 for any @xmath603 to this end let @xmath495 and @xmath496 be the two unknots used in the construction of @xmath572 and @xmath308 the annulus , see construction [ constructnk ] . from the construction we know that @xmath572 is obtained by taking the union of arbitrarily small neighborhoods @xmath573 of @xmath544 and @xmath473 of @xmath308 ( and rounding corners ) .
consider the 2-complex @xmath609 obtained from @xmath544 by attaching ( an extension of ) @xmath414 clearly @xmath572 can be isotoped to be contained in any arbitrarily small neighborhood of @xmath610 we now consider the intersection of @xmath609 with the bypasses above .
first we notice there is a contact isotopy of the @xmath611 making them transverse to @xmath610 so the intersection consists of closed curves , vertices ( corresponding to the intersection of @xmath275 with @xmath544 ) and arcs . we may now choose standard neighborhoods @xmath423 of the legendrian knots @xmath128 ( and possibly isotope the interiors of the @xmath611 ) so that @xmath423 intersects the bypass disks in disks ( that is each vertex of @xmath612 becomes a disk ) that are disjoint from the simple closed curves in @xmath612 .
we may now isotope @xmath609 so that @xmath613 is a convex annulus @xmath583 with legendrian boundary ruling curves on @xmath614 and intersects the bypass disks as @xmath613 does .
let @xmath275 denote one of these bypasses .
we will show how to isotope @xmath275 to be disjoint from @xmath572 and observe that this argument can be applied to each of the @xmath611 resulting in the desired contradiction .
it is clear that if @xmath615 then @xmath275 may be assumed to be contained in @xmath600 thus we show how to eliminate the intersections between @xmath275 and @xmath609 .
we first show how to remove the closed curves from the intersection .
let @xmath375 be an innermost closed curve in @xmath616 .
( that is @xmath375 bounds a disk on @xmath275 that does not contain any other points of intersection between @xmath609 and @xmath275 . )
notice that from the set - up above @xmath375 is an intersection between @xmath583 and @xmath299 we can isotope @xmath583 , rel boundary , so as to eliminate @xmath375 from @xmath617 ( notice along the way , we might also eliminate some intersections between @xmath609 and other @xmath611 but we do not increase the number of intersections between @xmath611 and @xmath609 . ) by lemma [ movea ] we see that this isotopy can be done by a contact isotopy , thus resulting in a new @xmath609 with all the above properties but fewer intersections with the disk @xmath299 continuing we can assume that @xmath616 contains no simple closed curves . now suppose that @xmath375 is an arc in @xmath616 that connects two vertices .
we can take an interval in @xmath375 that is disjoint from the intersection of @xmath275 and @xmath573 and then isotope @xmath583 as above to remove this interval from the intersection of @xmath609 and @xmath299 thus @xmath618 consists of `` stars '' and arcs ; that is , each connected component of the intersection is either an arc ( with both endpoints on @xmath619 ) or has a single vertex with several edges ( connecting the vertex to @xmath619 ) .
we again notice that the arcs of intersection are intersections between @xmath583 and @xmath275 and thus we may remove them as above if they are outermost ( that is , separates off a disks from @xmath275 that does not contain any points of intersection between @xmath275 and @xmath609 ) .
we are now left to consider outermost `` stars '' .
given such a star we assume that the vertex comes from an intersection between @xmath275 and @xmath620 so we have a disk @xmath621 corresponding to the vertex and the @xmath362 edges corresponding to @xmath622 ( we would have @xmath512 edges if @xmath275 intersected @xmath496 at the vertex under consideration ) .
recall @xmath572 is obtained by taking the union of an @xmath257 invariant neighborhood of @xmath583 and @xmath573 and rounding corners .
so we can isotope @xmath275 slightly near @xmath583 so that @xmath623 consists of @xmath624 union @xmath362 strips corresponding to thickenings of the edges of @xmath625 . from this
it is easy to see that @xmath626 consists of @xmath362 arcs , @xmath627 one of these arcs , which we denote @xmath628 , divides @xmath275 into two disks , one of which contains all the other @xmath629 s ( and no other intersections with @xmath609 ) .
denote this disk @xmath630 .
notice that @xmath631 does not intersect @xmath632 each arc @xmath633 separates a disk @xmath634 from @xmath630 that is disjoint from the interior of @xmath635 if we push @xmath636 across the disk @xmath634 then we get a new torus @xmath135 in @xmath637 recall that the ruling slope on @xmath448 was by @xmath412-curves and that the isotopy of @xmath636 to @xmath135 can be done fixing one of these curves
. thus the contact twisting of the ruling curve is still @xmath510 , however , the ruling curve on a convex torus with dividing slope in @xmath638 will always have twisting less than or equal to @xmath510 with equality if and only if the dividing slope is @xmath167 thus we see that @xmath135 has dividing slope @xmath99 and hence is contact isotopic to @xmath639 that is we can find a contact isotopy that eliminates one of the arcs of intersection . continuing in this way we push @xmath640 across the other disks @xmath634 by a contact isotopy resulting in the disk @xmath630 being contained in @xmath635 now pushing @xmath640 across @xmath630 will not change the dividing set since @xmath153 is a non - thickenable torus . combining these isotopies
we have removed the outermost `` star '' in @xmath641 by successively removing outermost arcs or `` stars '' from @xmath616 we can eventually make @xmath275 disjoint from @xmath609 and thus contained in @xmath637 let @xmath317 be a solid torus in @xmath153 with standard convex boundary having dividing slope @xmath642 then @xmath317 will thicken to the solid torus @xmath157 ( which is a standard neighborhood of the maximal thurston - bennequin invariant legendrian @xmath7-torus knot ) . given such a torus @xmath317 we know from the construction and discussion in subsection [ ss : nont ] that we can thicken @xmath317 to a solid torus @xmath320 whose boundary is convex with two dividing curves of slope @xmath170 and in the complement of @xmath320 we will have @xmath572 . now taking an annulus from @xmath320 to @xmath447 ( using the notation from construction [ constructnk ] )
we will see that there is a bypass for @xmath447 and thus we can increase the thurston - bennequin of @xmath620 as in the proof of lemma [ basecasethickening ] we see that @xmath320 will thicken to some @xmath643 with @xmath603 thus we know we can thicken past @xmath643 unless @xmath644 , and hence we can thicken to @xmath645 we are now ready to establish the main results stated in the introduction concerning partially thickenable tori .
the statements in the theorem just collect the facts from proposition [ prop : pt ] , remark [ rem : pt ] and lemma [ lem : basicrange ] . for statement
( 1 ) notice that if @xmath172 then a convex torus with two dividing curves of slope @xmath19 will lie inside one of the @xmath646 for @xmath647 or @xmath157 . from the classification of the @xmath646 we know
there is a convex torus with two dividing curves and infinite dividing slope inside each of the @xmath646 and it will cobound with @xmath648 a unique basic slice , @xcite .
moreover there are two distinct such tori in @xmath157 and each of these two will cobound with @xmath649 a unique basic slice . inside a basic slice
there is a unique , up to contactomorphism , convex torus of slope @xmath19 . thus given any convex torus @xmath134 with two dividing curves of slope
@xmath19 we can use this data to construct a contactomorphism of @xmath148 taking @xmath134 to one of the tori described above .
then the discussion in subsection [ sssec : isotopy ] gives a contact isotopy from @xmath134 to one of these tori .
as there are @xmath173 such tori this establishes statement ( 1 ) of the theorem .
the other statements in the corollary have analogous proofs . in this section
we prove two fundamental propositions about legendrian knots on partially thickenable , and non - thickenable , tori that will be necessary in our classification of cables of torus knots .
[ curvesonpt1 ] suppose @xmath13 is a positive @xmath7-torus knot and @xmath650 is a solid torus constructed above in subsection [ ss : nont ] , for some @xmath651 with @xmath652 .
let @xmath653 and @xmath654 .
if @xmath134 is the convex torus in @xmath650 with two dividing curves and dividing slope @xmath43 and @xmath259 is a legendrian divide on @xmath134 , then : 1 . for any @xmath655 and @xmath656 , any convex torus @xmath135 on which the legendrian knot @xmath657 sits bounds a partially thickenable , or non - thickenable , torus in @xmath650 .
the legendrian knot @xmath658 sits on a convex torus @xmath135 that bounds a solid torus that thickens to @xmath157 .
we will concentrate on the legendrian divide @xmath259 on a torus @xmath134 inside @xmath659 below , but analogous arguments also work for @xmath660 . recall that inside the solid torus @xmath659 there is a convex torus @xmath135 with two dividing curves and dividing slope @xmath661 .
let @xmath662 be a legendrian ruling curve on @xmath135 of slope @xmath43 .
using an annulus @xmath308 that @xmath259 and @xmath662 cobound , it is easy to see that @xmath662 is obtained from @xmath259 by stabilizing @xmath663 times .
we want to compute the difference between the rotation number of @xmath259 on @xmath134 and @xmath662 on @xmath135 .
the region between @xmath134 and @xmath135 is a thickened torus and the difference in these rotation numbers will be given by the value of the relative euler class of the thickened torus evaluated on the annulus @xmath308 .
to compute this we use the classification of tight contact structures on thickened tori , as given in @xcite , and the fact that @xmath659 is universally tight .
in particular , we can compute the relative euler class @xmath664 of the thickened torus cobounded by @xmath665 and @xmath134 : @xmath666 where @xmath667 stands for the poincar dual and we are using the basis for @xmath668 given by the meridian and longitude and @xmath669 .
we can use this to compute the difference between the rotation number of the @xmath0 curve on @xmath665 and on @xmath134 which is @xmath670 .
that is , @xmath662 is obtained from @xmath259 by @xmath24 positive stabilizations . according to theorem [ thm :
partially ] the solid torus that @xmath135 bounds can be thickened to @xmath157 . as any further negative stabilizations of @xmath259 can be seen on @xmath135 as well ( by having @xmath259 intersect the dividing curves in a non - minimal way ) we have established the second point in the proposition . for the first point in the proposition notice
that the discussion above shows that @xmath671 , with @xmath672 , can not sit as a legendrian curve on a convex torus with dividing slope @xmath661 ( since otherwise @xmath673 ) .
suppose that @xmath671 is also isotopic to a curve on a convex torus @xmath135 that is neither a partially thickenable , nor a non - thickenable , torus in @xmath659 .
( this @xmath135 is not the same as in the previous paragraph . )
we can extend the isotopy of @xmath671 to an ambient contact isotopy and thus we may assume that one fixed copy of @xmath671 sits on both a partially ( or non- ) thickenable torus @xmath134 in @xmath659 and on a torus @xmath135 that is not a partially ( or non- ) thickenable torus in @xmath659 .
we may isotope @xmath135 near @xmath671 so that it agrees with @xmath134 .
let @xmath317 be a standard neighborhood of @xmath671 that intersects @xmath134 and @xmath135 on a subset of @xmath674 .
let @xmath308 and @xmath475 be the annuli in the complement of @xmath317 given by @xmath134 and @xmath135 , respectively .
we may further assume that @xmath675 are ruling curves on @xmath319 and that all ruling curves on @xmath319 are parallel to @xmath676 .
these annuli are properly topologically isotopic in the complement of a neighborhood of @xmath677 .
( this follows from standard results concerning incompressible surface in seifert fibered spaces . )
we can use isotopy discretization as discussed in subsection [ sssec : discritize ] to find a sequence of annuli @xmath678 such that @xmath679 each @xmath597 is a convex annulus with boundary consisting of legendrian ruling curves parallel to @xmath550 and for each @xmath680 and @xmath599 are disjoint and related by a bypass attachment .
notice that this gives us a sequence of tori @xmath681 that are related by bypass attachments in the complement of @xmath671 .
the torus @xmath682 is partially ( or non- ) thickenable inside of @xmath659 .
we inductively show that @xmath683 is also such a convex torus .
assume that we have shown that @xmath684 is such a torus ; then recall @xmath683 is obtained from @xmath684 by attaching a bypass from the outside ( that is from the outside of the solid torus @xmath684 bounds ) or from the inside .
if we attach the bypass to @xmath684 from the outside we get a new convex torus that bounds a thickening of the solid torus that @xmath684 bounds , and so is also a partially ( or non- ) thickenable torus in @xmath659 . if we attach the bypass from the inside then as there is an edge in the farey tessellation between @xmath107 and @xmath661 ( and the dividing slope of @xmath684 is contained in the interval @xmath685 ) we see that the dividing slope of @xmath683 is in @xmath686.$ ] but as in the previous paragraph the restriction on the rotation number and thurston - bennequin invariant implies that the dividing slope can not be @xmath661 .
thus the dividing slope of @xmath683 is in @xmath687 .
in particular it bounds a partially ( or non- ) thickenable solid torus in @xmath659 .
thus @xmath688 bounds a partially ( or non- ) thickenable solid torus in @xmath659 , which contradicts our assumption on @xmath135 . from this
we see that any convex solid torus on which @xmath671 sits bounds a partially ( or non- ) thickenable solid torus in @xmath659 .
[ curvesonpt2 ] suppose @xmath13 is a positive @xmath7-torus knot and @xmath650 is a solid torus constructed above in subsection [ ss : nont ] , for some @xmath651 with @xmath652 .
let @xmath689 and @xmath690 .
if @xmath259 is a ruling curve on @xmath691 with slope @xmath43 , then : 1 . for any @xmath692 and @xmath656 the convex torus containing @xmath657
has dividing slope @xmath107 and is contained in @xmath650 .
the legendrian knot @xmath693 sits on a convex torus @xmath134 that bounds a solid torus that thickens to @xmath157 .
we will concentrate on a legendrian ruling curve @xmath259 on @xmath694 below , but analogous arguments also work for @xmath660 .
the proof of the second point in the proposition follows exactly as in the proof of proposition [ curvesonpt1 ] and in particular , @xmath695 sits on a convex torus @xmath135 inside of @xmath659 with dividing slope @xmath661 .
moreover , any legendrian knot that is a stabilization of @xmath259 that sits on @xmath135 will have at least @xmath24 positive stabilizations .
the first point follows the same outline as the proof of claim 6.5 in @xcite , but is augmented by what we know from proposition [ prop : pt ] . more specifically ,
if @xmath135 also contains @xmath259 and is isotopic to @xmath665 then standard properties of incompressible surfaces in seifert fibered spaces ( recall that the sub - annulus of @xmath135 contained in the complement of a neighborhood of @xmath259 is incompressible in the complement of @xmath259 ) imply that @xmath135 must be isotopic to @xmath665 relative to @xmath259 .
therefore , it suffices to show that the slope of the dividing set does not change under any isotopy of @xmath665 relative to @xmath259 .
although we would like to say that the isotopy leaves the dividing set of @xmath665 invariant , this is not true , see @xcite , though we will show the dividing slope does not change .
if @xmath135 is isotopic to @xmath665 relative to @xmath259 then the standard isotopy discretization used above implies that there is a sequence of surfaces @xmath696 such that each @xmath312 is convex and obtained from the previous @xmath697 by a bypass attachment .
we inductively assume the following : 1 .
@xmath253 is a convex torus which contains @xmath259 and satisfies @xmath698 and @xmath699 .
@xmath253 is contained in a @xmath483$]-invariant @xmath337 $ ] with @xmath700 and @xmath701 and is parallel to @xmath702 .
3 . there is a contact diffeomorphism @xmath703 which takes @xmath704 $ ] to a standard @xmath257-invariant neighborhood of @xmath665 and matches up their complements .
notice that if we prove all the @xmath312 satisfy these conditions then @xmath135 will satisfy the conclusions of the first point of the proposition , thus completing our proof .
we assume that @xmath312 satisfies the inductive hypothesis above . using the terminology from the proof of proposition [ curvesonpt1 ]
we notice that if a bypass is attached to @xmath312 from the outside then the dividing slope can not change or this would give a thickening of our non - thickenable solid torus .
if the bypass is attached from the inside , then let @xmath307 be the torus obtained after the bypass is attached .
by lemma [ lem : eaec ] we see that @xmath705 must lie in @xmath706 $ ] .
since the argument in the first paragraph of this proof disallows @xmath707 , we know that @xmath708 .
suppose that @xmath709 .
let @xmath710 be a convex torus of slope @xmath661 and
@xmath711 in the interior of the solid torus bounded by @xmath307 . take a legendrian curve @xmath662 on @xmath307 which is parallel to and disjoint from @xmath259 , and intersects @xmath712 minimally .
( the existence of such a curve is easily established by noting that @xmath259 is obtained from a curve @xmath662 that minimally intersects @xmath712 by a sequence of finger moves " across @xmath712 .
inducting on the number of such moves one may show that a parallel copy of @xmath662 can be made disjoint from these moves . ) similarly , consider @xmath713 on @xmath710 . using lemma [ lem : basicrange ]
we see that latexmath:[$|\gamma_{\sigma'}\cap l'| > @xmath662 and @xmath713 will contain bypasses for @xmath307 that are disjoint from @xmath259 .
after successive attachments of such bypasses , we eventually obtain @xmath715 of slope @xmath661 containing @xmath259 , a contradiction . therefore ( observing the restriction on the number of components of @xmath716 are dictated by @xmath717 ) we see that condition ( 1 ) is preserved .
suppose @xmath307 is obtained from @xmath253 by a single bypass move .
since @xmath718 , either the bypass attachment was trivial or @xmath719 is either increased or decreased by 2 .
suppose first that @xmath720 , where @xmath317 is the solid torus bounded by @xmath253 . for convenience ,
suppose @xmath721 inside @xmath337 $ ] satisfies conditions ( 2 ) and ( 3 ) of the inductive hypothesis .
in particular @xmath682 is a torus outside of @xmath317 with two dividing curves .
the tori @xmath682 and @xmath307 cobound a thickened torus @xmath722 $ ] with non - rotative contact structure .
thus by the classification of tight contact structures on solid tori , we can factor a non - rotative outer layer which is the new @xmath723 $ ] .
it is easy to see that this new @xmath704 $ ] satisfies conditions ( 2 ) and ( 3 ) of the inductive hypothesis .
now suppose @xmath724 .
if @xmath320 is the solid torus @xmath307 bounds then we prove that there exists a non - rotative outer layer @xmath722 $ ] for @xmath725 , where @xmath726 .
this follows from repeating the procedure in the proof of lemma [ basecasethickening ] , where legendrian representatives of @xmath402 and @xmath497 were thickened and then connected by a vertical annulus .
this time the same procedure is carried out with the provision that the representatives of @xmath402 and @xmath497 lie in @xmath725 .
once the maximal thickness for representatives of @xmath402 and @xmath497 is obtained , after rounding we get a convex torus in @xmath725 parallel to @xmath307 but with @xmath711 .
therefore we obtain a non - rotative outer layer @xmath722 $ ] .
in this section we classify the simple cables of positive torus knots . these classification results and their proofs are very similar to those in @xcite and the first two of them follow directly from @xcite .
we include sketches here to demonstrate the classification strategy discussed in subsection [ sssec : strategy ] and as a warm - up for the more intricate results in the next section .
[ poscableclass ] suppose @xmath13 is a positive @xmath7-torus knot
. if @xmath727 are relatively prime integers with @xmath728 then @xmath22 is legendrian simple .
moreover , there is a unique maximal thurston - bennequin invariant representative @xmath259 of @xmath22 which has invariants @xmath729 and @xmath730 all other legendrian representatives of @xmath22 destabilize to @xmath53 we establish the theorem by ( 1 ) proving the above formula for @xmath731 ( 2 ) showing there is a unique legendrian knot @xmath259 with this as its thurston - bennequin invariant and ( 3 ) showing that any other legendrian knot in this knot type is a stabilization of @xmath259 . to show ( 1 ) we let @xmath330 be any legendrian knot in the knot type @xmath22 .
there is a solid torus @xmath141 realizing the knot type @xmath13 that contains @xmath330 in @xmath732 we know there is a seifert surface for @xmath22 with euler characteristic @xmath733 thus the bennequin inequality implies @xmath734 from this we see that the twisting of the contact planes along @xmath330 measured with respect to @xmath149 is less than or equal to @xmath735 our condition that @xmath736 implies that @xmath737 , from which we can conclude that @xmath149 can be made convex without moving @xmath738 let @xmath94 be the slope of the dividing curves on @xmath732 we know @xmath739 or negative .
moreover , @xmath740 with equality if and only if @xmath741 .
since we know that @xmath742 is @xmath56 plus @xmath743 and @xmath743 is @xmath744 times the number of dividing curves , we clearly see that the maximal possible thurston - bennequin invariant is realized on the the boundary of a solid torus @xmath141 with convex boundary having two dividing curves of slope @xmath745 if @xmath141 is the standard neighborhood of a legendrian knot in the knot type @xmath13 with maximal thurston - bennequin invariant then a ruling curve of slope @xmath43 will give a legendrian knot @xmath259 in the knot type @xmath22 realizing this bound as its thurston - bennequin invariant .
thus we have computed @xmath746 .
notice we have also shown that if @xmath330 is any other legendrian knot with @xmath747 then @xmath330 will sit on the boundary of a standard neighborhood of a maximal thurston - bennequin invariant legendrian knot representing @xmath13 . since there is a unique such knot , standard arguments , like those in @xcite and discussed in subsection [ sssec : isotopy ] , show that @xmath330 is legendrian isotopic to @xmath53 thus we have shown there is a unique legendrian representative with maximal thurston - bennequin invariant .
we are left to check ( 3 ) . to this end
let @xmath330 be a legendrian knot in the knot type @xmath22 with @xmath748 and let @xmath141 be a solid torus in the knot type @xmath13 such that @xmath330 sits on @xmath732 as mentioned above we can assume that @xmath149 is convex .
let @xmath94 be the dividing slope for @xmath149 .
if @xmath94 is positive then there is some integer @xmath749 such that @xmath750 .
( a similar argument will hold for @xmath94 negative . )
thus there is a convex torus @xmath134 inside @xmath141 with two dividing curves of slope @xmath751 .
as @xmath752 for any slope @xmath753 $ ] with equality if and only if @xmath754 we see that the @xmath0 ruling curve on @xmath134 has thurston - bennequin invariant less than or equal to @xmath742 and it is strictly less than @xmath742 unless @xmath755 .
taking an annulus between @xmath330 and a ruling curve on @xmath134 we can find a bypass to show that @xmath330 destabilizes unless @xmath755 . in this case
we can assume that @xmath134 is @xmath149 and @xmath141 is a standard neighborhood of a legendrian knot in the knot type @xmath143 as @xmath13 is legendrian simple and @xmath26 is not the maximal thurston - bennequin invariant we can thicken @xmath141 to a solid torus @xmath145 that is a standard neighborhood of a legendrian knot with @xmath756 we can now use the ruling curve on @xmath757 to show that @xmath330 destabilizes . [ negcableclass ] suppose @xmath13 is a positive @xmath7-torus knot .
if @xmath727 are relatively prime integers with @xmath758 and @xmath54 , then @xmath22 is also legendrian simple .
moreover , @xmath759 and the set of rotation numbers realized by @xmath760 is @xmath761 where @xmath26 is the integer that satisfies @xmath58 all other legendrian knots destabilize to one of these maximal thurston - bennequin knots .
notice that the restriction @xmath758 is reasonable as when @xmath762 we know @xmath763 this theorem is essentially theorem 3.6 from @xcite , the only difference being that @xmath13 is not uniformly thick . as we saw in the previous proof
the only real difference in this case where @xmath13 is not uniformly thick is that we have to be careful to argue that legendrian knots with non - maximal thurston - bennequin invariants destabilize .
but in this case we see that if @xmath330 is any legendrian knot in the knot type @xmath764 then it sits on a convex torus @xmath134 bounding a solid torus @xmath141 in the knot type @xmath13 and there is either a torus @xmath135 parallel to @xmath134 inside @xmath141 or outside @xmath141 such that @xmath135 is convex with dividing slope @xmath3 we can use @xmath135 to find a destabilization of @xmath738 [ intermediatecableclass ] suppose @xmath13 is a positive @xmath7-torus knot with @xmath42 .
if @xmath727 are relatively prime positive integers with @xmath765 but @xmath766 , where @xmath767 is as in theorem [ main ] , then @xmath22 is also legendrian simple .
moreover , @xmath759 and the set of rotation numbers realized by @xmath768 is @xmath769 where @xmath26 is the integer that satisfies @xmath770 all other legendrian knots destabilize to one of these maximal thurston - bennequin knots . establishing the classification of maximal thurston - bennequin legendrian knots in this knot type
can be done exactly as in theorem 3.6 from @xcite , see @xcite for details , except when @xmath771 for some @xmath26 not relatively prime to @xmath772 .
if @xmath259 is a legendrian knot in the knot type @xmath22 for such an @xmath773 and @xmath259 has maximal thurston - bennequin invariant , then , as discussed above , @xmath259 will sit as a legendrian divide on a convex torus @xmath134 in the knot type @xmath13 .
such a torus bounds a solid torus @xmath141 that can be thickened to a solid torus with convex boundary having two dividing curves of slope @xmath107 .
as mentioned in corollary [ cor : tori ] , see also remark [ rem : pt ] , we see that this torus further thickens to @xmath157 .
thus the reasoning in theorem 3.6 in @xcite applies .
if @xmath259 is a legendrian knot in the knot type @xmath22 with @xmath774 , then it again sits as a legendrian divide on a convex torus @xmath134 .
if @xmath134 is not @xmath691 then according to corollary [ cor : tori ] it will bound a solid torus that thickens to @xmath157 . if @xmath775 then since @xmath776 , by assumption , we know @xmath777 and hence @xmath134 has more than two dividing curves .
below we show that we can find a torus @xmath135 , inside the solid torus @xmath134 bounds , with two less dividing curves on which @xmath259 also sits .
of course this new torus will thicken to @xmath157 and hence we are done as above . to find @xmath135 notice that according to the classification of contact structures on thickened tori we can find a convex torus @xmath778 inside of @xmath141 , the solid torus @xmath134 bounds , with two dividing curves of slope @xmath107 .
let @xmath779 $ ] be the thickened torus that @xmath134 and @xmath778 cobound .
take a simple closed curve @xmath375 on @xmath778 that intersects a curve of slope @xmath107 one time .
let @xmath780 $ ] be an annulus in @xmath781 running from @xmath375 on @xmath778 to @xmath134 .
we can arrange that @xmath550 consists of ruling curves on @xmath778 and @xmath134 .
now if @xmath782 then there will be at least 2 non - adjacent bypasses on @xmath308 for @xmath134 .
thus one of them will be disjoint from @xmath259 .
pushing @xmath134 across this bypass will result in the torus @xmath135 with fewer dividing curves than @xmath134 and on which @xmath259 sits .
since we are considering @xmath7-torus knots notice that @xmath772 is odd and thus @xmath783 can not be even , thus the condition that @xmath782 is satisfied .
we are left to show that any legendrian knot with non - maximal thurston - bennequin invariant destabilizes .
let @xmath330 be a legendrian knot in the knot type @xmath22 with @xmath784 we know that @xmath330 can be put on a convex torus @xmath134 that bounds a solid torus @xmath141 representing the knot type @xmath143 let @xmath94 be the dividing slope of @xmath358 if @xmath785
then there is a torus @xmath135 parallel to @xmath134 inside @xmath141 with dividing slope @xmath3 we can use an annulus that cobounds @xmath330 and a legendrian divide on @xmath135 to show that @xmath330 destabilizes .
now suppose that @xmath786 if @xmath787 for some @xmath26 then from lemma [ lem : basicrange ] we see that @xmath788 with equality if and only if @xmath789 . since @xmath790 we can let @xmath135 be a torus inside @xmath141 that is parallel to @xmath134 and has dividing slope @xmath661 and use an annulus between @xmath330 and a ruling curve on @xmath135 to show @xmath330 destabilizes .
if @xmath94 is not in @xmath791 for any @xmath26 then from theorem [ thm : partially ] we know there is a torus @xmath135 outside @xmath141 that is parallel to @xmath134 and has dividing slope @xmath792 .
thus between @xmath134 and @xmath135 we have a convex torus @xmath793 with dividing slope @xmath3 as above we can use this torus to show @xmath330 destabilizes .
recall if @xmath13 is the knot type of the positive @xmath7-torus knot and @xmath42 then we set @xmath794 @xmath795 , @xmath796 and @xmath797 much of theorem [ main ] was proven in the previous section . to complete the proof we need to classify legendrian knots in the @xmath0-cable of the @xmath7-torus knot type @xmath13 when @xmath115 for some @xmath798 in the next two propositions we do this first for the case when @xmath138 , and then for the case when @xmath140 .
[ prop : classify1 ] with the notation above , suppose @xmath799 for some @xmath800 then there is some @xmath116 such that @xmath801 and @xmath55 admits the following classification . 1 .
the maximal thurston - bennequin invariant is @xmath62 .
2 . for each integer
@xmath119 in the set @xmath802 there is a legendrian @xmath121 with @xmath122 3 .
there are two legendrian knots @xmath123 with @xmath803 4 .
all legendrian knots in @xmath55 destabilize to one of the @xmath128 or @xmath129 5 .
[ item5 ] let @xmath804 .
for any @xmath692 and @xmath72 the legendrian @xmath80 is not isotopic to a stabilization of any of the other maximum thurston - bennequin invariant legendrian knots in @xmath55 .
any two stabilizations of maximal thurston - bennequin invariant legendrian knots in @xmath55 , except those mentioned in item ( [ item5 ] ) , are legendrian isotopic if they have the same @xmath59 and @xmath60 .
we follow the standard approach to classifying legendrian knots in a given knot type outlined in section [ sec : pre ] .
* step i * _ identify the maximal thurston - bennequin invariant of the knot type and classify legendrian knots realizing this : _ the computation of the maximal thurston - bennequin invariant is done in lemma [ lem : maxtb ] . _
@xmath805 construction of maximal thurston - bennequin invariant knots in @xmath55 : _ let @xmath646 be the non - thickenable solid tori representing @xmath13 that were constructed in subsection [ ss : nont ] .
recall @xmath157 is a standard neighborhood of the maximal thurston - bennequin invariant legendrian @xmath7-torus knot @xmath259 ( and that there is only one @xmath157 so the @xmath501 is ignored here ) .
inside @xmath157 there are solid tori corresponding to stabilizing @xmath259 , @xmath806 times .
the range of the rotation numbers for the legendrian @xmath7-torus knots represented by these tori is @xmath807 .
denote these tori @xmath808 for @xmath809 inside each @xmath808 there are two tori @xmath810 that come from positively or negatively stabilizing the legendrian knot corresponding to @xmath808 . in the thickened torus @xmath811
there is a unique convex torus @xmath812 with dividing slope @xmath43 .
let @xmath813 where @xmath814 is the remainder .
denote by @xmath128 a legendrian divide on @xmath812 .
we clearly have that @xmath815 and the computation in the proof of lemma 3.8 in @xcite ( or similar to the one given below for @xmath69 ) gives that @xmath816 .
now consider the two tori @xmath650 .
inside each one there is a convex torus @xmath817 with dividing slope @xmath3 let @xmath69 be a legendrian divide on @xmath817 . again it is clear that @xmath818 recall that from lemma [ rcomp ] we know that @xmath819 where @xmath275 is a meridional disk for @xmath817 with legendrian boundary and @xmath253 is a surface , outside the solid torus @xmath817 that bounds , with legendrian boundary on @xmath820 if @xmath624 and @xmath307 are the corresponding surfaces for @xmath691 then we know from lemma [ candidates ] that @xmath821 and @xmath822 thus the rotation number of an @xmath0-ruling curve on @xmath691 is @xmath823 . to compute the rotation number for the legendrian divide on @xmath817 we use the classification of tight contact structures on thickened tori , as given in @xcite , and the fact that @xmath650 is universally tight . in particular
, we can compute the relative euler class @xmath664 of the thickened torus cobounded by @xmath650 and @xmath817 : @xmath824 where @xmath667 stands for the poincar dual and we are using the basis for @xmath668 given by the meridian and longitude .
we can use this to compute the difference between the rotation number of the @xmath0 curve on @xmath691 and on @xmath817 which is @xmath825 .
thus we have that @xmath826 .
_ @xmath805 classification of maximal thurston - bennequin invariant knots in @xmath55 : _ if @xmath827 with @xmath828 then @xmath330 sits on a convex torus with dividing slope @xmath43 .
theorem [ thm : partially ] and corollary [ cor : tori ] say that such a torus is one of the ones considered when constructing @xmath69 and @xmath128 . thus , a by now standard argument , see @xcite and subsection [ sssec : isotopy ] above , says the torus must be isotopic to one of the ones used in those constructions from which we can also conclude that @xmath330 is isotopic to one of @xmath69 or @xmath128 .
* step ii * _ prove all non - maximal thurston - bennequin invariant knots in @xmath55 destabilize : _
let @xmath330 be any legendrian knot in @xmath55 with thurston - bennequin invariant less than @xmath56 .
let @xmath134 be a torus bounding a solid torus @xmath141 in the knot type @xmath13 on which @xmath330 sits .
since @xmath829 we know that we can perturb @xmath134 relative to @xmath330 so that it is convex .
if the dividing slope @xmath14 of @xmath134 is equal to @xmath43 then @xmath330 intersects the dividing curves inefficiently and we can find a bypass for @xmath330 on @xmath134 .
thus we can destabilize @xmath330 .
if @xmath830 then we have three cases to consider .
case one is when @xmath831 for any @xmath25 . in this case theorem [ thm : partially ] tells us that @xmath141 can be thickened to a standard neighborhood of a maximal thurston - bennequin knot in @xmath16 .
thus there is a convex torus @xmath135 parallel to @xmath134 ( either inside @xmath141 or outside @xmath141 depending on @xmath14 ) with dividing slope @xmath3 we can use an annulus between @xmath134 and @xmath135 with boundary on @xmath330 and a legendrian divide on @xmath135 to find a bypass for @xmath832 and hence @xmath330 destabilizes .
case two is when @xmath833 for @xmath834 lemma [ lem : basicrange ] says that @xmath835 is strictly greater than @xmath836 and @xmath837 ( since @xmath14 is on the interior of @xmath838 $ ] ) .
thus there is a torus @xmath135 in @xmath141 with dividing slope @xmath839 . using an annulus between @xmath330 on @xmath134 and a @xmath43 ruling curve on @xmath135
we find a bypass for @xmath330 and hence a destabilization .
finally in case three we consider @xmath840 . in this case
we can find a torus @xmath135 as in case one to destabilize @xmath330 .
* step iii * _ determine which stabilizations of the @xmath69 and @xmath128 are legendrian isotopic : _ we first notice that exactly as in lemma 4.12 of @xcite and theorem 3.6 in @xcite we see that stabilizations of the @xmath128 are legendrian isotopic whenever they have the same thurston - bennequin invariants and rotation numbers .
( recall this is easily established by showing that when two of the @xmath128 are stabilized a minimal number of times to have the same invariants they can both be realized as a ruling curve on the boundary of a standard neighborhood of the same legendrian knot in the knot type @xmath13 . ) we will now concentrate on @xmath841 below , but analogous arguments also work for @xmath842 . from proposition [ curvesonpt1 ] we see that @xmath677 sits on a torus @xmath134 that bounds a solid torus that thickens to @xmath157 if @xmath843 .
in particular @xmath134 sits inside a solid torus @xmath141 used in the construction of one of the @xmath128 ( that is @xmath128 is a legendrian dividing curve on @xmath149 ) .
thus we may use an annulus that @xmath677 and @xmath128 cobound to see that @xmath677 destabilizes to @xmath128 .
we are left to see that @xmath677 is not isotopic to any stabilization of the other maximal thurston - bennequin invariant knots if @xmath844 but this is clear from part one of proposition [ curvesonpt1 ] since any stabilization of one of the @xmath128 or @xmath842 sits on a convex torus that does not bound a partially ( or non- ) thickenable torus contained in @xmath659 .
[ prop : classify2 ] with the notation above , suppose @xmath845 for some @xmath800 then there is some @xmath116 such that @xmath801 and @xmath55 admits the following classification . 1 . the maximal thurston - bennequin invariant is @xmath62 .
2 . for each integer
@xmath119 in the set @xmath802 there is a legendrian @xmath121 with @xmath122 3 .
there are two legendrian knots @xmath123 that do not destabilize but have @xmath846 4 .
all legendrian knots in @xmath55 destabilize to one of the @xmath128 or @xmath129 5 .
[ item52 ] let @xmath847 . for
any @xmath692 and @xmath72 the legendrian @xmath80 is not isotopic to a stabilization of any of the maximum thurston - bennequin invariant legendrian knots in @xmath55 or stabilizations of @xmath76 .
any two stabilizations of the non - destabilizable thurston - bennequin invariant legendrian knots in @xmath55 , except those mentioned in item ( [ item52 ] ) , are legendrian isotopic if they have the same @xmath59 and @xmath60 .
we follow the standard approach to classifying legendrian knots in a given knot type outlined in section [ sec : pre ] .
* step i * _ identify the maximal thurston - bennequin invariant of the knot type and classify legendrian knots realizing this : _ the computation of the maximal thurston - bennequin invariant is done in lemma [ lem : maxtb ] . _
@xmath805 construction of maximal thurston - bennequin invariant knots in @xmath55 : _ this is identical to part of the construction in proposition [ prop : classify1 ] .
let @xmath157 be a standard neighborhood of the maximal thurston - bennequin invariant legendrian @xmath7-torus knot .
inside @xmath157 there are solid tori corresponding to stabilizing @xmath259 , @xmath806 times .
the range of the rotation numbers for the legendrian @xmath7-torus knots represented by these tori is @xmath848 .
denote these tori @xmath808 for @xmath809 inside each @xmath808 there are two tori @xmath810 that come from positively or negatively stabilizing the legendrian knot corresponding to @xmath808 . in the thickened torus @xmath811
there is a unique convex torus @xmath812 with dividing slope @xmath43 .
let @xmath813 where @xmath814 is the remainder .
denote by @xmath128 a legendrian divide on @xmath812 .
we clearly have that @xmath815 and the computation in the proof of lemma 3.8 in @xcite gives that @xmath816 .
_ @xmath805 classification of maximal thurston - bennequin invariant knots in @xmath55 : _ if @xmath827 with @xmath828 then @xmath330 sits on a convex torus with dividing slope @xmath43 .
theorem [ thm : partially ] and corollary [ cor : tori ] say that such a torus is one of the ones considered when constructing the @xmath128 .
thus , a by now standard argument , see @xcite , says the torus must be isotopic to one of the ones used in those constructions from which we can also conclude that @xmath330 is isotopic to one of @xmath128 .
* step ii * _ identify and classify the non - destabilizable , non - maximal thurston - bennequin legendrian knots in @xmath55 and then show the rest destabilize to one of these or a maximal thurston - bennequin legendrian knot : _ let @xmath646 be the non - thickenable solid tori representing @xmath13 that were constructed in subsection [ ss : nont ] .
@xmath805 _ constructing the non - destabilizable legendrian knots : _ consider the two tori @xmath650 .
let @xmath69 be a ruling curve of slope @xmath0 on @xmath691 .
it is clear that the twisting of the contact planes along @xmath69 with respect to the framing of @xmath69 coming from @xmath691 is @xmath849 thus the thurston - bennequin invariant ( that is the twisting with respect to the seifert surface for @xmath69 ) is @xmath850 just as in step i of the proof of proposition [ prop : classify1 ] we compute @xmath851 @xmath805 _ proving all non - maximal thurston - bennequin invariant knots either destabilize or have @xmath852 and sit as a ruling curve on @xmath691 : _ let @xmath259 be a legendrian knot in @xmath853 with @xmath854 .
let @xmath141 be a solid torus representing the knot type @xmath13 that contains @xmath259 in its boundary .
we know that the twisting of the contact planes with respect to @xmath855 is negative so we can make @xmath149 convex without moving @xmath259 .
if @xmath259 does not intersect the dividing curves @xmath856 minimally ( for curves in their homology classes ) then we will see a bypass for @xmath259 on @xmath149 and hence @xmath259 destabilizes .
so we can assume that @xmath259 intersects @xmath857 minimally .
now if the dividing slope @xmath14 of @xmath149 is not @xmath107 then there are three cases to consider .
case one is when @xmath831 for any @xmath25 . in this case theorem [ thm : partially ] tells us that @xmath141 can be thickened to a standard neighborhood of a maximal thurston - bennequin knot in @xmath16 .
thus there is a convex torus @xmath134 parallel to @xmath149 ( either inside @xmath141 or outside @xmath141 depending on @xmath14 ) with dividing slope @xmath3 we can use an annulus between @xmath134 and @xmath149 with boundary on @xmath259 and a legendrian divide on @xmath134 to find a bypass for @xmath259 and hence @xmath259 destabilizes .
case two is when @xmath833 for @xmath834 lemma [ lem : basicrange ] says that @xmath858 is strictly greater than @xmath836 and @xmath859 ( since @xmath14 is on the interior of @xmath838 $ ] ) .
thus there is a torus @xmath134 in @xmath141 with dividing slope @xmath839 .
using an annulus between @xmath330 on @xmath134 and a @xmath43 ruling curve on @xmath134 we find a bypass for @xmath259 and hence a destabilization .
finally in case three we consider @xmath860 . in this case
we have that latexmath:[$|\frac sr \cdot t| > @xmath259 on @xmath149 and a @xmath43 ruling on @xmath691 to find a bypass for @xmath259 . if @xmath862 then @xmath259 is a ruling curve on @xmath149
. if @xmath141 is not @xmath650 then @xmath141 will thicken to @xmath157 and thus we can again destabilize @xmath259 as in case one of the previous paragraph .
so we see that @xmath259 will destabilize unless it is a ruling curve on @xmath650 . of course in this case @xmath863 .
@xmath805 _ proving the knots @xmath69 do not destabilize : _
it @xmath69 destabilized then by the above work they would be stabilizations of one of the @xmath128 .
thus @xmath69 could be put on some convex torus other than @xmath691 , but this contradicts proposition [ curvesonpt2 ] .
@xmath805 _ proving any legendrian knots with @xmath864 either destabilize or are isotopic to @xmath69 : _ this is immediate from the work above and corollary [ cor : tori ] .
* step iii * _ determine which stabilizations of the @xmath69 and @xmath128 are legendrian isotopic : _ the stabilizations of the @xmath128 all become legendrian isotopic whenever they have the same thurston - bennequin invariants as discussed in step iii of the proof of proposition [ prop : classify1 ] . from proposition [ curvesonpt2 ] we know that @xmath80 , for any @xmath655 and @xmath72 , can be put only on the convex torus @xmath691 .
thus it is clear that @xmath80 is not isotopic to any stabilization of a @xmath128 or of @xmath76 .
we also know from propositon [ curvesonpt1 ] that @xmath865 can be put on a convex torus that bounds a solid torus that thickens to @xmath157 and thus is a stabilization of the @xmath128 .
theorem [ main ] is an immediate consequence of propositions [ prop : classify1 ] and [ prop : classify2 ] together with theorems [ poscableclass ] , [ negcableclass ] and [ intermediatecableclass ] .
theorem [ qual3 ] is clear from the statement of theorem [ main ] .
theorem [ tequivaletsl ] tells us that the classification of transverse knots is equivalent to the classification of legendrian knots up to negative stabilization .
thus the theorem [ maint ] is a corollary of theorem [ main ] .
theorem [ qual4 ] follows from theorem [ maint ] once one observes that that if we choose @xmath866 ( where the addition is done as on the farey tessellation ) , then @xmath867 . as a result
, the non - destabilizable transverse knot will have self - linking number at least @xmath173 less than maximal ; furthermore , it will take @xmath868 stabilizations before it becomes isotopic to a stabilization of the maximal self - linking number transverse knot .
we are now ready to classify non - legendrian simple cables of the positive trefoil knot .
[ class23 ] let @xmath13 be the positive trefoil knot .
suppose that @xmath869 and @xmath0 is a pair of relatively prime integers such that @xmath870 .
then the @xmath0-cable of @xmath13 , @xmath22 , is not legendrian simple and legendrian knots in this knot type have the following classification . 1 .
the maximal thurston - bennequin number is @xmath62 .
2 . there are @xmath26 legendrian knots @xmath63 , with @xmath64 3 . if @xmath65 then there are two legendrian knots @xmath66 that do not destabilize but have @xmath871 4 .
all legendrian knots in @xmath55 destabilize to one of the @xmath68 or @xmath69 .
[ item511 ] let @xmath70 . for
any @xmath872 and @xmath72 the legendrian @xmath73 is not isotopic to a stabilization of any of the other @xmath74 s the @xmath75 , @xmath69 or @xmath76 . 6 . [ item512 ] let @xmath873 .
for any @xmath655 and @xmath79 the legendrian @xmath80 is not isotopic to a stabilization of any of the @xmath74 s or @xmath76 .
any two stabilizations of the @xmath81 or @xmath69 , except those mentioned in item ( [ item511 ] ) and ( [ item512 ] ) , are legendrian isotopic if they have the same @xmath59 and @xmath60 .
* step i * _ identify the maximal thurston - bennequin invariant of the knot type and classify legendrian knots realizing this : _ the computation of the maximal thurston - bennequin invariant is done in lemma [ lem : maxtb ] . _
@xmath805 construction of maximal thurston - bennequin invariant knots in @xmath55 : _ this is identical to the construction from the last section .
let @xmath157 be a standard neighborhood of the maximal thurston - bennequin invariant legendrian positive trefoil knot .
inside @xmath157 there are two solid tori @xmath874 that come from positively or negatively stabilizing the legendrian knot corresponding to @xmath157 . in the thickened torus @xmath875
there is a unique convex torus @xmath817 with dividing slope @xmath43 .
let @xmath876 be a legendrian divide on @xmath817 .
we clearly have that @xmath815 and the computation in the proof of lemma 3.8 in @xcite gives that @xmath877 .
now consider the non - thickenable tori @xmath878 from theorem [ thm : partially ] .
for @xmath879 we can find a convex torus @xmath880 with dividing slope @xmath3 let @xmath68 be a legendrian divide on @xmath880 .
again it is clear that @xmath818 recall that from lemma [ rcomp ] we know that @xmath881 where @xmath275 is a meridional disk for @xmath880 with legendrian boundary and @xmath253 is a surface outside the solid torus @xmath880 bounds with legendrian boundary on @xmath882 if @xmath624 and @xmath307 are the corresponding surfaces for @xmath883 then we know from lemma [ candidates ] that @xmath884 and @xmath822 thus the rotation number of an @xmath0-ruling curve on @xmath691 is @xmath885 . to compute the rotation number for the legendrian divide on @xmath880 we use the classification of tight contact structures on thickened tori , as given in @xcite , and the fact that @xmath878 is universally tight . in particular
, we can compute the relative euler class @xmath664 of the thicken torus cobranded by @xmath878 and @xmath817 : @xmath886 where @xmath667 stands for the poincar dual and we are using the basis for @xmath668 given by the meridian and longitude .
we can use this to compute the difference between the rotation number of the @xmath0 curve on @xmath883 and on @xmath880 which is @xmath887 .
thus we have that @xmath888 .
_ @xmath805 classification of maximal thurston - bennequin invariant knots in @xmath55 : _ if @xmath827 with @xmath828 then @xmath330 sits on a convex torus with dividing slope @xmath43 .
theorem [ thm : partially ] and corollary [ cor : tori ] say that such a torus is one of the ones considered when constructing the @xmath68 .
thus , a by now standard argument , see @xcite , says the torus must be isotopic to one of the ones used in those constructions from which we can also conclude that @xmath330 is isotopic to one of the @xmath68 . * step ii * _ identify and classify the non - destabilizable , non - maximal thurston - bennequin legendrian knots in @xmath55 and then show the rest destabilize to one of these or a maximal thurston - bennequin legendrian knot : _ let @xmath646 be the non - thickenable solid tori representing @xmath13 that were constructed in subsection [ ss : nont ] .
@xmath805 _ constructing the non - destabilizable legendrian knots : _
if @xmath889 then there are no non - destabilizable knots
. otherwise consider the two tori @xmath890 .
let @xmath69 be a ruling curve of slope @xmath0 on @xmath891 .
it is clear that the twisting of the contact planes along @xmath69 with respect to the framing of @xmath69 coming from @xmath691 is @xmath892 thus the thurston - bennequin invariant ( that is the twisting with respect to the seifert surface for @xmath69 ) is @xmath893 just as in the proof of proposition [ prop : classify1 ] we compute @xmath894 or , more simply , @xmath895 .
@xmath805 _ proving all non - maximal thurston - bennequin invariant knots either destabilize or have @xmath896 and sit as a ruling curve on @xmath891 : _ assume that @xmath897 ( since otherwise there are no - non - destabilizable knots ) .
let @xmath259 be a legendrian knot in @xmath853 with @xmath854 .
let @xmath141 be a solid torus representing the knot type @xmath13 that contains @xmath259 in its boundary .
we know that the twisting of the contact planes with respect to @xmath149 is negative so we can make @xmath149 convex without moving @xmath259 .
if @xmath259 does not intersect the dividing curves @xmath857 minimally ( for curves in their homology classes ) then we will see a bypass for @xmath259 on @xmath149 and hence @xmath259 destabilizes .
so we can assume that @xmath259 intersect @xmath856 minimally .
now if the dividing slope @xmath14 of @xmath149 is not @xmath898 then there are three cases to consider . if @xmath899 then @xmath141 thickens to @xmath157 and in particular there is a convex torus with dividing slope @xmath43 either inside or outside @xmath141 .
we may use an annulus between @xmath259 and a dividing curve on this torus to destabilize @xmath259 .
if @xmath900 , then @xmath141 contains a solid torus @xmath145 with convex boundary having infinite dividing slope .
lemma [ lem : basicrange ] guarantees that @xmath901 is greater than @xmath902 .
thus we may take a convex annulus from @xmath259 to a ruling curve on @xmath757 and use the imbalance principle to find a bypass , and hence a destabilization , for @xmath259 .
finally if @xmath903 , then there is a torus with dividing slope @xmath43 either inside or outside of @xmath141 , and we may use an annulus between @xmath259 and a dividing curve on this torus to destabilize @xmath259 . if @xmath904 then @xmath259 is a ruling curve on @xmath149
. if @xmath141 is not @xmath890 then @xmath141 will thicken to @xmath486 for some @xmath905 and thus we can again destabilize @xmath259 as in case one of the previous paragraph .
so we see that @xmath259 will destabilize unless it is a ruling curve on @xmath890 . of course in this case @xmath906 .
@xmath805 _ proving the knots @xmath69 do not destabilize : _ if @xmath69 destabilized then by the above work they would be stabilizations of one of the @xmath68 . thus @xmath69 could be put on some convex torus other than @xmath691 , but this contradicts proposition [ curvesonpt2 ] . * step iii * _ determine which stabilizations of the @xmath69 and @xmath128 are legendrian isotopic : _ the stabilizations of the @xmath128 are shown to be legendrian isotopic when they have the same classical invariants in the usual fashion as discussed in the proof of proposition [ prop : classify1 ] . from proposition [ curvesonpt1 ]
we know that @xmath73 , for any @xmath692 and @xmath72 , can be put only on the convex torus @xmath883 .
thus it is clear that @xmath80 is not isotopic to any stabilization of any of the other @xmath68 , @xmath69 or @xmath76 .
similarly if @xmath908 then for @xmath79 , proposition [ curvesonpt2 ] says that @xmath80 can only be put on the convex torus @xmath891 and hence is not isotopic to any stabilization of the @xmath68 or to @xmath76 .
we also know from propositons [ curvesonpt1 ] and [ curvesonpt2 ] that @xmath909 and @xmath910 can be put on a convex torus that bounds a solid torus that thickens to @xmath157 and thus is a stabilization of the @xmath876 .
theorem [ main23 ] simply collects the results from proposition [ class23 ] and theorems [ poscableclass ] and [ negcableclass ] . for theorem [ qual1 ]
we can choose @xmath911 .
one may easily check using theorem [ main23 ] that @xmath55 contains @xmath86 legendrian knots @xmath912 with maximal thurston - bennequin invariant ( which will be @xmath56 in this case ) and rotation number @xmath913 .
it also contains one non - destabilizable knot @xmath662 with @xmath914 and rotation number @xmath915 .
moreover , one must stabilize @xmath662 positively @xmath24 times before it becomes isotopic to a stabilization of one of the @xmath128 .
theorem [ tequivaletsl ] tells us that the classification of transverse knots is equivalent to the classification of legendrian knots up to negative stabilization .
thus the theorem [ main23 t ] is a corollary of theorem [ main23 ] . turning to theorem [ qual2 ]
we see that choices similar to those in the previous proof yield the desired result . | in this paper we classify legendrian and transverse knots in the knot types obtained from positive torus knots by cabling .
this classification allows us to demonstrate several new phenomena .
specifically , we show there are knot types that have non - destabilizable legendrian representatives whose thurston - bennequin invariant is arbitrarily far from maximal . we also exhibit legendrian knots requiring arbitrarily many stabilizations before they become legendrian isotopic .
similar new phenomena are observed for transverse knots .
to achieve these results we define and study partially thickenable " tori , which allow us to completely classify solid tori representing positive torus knots . |
low - mass x - ray binaries ( lmxbs ) consist of a neutron star ( ns ) or black hole ( bh ) accreting from a low - mass ( @xmath6 ) stellar companion via roche - lobe overflow .
they may be divided into two categories : persistent accretors with x - ray luminosity @xmath710@xmath8 erg s@xmath2 , and transient systems .
transient lmxbs undergo recurrent bright ( @xmath710@xmath8 erg s@xmath2 ) outbursts lasting days to weeks and then return to long intervals of x - ray quiescence ( @xmath910@xmath10 erg s@xmath2 ) lasting months to years .
the long - term average mass accretion rate of the transients is thus substantially lower than in the persistent systems , owing to their low duty cycle .
transient behavior is understood to arise from a thermal instability in the outer accretion disk wherein the viscosity ( and thus the mass accretion rate @xmath11 through the disk ) jumps to a higher value when a critical surface density is reached as the disk fills up ( see * ? ? ? * and references therein ) .
the persistent lmxbs avoid this instability because their higher accretion rates lead to increased x - ray heating , keeping the disks permanently ionized @xcite . for the ns systems , the 0.510 kev x - ray spectrum of quiescent lmxb transients typically consists of two components : a low - energy ( `` soft '' ) @xmath120.1 kev thermal component , and a high - energy ( `` hard '' ) power - law component with photon index @xmath13 , where photon flux @xmath14 .
the soft component is generally well fit by a hydrogen atmosphere model for the ns .
the leading explanation for the energy source of the soft component is a deep crustal heating model @xcite in which the emission is powered by heat injected into the ns crust by pycnonuclear reactions driven by accretion during transient outbursts . in this model ,
the contribution of quiescent accretion is negligible .
x - ray spectroscopy of soft thermal emission in quiescent ns transients has been used to infer ns radii @xcite and to study the thermal relaxation of ns crusts ( see * ? ? ?
* and references therein ) .
however , a possible problem for such studies is that accretion may not have completely shut off during quiescence , as suggested by the detection of quiescent variability in the two brightest quiescent ns / lmxbs , aql x-1 @xcite and cen x-4 @xcite .
there has been considerable debate as to whether this variability is primarily in the soft thermal component , the hard power - law component , or both ( e.g. , * ? ? ?
* ; * ? ? ?
the origin of the hard power - law tail is unclear .
it is not predicted by the deep crustal heating model @xcite .
two explanations have been discussed : synchrotron shock emission from a radio pulsar wind and comptonization of the soft thermal photons by a hot corona @xcite .
the synchrotron model is of particular interest given the recent confirmation that ns / lmxbs can turn on as radio pulsars at low accretion rates @xcite .
a difficulty in discriminating between different models has been the absence of knowledge about how high the power - law component extends in energy , owing to lack of sufficient observational sensitivity above 10 kev .
the recent launch of the _ nustar _ hard x - ray telescope provides the first opportunity to explore this question .
the ideal target with which to address this is cen x-4 = x1455@xmath5314 ( galactic coordinates @xmath15 , @xmath16 ) , the brightest quiescent ns / lmxb . it was discovered in 1969 in the 312 kev band with the _
vela 5a/5b _ satellites during an extremely bright ( @xmath1220 crab at peak ) x - ray outburst lasting over two months @xcite .
a second bright ( @xmath124 crab at peak ) x - ray outburst was detected in 1979 @xcite along with counterparts in the optical @xcite and radio @xcite , but the source has been in x - ray quiescence ( @xmath17 fainter ) ever since .
bright x - ray flashes , now understood as thermonuclear ( type i ) x - ray bursts , were observed around the time of both the 1969 and 1979 outbursts @xcite , conclusively establishing the source as a ns and setting an upper limit on the distance of [email protected] kpc @xcite .
a third burst may have been observed by the _ apollo 15 _ lunar mission in 1971 @xcite .
the presence of thermonuclear bursts indicates that the surface dipole magnetic field is weak , with @xmath19 g @xcite and most likely @xmath20 g ( by analogy with other type i bursters ) .
optical photometry and spectroscopy indicate that the binary companion v822 cen is a k3 - 7 v dwarf , the binary period is 15.1 hr , and the binary mass ratio is @xmath21 @xcite .
the best - fit ns mass is @xmath22 @xcite .
given the proximity and high galactic latitude of the source , it has extremely low interstellar extinction and absorption , allowing more sensitive observations in the ultraviolet and soft x - ray bands than usually possible for lmxbs @xcite .
the integrated values along this line of sight are @xmath23 @xcite and @xmath24 @xmath25 @xcite .
cen x-4 has been observed extensively during x - ray quiescence since the 1979 outburst , with deep x - ray spectra in the 0.510 kev range previously obtained on six occasions since 1994 ( see * ? ? ?
* and references therein ; see also 5.2 ) . a long - term daily monitoring campaign with _ swift
_ recently demonstrated that the thermal and power - law components vary together on time scales from days to months , with no spectral change observed and each component contributing roughly half the flux @xcite .
these authors concluded that the quiescent x - ray emission in cen x-4 is primarily generated by accretion . in this paper , we present the first sensitive hard x - ray observation of cen x-4 in quiescence with _
, obtained simultaneously with a deep _ xmm - newton _ soft x - ray observation .
we describe the observations in 2 and our spectral analysis and results in 3 .
we interpret our results in 4 and discuss their implications in 5 .
_ nustar _ , the first focusing hard x - ray observatory in orbit , was launched in 2012 and operates in the 379 kev range @xcite .
it consists of two co - aligned telescopes , and the two focal planes ( fpma and fpmb ) are each covered by a 2@xmath262 array of cadmium - zinc - telluride ( czt ) pixel detectors .
our _ nustar _ observation of cen x-4 ( obsid 30001004002 ) began on 2013 january 20 , 20:20 ut and had an elapsed duration of 219 ks , with an on - source exposure time of 114 ks . the source was imaged on detector 0 in both fpma and fpmb .
the data were processed and screened using the standard pipeline for on - axis point sources in the _ nustar _ data analysis system ( nustardas ) version 1.2.0 , along with the _ nustar _ calibration database ( caldb ) version 20130509 .
light curves and spectra from both fpma and fpmb were extracted from a circular region centered on the source position with a radius of 75 arcsec . a detailed background model for the source position in each of the two telescopes was derived by using the nuskybgd tool @xcite to fit blank sky regions covering the entire field of view for each focal plane .
for faint point sources like cen x-4 , this is more accurate than the usual method of simply scaling from the background of a nearby blank sky region because it correctly accounts for the highly non - uniform background gradients across the detectors . in both telescopes ,
the background was brighter than the source above around 20 kev .
the background - subtracted count rates in fpma and fpmb were @xmath27 count s@xmath2 and @xmath28 count s@xmath2 , respectively .
the _ nustar _
spectra were rebinned so that all but the highest energy bin had a background - subtracted significance of at least 10@xmath29 .
the highest energy bin had 5.2@xmath29 significance in fpma ( 2079 kev ) and 4.3@xmath29 significance in fpmb ( 1779 kev ) , demonstrating that the source was detected beyond 20 kev .
the _ xmm - newton _ observatory , launched in 1999 , is a focusing x - ray telescope operating in the 0.312 kev range @xcite .
our _ xmm - newton _ observation of cen x-4 ( obsid 0692790201 ) began on 2013 january 21 , 13:01 ut and had a duration of 35 ks ; this was simultaneous with part of our _ nustar _ observation .
we used the data from all three co - aligned imaging x - ray cameras ( epic - pn , epic - mos1 , and epic - mos2 ; * ? ? ?
* ; * ? ? ?
the cameras were operated in full - frame mode with the thin optical - blocking filter in place , resulting in a time resolution of 73 ms for epic - pn and 2.6 s for the mos cameras .
the data were reduced using the _ xmm - newton _ scientific analysis system ( sas ) v13.0.1 , along with the latest calibration files available as of 2013 july 13 .
we reprocessed the data using epproc and emproc to produce new event files and applied standard event filtering .
we identified background flare intervals by constructing a light curve of the 1012 kev epic - pn data using events from the entire field of view , and searching for intervals where the count rate exceeded 1 count s@xmath2 .
we found three short flares ; these intervals were removed from our cen x-4 event lists .
the net exposure time was 27ks for epic - pn and 30ks for each of the epic - mos units .
for all three detectors , we initially extracted light curves and spectra from a circular region centered on the source position with a radius of 43.5 arcsec . for the mos detectors , we measured the background using a square blank - sky region 3@xmath263 arcmin in size .
for the epic - pn detector , an important consideration is that the outer parts of the field - of - view include photons due to cu fluorescence in the instrument while the inner parts do not . as cen
x-4 was in the region without the cu emission line , we chose a rectangular region near cen x-4 for determining the epic - pn background .
the background - subtracted count rates in the 0.310 kev range were [email protected] s@xmath2 in epic - pn , [email protected] s@xmath2 in epic - mos1 , and [email protected] s@xmath2 in epic - mos2 .
these count rates are higher than the those observed in previous _ xmm - newton _ observations @xcite . in fact , the count rates in both the epic - mos and epic - pn detectors are close to the threshold where photon pileup effects may begin to distort the measured spectra , particularly during the flares .
moreover , our preliminary analysis found that the two epic - mos spectra each have significant systematic differences with the epic - pn spectrum above 2 kev .
as a precaution , we reextracted both the mos and pn data from an annular region centered on the source position with an outer radius of 43.5 arcsec and an inner radius of 10 arcsec , thus excluding the core of the point - spread function ( psf ) where any pile - up would occur ( at the expense of reduced counting statistics ) .
the mos and pn spectra from the annular region are mutually consistent .
this annulus - only data set still includes sufficient counts to obtain a good measurement of the soft x - ray spectrum .
) . [ fig_lc],scaledwidth=47.0% ] 0.16 counts s@xmath2 .
the short data gaps are intervals that were excluded owing to strong background flares .
the source intensity shows strong flaring behavior , varying by more than a factor of two on time scales as short as a few minutes . the brightest flare , around mjd 56313.8 , is also visible in the _ nustar _ light curve ( see figure [ fig_lc ] ) . _ bottom panel : _ spectral hardness of the count rate shown in the top panel , constructed by taking the ratio of the 210 kev and 0.31.0 kev count rates .
there is no evidence for significant spectral changes during the flaring intervals , although there is weak evidence for a small change during the flare at mjd 56313.63 .
[ fig_lc_xmm],scaledwidth=47.0% ] the light curves for the _ nustar _ and _ xmm - newton _ observations are shown in figure [ fig_lc ] and the top panel of figure [ fig_lc_xmm ] , respectively .
( the _ xmm - newton _ light curve uses the full data set , not the annulus - only data . )
both light curves vary significantly on time scales of a few minutes .
the _ xmm - newton _ light curve clearly exhibits flaring activity ; the strongest of these flares , around mjd 56313.8 , is also easily visible in the _ nustar _ light curve .
the fractional excess root - mean - squared ( rms ) variability @xmath30 was 37@xmath186% in the _ nustar _ light curve and 37@xmath182% in the _ xmm - newton_/epic - pn light curve . for comparison
, a value of [email protected]% was measured in 60 short _ swift _ observations made over three months @xcite .
a comparison of the 0.31 kev and 210 kev _ xmm - newton_/epic - pn light curves shows no evidence for significant spectral changes during these flares , although there is weak evidence for a small change during the flare at mjd 56313.63 ( figure [ fig_lc_xmm ] , bottom panel ; figure [ fig_hi ] ) .
a cross - correlation analysis indicates that the flares in these two bands are simultaneous to within @xmath3130 s. we searched the _ nustar _ light curve for evidence of orbital variability by folding at the 15.1 hr binary period @xcite .
no orbital modulation was detected .
-axis shows the 0.310 kev count rate , and the @xmath32-axis shows the spectral hardness ( defined as the ratio of the 210 kev and 0.31.0 kev count rates ) .
there is no evidence for significant spectral changes as a function of source intensity .
[ fig_hi],scaledwidth=47.0% ] since there is no evidence for significant spectral variability during the flares ( figure [ fig_lc_xmm ] , bottom panel ; figure [ fig_hi ] ) , we chose to include the flare intervals for our spectral analysis .
we performed joint spectral fits of the _ nustar _ and _ xmm - newton _
( annulus ) data with xspec version 12.8 spectral fitting software @xcite using @xmath33-minimization . to allow for systematic calibration differences between the different detectors
, we included a constant multiplicative factor in the model .
this constant was set to unity for _ xmm - newton_/epic - pn but allowed to vary for the other detectors .
all other spectral model parameters were tied together across the three instruments .
interstellar absorption was modeled using the tbabs model @xcite along with photoionization cross - sections from @xcite .
we tried two different models for the soft thermal component .
we first fit a passively cooling neutron star atmosphere using the nsatmos model @xcite , which assumes a pure h atmosphere and a negligible surface magnetic field ( @xmath34 g ) , and also includes the effects of surface gravity , heat conduction by electrons , and self - irradiation . in applying nsatmos , we fixed the source distance at 1.2 kpc @xcite and assumed that the atmospheric emission was coming from the entire ns surface .
we found that it was not possible to constrain the ns mass @xmath35 and radius @xmath36 when both were allowed to vary , as a wide range of @xmath35-@xmath36 pairs gave acceptable fits .
we therefore fixed the ns mass to @xmath37 @xcite .
as an alternative model , we also tried fitting to synthetic spectra of @xcite for thermal emission from unmagnetized nss with a pure h atmosphere in the presence of very low accretion rates .
these spectra have been implemented as the xspec additive table model zamp and are parametrized in terms of the observed accretion luminosity scaled to the eddington rate ( @xmath38 ) , with fixed ns mass @xmath39 and true radius @xmath40 km . in practice
, the model shapes from nsatmos and zamp are essentially identical ( see , e.g. * ? ? ? * ) , but it is useful to demonstrate that models that explicitly include accretion are consistent with the data . , scaledwidth=47.0% ] .
the fit parameters are shown in table 1 .
[ fig_spec],scaledwidth=47.0% ] spectrum of cen x-4 corresponding to the tbabs*(nsatmos+bremss ) fit shown in figure [ fig_spec ] .
the individual model components are denoted by the dotted lines .
the cutoff of the hard spectral component above 10 kev is clearly visible.[fig_uf],scaledwidth=47.0% ] although previous 0.510 kev x - ray observations of cen x-4 in quiescence were all well fit by an absorbed passive ns atmosphere plus power - law model ( see * ? ? ? * and references therein ) , the tbabs*(nsatmos+powerlaw ) model does not provide a satisfactory fit to the high - energy data in our combined _ nustar_+_xmm - newton _ data covering the 0.379 kev range .
the residuals indicate the presence of a spectral break or cutoff in the 1020 kev range ( see figure [ fig_pl ] ) .
we found several different models provided a good fit for the hard component : a cut - off power - law ( cutoffpl ) , a broken power - law ( bknpower ) , thermal comptonization ( comptt ; titarchuk 1994 ; hua & titarchuk 1994 ) , and thermal bremsstrahlung ( bremss ) . in each case , this component was combined with tbabs and either nsatmos or zamp . in all cases ,
the @xmath41 value is consistent with the integrated interstellar value along the line of sight @xcite .
we found no evidence for fluorescent fe line emission in the spectrum .
the upper limit on the equivalent width of a narrow gaussian fe emission line at 6.4 kev is @xmath0120 ev ( 90%-confidence ) .
the spectral fits with nsatmos are summarized in table 1 ; those with zamp are summarized in table 2 .
a typical fit is shown in figure [ fig_spec ] .
the corresponding unfolded spectrum is shown in figure [ fig_uf ] .
we are able to obtain reasonably good fits with a wide variety of models .
we note that the only spectral shape parameter in the zamp model is the accretion luminosity @xmath42 observed at infinity ; one must compare this with the measured flux and source distance to check for self - consistency .
we find that the best - fit values of @xmath42 are roughly consistent with the measured soft flux for the assumed distance of 1.2 kpc .
using the tbabs*(nsatmos+bremss ) model , we find an average ( absorbed ) flux of @xmath43 erg @xmath25 s@xmath2 in the soft ( 0.33 kev ) band and @xmath44 erg @xmath25 s@xmath2 in the hard ( 379 kev ) band .
the unabsorbed soft luminosity was @xmath45 erg s@xmath2 ( 0.33 kev ) , the unabsorbed hard luminosity was @xmath46 erg s@xmath2 ( 379 kev ) , and the total unabsorbed luminosity was @xmath47 erg s@xmath2 ( 0.379 kev ) , where @xmath48 is the source distance in kiloparsecs .
for comparison with previous observations , the 0.510 kev unabsorbed luminosity was @xmath49 erg s@xmath2 .
this is the highest 0.510 kev luminosity ever measured from cen x-4 in quiescence .
the next highest observation was 2.6 times fainter in 2001 @xcite , and the faintest observation was 11.5 times fainter in 2009 @xcite .
the thermal ( ns atmosphere ) contribution to the 0.510 kev luminosity on our observation is 59% .
this is consistent with previous observations , where the thermal fraction has always been 5060% @xcite .
lcccccccc + absorption column & @xmath50 & @xmath51 @xmath25 & 0.88(4 ) & 0.89(3 ) & 0.92(4 ) & 0.92(4 ) & 0.87(5 ) + temperature ( unredshifted ) & @xmath52 & k & 6.16(4 ) & 6.18(1 ) & 6.20(27 ) & 6.19(23 ) & 6.20(10 ) + mass ( fixed ) & @xmath35 & @xmath53 & 1.9 & 1.9 & 1.9 & 1.9 & 1.9 + radius & @xmath36 & km & 9.2(1.0 ) & 8.6(2.6 ) & 8.5(5.7 ) & 8.6(4.9 ) & 8.1(1.8 ) + covering fraction ( fixed ) & @xmath54 & @xmath55 & 1 & 1 & 1 & 1 & 1 + distance ( fixed ) & @xmath56 & kpc & 1.2 & 1.2 & 1.2 & 1.2 & 1.2 + + + normalization & @xmath57 & @xmath58 & 2.3(2 ) & 2.2(2 ) & 0.7(1 ) & 0.7(1 ) & 4.1(1 ) + photon index & @xmath59 & @xmath55 & 1.02(10 ) & 1.26(10 ) & @xmath55 & @xmath55 & @xmath55 + cutoff or break energy & @xmath60 , @xmath61 & kev & 10.4(1.4 ) & 5.8(3 ) & @xmath55 & @xmath55 & @xmath55 + photon index 2 & @xmath62 & @xmath55 & @xmath55 & 2.02(5 ) & @xmath55 & @xmath55 & @xmath55 & + electron temperature & @xmath4 & kev & @xmath55 & @xmath55 & 6.4(1.1 ) & 6.4(1.1 ) & 18.2(1.0 ) + seed temperature & @xmath63 & kev & @xmath55 & @xmath55 & 0.74(8 ) & 0.74(7 ) & @xmath55 + scattering optical depth & @xmath64 & @xmath55 & @xmath55 & @xmath55 & 4.0(5 ) & 8.8(1.0 ) & @xmath55 + + + _ nustar_/fpma & @xmath65 & @xmath55 & 1.04(3 ) & 1.03(3 ) & 1.02(3 ) & 1.02(3 ) & 1.09(3 ) + _ nustar_/fpmb & @xmath66 & @xmath55 & 1.07(4 ) & 1.05(4 ) & 1.05(4 ) & 1.05(4 ) & 1.12(4 ) + _ xmm_/epic - pn ( fixed ) & @xmath67 & @xmath55 & 1 & 1 & 1 & 1 & 1 + _ xmm_/epic - mos1 & @xmath68 & @xmath55 & 1.04(1 ) & 1.04(1 ) & 1.04(1 ) & 1.03(1 ) &
1.04(1 ) + _ xmm_/epic - mos2 & @xmath69 & @xmath55 & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) + + fit statistic & @xmath70/dof & @xmath55 & 1.155/380 & 1.123/379 & 1.130/379 & 1.129/379 & 1.167/381 lcccccccc + absorption column & @xmath50 & @xmath51 @xmath25 & 0.89(3 ) & 0.87(6 ) & 0.69(5 ) & 0.69(5 ) & 0.93(5 ) + luminosity & @xmath71 & @xmath55 & @xmath55.15(3 ) & @xmath55.13(4 ) & @xmath54.98(3 ) & @xmath54.98(3 ) & @xmath55.19(3 ) + mass ( fixed ) & @xmath35 & @xmath53 & 1.4 & 1.4 & 1.4 & 1.4 & 1.4 + radius ( fixed ) & @xmath36 & km & 12.4 & 12.4 & 12.4 & 12.4 & 12.4 + normalization & @xmath72 & @xmath73 & 1.25(2 ) & 1.25(2 ) & 1.39(2 ) & 1.39(2 ) & 1.23(2 ) + + + normalization & @xmath57 & @xmath58 & 2.5(2 ) & 2.3(2 ) & 0.6(1 ) & 0.6(1 ) & 4.2(1 ) + photon index & @xmath59 & @xmath55 & 1.08(9 ) & 1.29(7 ) & @xmath55 & @xmath55 & @xmath55 + cutoff or break energy & @xmath60 , @xmath61 & kev & 11.2(1.5 ) & 5.9(3 ) & @xmath55 & @xmath55 & @xmath55 + photon index 2 & @xmath62 & @xmath55 & @xmath55 & 2.03(5 ) & @xmath55 & @xmath55 & @xmath55 & + electron temperature & @xmath4 & kev & @xmath55 & @xmath55 & 7.3(1.6 ) & 7.6(1.7 ) & 17.9(1.0 ) + seed temperature & @xmath63 & kev & @xmath55 & @xmath55 & 0.81(4 ) & 0.81(4 ) & @xmath55 + scattering optical depth & @xmath64 & @xmath55 & @xmath55 & @xmath55 & 3.6(6 ) & 7.8(1.2 ) & @xmath55 + + + _ nustar_/fpma & @xmath65 & @xmath55 & 1.04(3 ) & 1.03(3 ) & 1.01(3 ) & 1.01(3 ) & 1.08(3 ) + _ nustar_/fpmb & @xmath66 & @xmath55 & 1.07(3 ) & 1.05(4 ) & 1.04(4 ) & 1.04(4 ) & 1.10(4 ) + _ xmm_/epic - pn ( fixed ) & @xmath67 & @xmath55 & 1 & 1 & 1 & 1 & 1 + _ xmm_/epic - mos1 & @xmath68 & @xmath55 & 1.04(1 ) & 1.04(1 ) & 1.03(1 ) & 1.03(1 ) & 1.03(1 ) + _ xmm_/epic - mos2 & @xmath69 & @xmath55 & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) & 0.97(1 ) + + fit statistic & @xmath70/dof & @xmath55 & 1.156/380 & 1.129/379 & 1.109/379 & 1.109/379 & 1.162/381
we now discuss what can be inferred from our observed high - energy spectral cutoff .
we introduce dimensionless variables for parametrizing the ns mass @xmath74 and radius @xmath75 km .
it is also convenient to scale the luminosity and accretion rate to the eddington critical values , @xmath76 where @xmath77 is the hydrogen mass fraction for cosmic abundances .
our observed luminosities can then be written as @xmath78 , @xmath79 , and @xmath80 .
it is clear that the quiescent mass accretion rate @xmath81 onto the ns must also be very low .
we can set an upper limit by assigning all of the observed x - ray luminosity to accretion , @xmath82 .
we then have @xmath83 . at such extremely low accretion rates ,
the outer accretion disk is likely to transition into a quasi - spherical , radiatively - inefficient accretion flow ( riaf ) at a transition radius @xmath84 cm , where @xmath85 km is the schwarzschild radius @xcite . one way to explain a cutoff power - law spectrum is thermal comptonization , where soft thermal seed photons ( like those from the ns atmosphere ) are compton scattered from a corona of hot electrons . @xcite
have previously argued that comptonization of the ns atmosphere photons can not account for the 110 kev power law observed in cen x-4 because the observed luminosity in the power - law component is too high relative to the soft luminosity .
we show here that a thermal comptonization model for the hard spectral cutoff in cen x-4 is not physically self - consistent , despite providing an acceptable fit to our observed spectrum . for simplicity ,
we consider the spherical geometry case ; the results for a disk geometry are qualitatively similar .
our comptt fits yielded an electron temperature of @xmath868 kev and an electron scattering optical depth of @xmath879 for a spherical geometry ( see tables 1 and 2 ) .
this is an unusually low @xmath4 for a comptonized plasma , reflecting the observed spectral cutoff .
it is likewise an unusually high @xmath64 ; this follows from the fact that @xmath4 and @xmath64 are inversely proportional for a fixed power - law index @xmath59 @xcite . the optical depth in a corona of radius @xmath88 is related to the electron density @xmath89 by @xmath90 where @xmath91 is the thomson cross - section .
as we see below , it is our high value of @xmath64 that makes it difficult to find a physically self - consistent comptonization model .
we begin by considering the possibility of a hot , optically thin layer above the ns atmosphere . based on our measured @xmath64 ,
the electron density in the layer would be @xmath92 @xmath93 , where @xmath94 cm is the thickness of the layer .
this relatively dense scattering layer would itself be a source of significant thermal bremsstrahlung emission , with @xmath95 kev . despite the large scattering optical depth @xmath64 ,
the _ effective _ optical depth of the medium at 2 kev ( including both scattering and free - free absorption ) is only @xmath96 @xcite , where @xmath97 is the free - free absorption optical depth at 2 kev .
( at higher energies , @xmath98 will be even smaller . )
we can thus use the emissivity for optically thin thermal bremsstrahlung to compute the expected luminosity from the dense scattering layer , @xmath99 erg s@xmath2 @xmath100 .
this is orders of magnitude brighter than what we observe and can be ruled out .
we also consider a corona around the ns magnetosphere . in x - ray quiescence , we can scale the magnetospheric radius @xmath101 to the corotation radius ( where the keplerian and stellar angular velocities are equal ) , which is given by @xmath102 where we have written the ( unknown ) ns spin period as @xmath103 ms .
we assume a spherically symmetric corona with inner radius @xmath104 and scale size @xmath105 .
then , following the calculations in the previous paragraph , we find an electron density @xmath106 @xmath93 , an effective optical depth of @xmath107 at 2 kev , and a predicted bremsstrahlung luminosity from the scattering corona of @xmath108 erg s@xmath2 @xmath109 .
this is , again , orders of magnitude brighter than what we observe and can be ruled out .
we next consider comptonization in a riaf - like spherical accretion flow with a radially uniform mass inflow rate @xmath11 at infall velocity @xmath110 ( where @xmath111 ) .
the electron density will vary with distance @xmath112 from the ns as @xmath113 where @xmath114 is the mean molecular weight and @xmath115 is the proton mass .
we then find that the optical depth can not exceed @xmath116 independent of @xmath88 .
this is orders of magnitude smaller than our comptt fit values for @xmath64 .
the only way to obtain higher optical depths is to assume that the mass inflow rate @xmath11 varies with @xmath112 in such a way that only a small amount of mass actually reaches the ns , with the remainder being expelled in some sort of outflow . for convenience ,
we parametrize @xmath11 as a power - law in @xmath112 @xmath117 where @xmath36 is the ns radius , @xmath118 is the mass accretion rate onto the ns , and @xmath119 .
( the @xmath120 case corresponds to the radially uniform inflow rate that we just dismissed . )
this is the same parameterization used in the `` adios '' ( adiabatic inflow - outflow solution ) model of @xcite .
the electron density now varies as @xmath121 and the optical depth is @xmath122 for ionized gas with cosmic abundances , we have @xmath123 and @xmath77 .
we require that @xmath124 , where the binary separation is @xmath125 cm @xcite .
this is equivalent to requiring that @xmath126 .
then taking @xmath127 and @xmath128 , we find @xmath129 for @xmath130 .
even larger @xmath131-values are required for smaller coronas .
such high @xmath131-values yield high electron densities far from the ns , so that the scattering cloud would again itself become a source of significant thermal bremsstrahlung emission . as an example
, we consider the @xmath132 ( uniform density ) case , for which @xmath133 @xmath93 .
the effective optical depth is @xmath134 , so we expect optically thin bremsstrahlung emission .
taking @xmath135 and @xmath136 kev , we predict a bremsstrahlung emission measure @xmath137 @xmath93 , which corresponds to a luminosity of @xmath138 erg s@xmath2 @xmath139 .
this is nearly 7 orders of magnitude brighter than the observed quiescent emission !
we conclude that a thermal comptonization model is not physically self - consistent for the hard spectral component in cen x-4 in quiescence and can be ruled out .
another way of producing a cut - off power - law spectrum is through synchrotron emission .
radio pulsars dissipate their rotational energy via a relativistic wind comprising charged particles and poynting flux ( see * ? ? ?
* ; * ? ? ?
* and references therein ) .
quiescent ns / lmxb transients can turn on as radio pulsars under some circumstances @xcite ; this has recently been observed in at least three cases ( see [ sec - comps ] ) .
if cen x-4 turned on as a radio pulsar during x - ray quiescence , then synchrotron shock emission from the radio pulsar wind interacting with intrabinary material or the interstellar medium ( ism ) could produce a power - law x - ray spectrum @xcite .
the synchrotron power - law spectrum will have a high - energy exponential cutoff corresponding to the maximum energy of the electron population accelerated in the shock .
this is generally thought to occur at energies @xmath140100 kev in pulsar wind shocks .
indeed psr j1023 + 0023 , the only quiescent ns / lmxb in which synchrotron shock emission is definitely thought to have been observed , has an unbroken power - law spectrum out to at least 80 kev is observed @xcite . however , as we show below , it is possible to have lower cutoff energies in pulsar wind shocks .
we begin by reviewing the formalism for the synchrotron shock scenario developed by ( * ? ? ?
* hereinafter at93 ) for the case of the eclipsing `` black widow '' millisecond pulsar psr b1957 + 20 .
the pulsar wind is powered by spin - down of the pulsar with an energy loss rate set by magnetic dipole radiation , @xmath141 where @xmath142 and @xmath143 are the pulsar spin period and its derivative is the spin period derivative due to magnetic dipole spin - down alone , in the absence of any accretion torques . ] , and we assume @xmath144 g @xmath145 for the ns moment of inertia . since we do not know @xmath142 and @xmath143 for cen x-4 , we will scale the surface dipole magnetic field as @xmath146 g and the spin period as @xmath147 ms , where we assume @xmath148 thus , for a millisecond pulsar , we have @xmath149 and @xmath150 erg s@xmath2 .
upstream from the shock , the lorentz factor in the wind is @xmath151 where we assume that the ions in the wind have charge @xmath152 and mass @xmath153 and are accelerated to a fraction @xmath154 of the open field line voltage of the ns .
possible ion values range from protons ( @xmath155 ) to partially - ionized iron ( @xmath156 , @xmath157 ) .
the relative energetic contribution of poynting flux and ions in the wind upstream of the shock is described by a magnetization parameter @xcite , @xmath158 where @xmath159 and @xmath160 are the magnetic field strength and density of the upstream wind . for a particle - dominated wind ( like the one in the crab nebula ) , @xmath161 , while for a magnetically dominated wind , @xmath162 .
the upstream magnetic field strength can then be written as @xcite @xmath163 where @xmath164 is the fractional solid angle into which the wind is emitted . from the shock jump conditions ,
the downstream field strength is @xcite @xmath165 the shock will give rise to a power - law electron population with energy distribution @xmath166 for @xmath167 , with @xmath168 .
this will , in turn , produce a synchrotron radiation spectrum with photon number index @xmath169 for photon energies above @xmath170 where @xmath171 and @xmath172 are the charge and mass of the electron .
if radiative losses are negligible , these power laws will extend up to a cutoff at lorentz factor @xmath173 and at photon energy @xmath174 however , if radiative losses are significant , then the electron population will extend only to @xmath175 . to find @xmath176
, we compare the acceleration time ( at93 ) @xmath177 and the synchrotron loss time @xmath178 and solve @xmath179 .
if @xmath175 , then radiative losses are important , and the photon power law will only extend up to @xmath180 given the above formalism , we now consider two possible sites for the pulsar wind shock location .
if the shock occurs where the pulsar wind is confined by ram pressure in the ism , then the shock radius will be ( * ? ? ?
* at93 ) @xmath181 where @xmath182 @xmath93 is the ism particle density and @xmath183 kms@xmath2 is the pulsar space velocity @xcite . at this large distance from the ns
, we assume that @xmath184 @xcite . then , the upstream magnetic field strength is @xmath185 and the magnetic field downstream of the shock is @xmath186 .
radiative losses are negligible in such a weak field , so the expected synchrotron power - law spectrum will extend from @xmath187 to a cutoff at @xmath188 the ism shock model produces a cutoff energy consistent with our observed spectrum for a reasonable range of pulsar parameters , as shown in shaded region ( a ) of figure [ fig_sync ] . however , the predicted synchrotron emission corresponding to that region is orders of magnitude weaker than our observed power - law luminosity ( @xmath189 erg s@xmath2 ) . scaling to @xmath142 and @xmath190
values that lie in the central strip of region ( a ) in figure [ fig_sync ] , the expected 0.320 kev synchrotron luminosity from the ism shock is ( at93 , equation [ 16 ] ) @xmath191 where @xmath192 is the conversion efficiency of pulsar wind luminosity into particle acceleration in the shock , and @xmath193 is the fraction of the bolometric synchrotron luminosity that lies in the 0.320 kev band , @xmath194^{1/2 } - \left[\frac{\mbox{0.3 kev}}{e_c}\right]^{1/2 } .\ ] ] note that contours of constant @xmath60 and constant @xmath195 have the same slope in figure [ fig_sync ] , so the numerical pre - factor in equation ( [ eq - synclum_ism ] ) characterizes the entire length of the shaded region .
we conclude that synchrotron emission from a shock in the ism can not explain the observed hard x - ray spectrum in cen x-4 . for a synchrotron shock spectrum , displayed as a function of the radio pulsar spin period @xmath142 and period derivative @xmath143 .
@xmath142 and @xmath143 are not known for cen x-4 .
the dotted lines show contours of constant pulsar magnetic field strength @xmath190 ( see equation [ [ eq : pulsarb ] ] ) .
the shaded regions have @xmath60 in the 520 kev range , consistent with the observed power - law cutoff energy in cen x-4 , and correspond to the three cases listed in the box on the upper left .
radiative losses are negligible in all these regions ( @xmath196 )
. contours of constant @xmath60 and constant synchrotron luminosity @xmath195 have the same slope on the @xmath142-@xmath143 plane . in all three regions ,
the predicted @xmath195 is orders of magnitude too small to explain the observed power - law luminosity ( @xmath197 erg s@xmath2 ) in cen x-4 .
additional regions , corresponding to the intrabinary shock cases where radiative losses dominate ( @xmath175 ) , are not shown ; they lie in an unphysical part of the phase space ( @xmath198 ms and @xmath199 g ) for @xmath200 kev .
all calculations were done with @xmath201 , @xmath202 , and @xmath203 .
[ fig_sync],scaledwidth=47.0% ] another possibility is that the pulsar wind collides with material lost from the binary companion or the accretion flow at a shock radius @xmath204 , where the magnetic field is considerably stronger than in the ism case . in psr j1023
+ 0038 , the intrabinary shock was modeled as occurring near the inner lagrangian ( @xmath205 ) point @xcite ; in cen x-4 , the @xmath205 point lies at @xmath206 .
it is unclear whether the pulsar wind would be particle - dominated or not in an intrabinary shock relatively close to the pulsar .
@xcite points out that theory predicts @xmath162 at the pulsar light cylinder , but that observations indicate @xmath184 in observed pulsar wind nebulae produced by ism shocks , and that it is not clear how or where the transition from one regime to the other occurs .
we note that @xcite inferred @xmath207 for the intrabinary shock in psr j1023 + 0038 .
we therefore consider both the @xmath208 and @xmath207 cases here .
@xcite wrote : `` large-@xmath29 shocks are effectively weak ... only when @xmath209 can a significant fraction of the total energy flux upstream be converted into thermal energy downstream and thereafter into synchrotron luminosity . '' ] . scaling to the binary separation in equation ( [ eq : b1 ] ) ,
the upstream magnetic field is now @xmath210 where @xmath211 cm .
the downstream field @xmath212 is given by equation ( [ eq : b2 ] ) . from equation ( [ eq : emin ] ) ,
the lower energy bound for the synchrotron power law is @xmath213 the upper bound depends on whether or not radiative losses are important , which is set by the ratio @xmath214 from equations ( [ eq : ecm ] ) and ( [ eq : ecs ] ) , the cutoff energy for the @xmath184 case is then @xmath215 and for the @xmath162 case is @xmath216 the intrabinary shock model with @xmath196 is able to produce a cutoff energy consistent with our observed spectrum for a reasonable range of pulsar parameters , as shown in shaded regions ( b ) and ( c ) of figure [ fig_sync ] .
however , again , the predicted synchrotron luminosity corresponding to those regions is orders of magnitude smaller than our observed power - law luminosity .
the expected 0.320 kev synchrotron luminosity from an intrabinary shock is ( at93 , equation [ 24 ] ) @xmath217 where @xmath218 is the fraction of the pulsar wind intercepted by intrabinary material . by analogy to psr j1023
+ 0038 , we expect @xmath219 to be in the range 0.010.1 @xcite . scaling to @xmath142 and @xmath190 values inside the shaded regions , we thus have @xmath220 for the @xmath196 , @xmath184 case [ region ( b ) ] , and @xmath221 for the @xmath196 , @xmath162 case [ region ( c ) ] .
the cases where radiative losses dominate ( @xmath175 ) only give low enough cutoff energies for an unphysical set of pulsar parameters ( @xmath198 ms with @xmath199 g , beyond the upper left corner of figure [ fig_sync ] ) , and they predict a synchrotron luminosity at least two orders of magnitude _ larger _ than what we observe .
taken together , we conclude that synchrotron emission from an intrabinary shock can not explain the cutoff power - law spectrum in cen x-4 .
a third way of modeling our cut - off spectrum is through thermal bremsstrahlung emission from an optically thin cloud of hot electrons .
our bremsstrahlung fits ( see tables 1 and 2 ) yielded an electron temperature @xmath22218 kev and an emission measure @xmath223 we examine two different possibilities for the emission site .
we again consider the possibility of a hot , optically thin layer above the ns atmosphere with geometric thickness @xmath94 cm .
the observed emission measure requires an electron density of @xmath224 this implies a free - free absorption optical depth of only @xmath225 at 2 kev , which is self - consistent for optically - thin thermal bremsstrahlung emission . in this scenario , both the soft ( thermal ) emission and the hard ( bremsstrahlung ) emission are formed in or above the ns atmosphere .
this can thus easily account for the fact that the soft and hard emission are observed to vary together on short time scales , with no detectable time lag .
@xcite calculated the spectrum of an unmagnetized ns atmosphere accreting at low rates through a riaf flow .
their work differs from the low-@xmath11 case considered by ( * ? ? ?
* see also 3.2 ) in that it includes coulomb heating of the atmosphere by energetic protons in the riaf flow .
interestingly , they _ predict _ a hot , optically thin surface layer above the ns atmosphere with @xmath226 kev and with significant bremsstrahlung emission expected for a certain range of @xmath81 , in addition to the soft thermal emission from the ns surface .
their most detailed calculations were made assuming a proton temperature of @xmath227 , where @xmath228 is the virial temperature .
for this case , they found that the bremsstrahlung emission is comparable in strength to the thermal component when @xmath229 , but that it becomes negligible when @xmath230 . by comparison ,
our observation of cen x-4 measured comparable luminosity in the the two components when the accretion rate was much lower , @xmath231 .
this is a factor of @xmath232 discrepancy with the @xcite calculation .
one way of reconciling this is to invoke magnetic channeling of the quiescent accretion flow onto the ns polar caps , thus increasing the local accretion rate per unit area .
however , this would require very small polar caps with area @xmath233 .
the measured spectral parameters of the soft thermal component in cen x-4 are not consistent with such small polar caps ( see tables 1 and 2 ) . on the contrary ,
they suggest that the thermal emission arises from a significant fraction of the stellar surface . another alternative is to reexamine the assumptions of the @xcite calculation , since the expected bremsstrahlung luminosity must depend on the details of the coulomb heating of the atmosphere .
the authors found that , for fixed accretion rate , reducing the proton temperature increases the heating in the upper atmosphere , since the protons do not penetrate as deeply .
however , they did not investigate this case in detail , so it is not clear whether a physically reasonable choice of proton temperature can produce significant bremsstrahlung emission at accretion rates as low as we observed in cen x-4 .
we conclude that bremsstrahlung emission from a hot layer above the ns atmosphere is consistent with our observed spectral cutoff .
however , it is not yet clear how to produce sufficient luminosity to match our data based on existing theoretical models for riaf accretion onto nss . as another alternative
, we again assume that @xmath11 varies with distance from the ns according to equation ( [ eq : bb99 ] ) and use equation ( [ eq : n_er ] ) for @xmath89 .
then , taking @xmath123 and @xmath77 for ionized gas with cosmic abundances , the expected emission measure for a cloud of radius @xmath88 is @xmath234 equating this to equation ( [ eq : bremss_em ] ) , we find that @xmath235 which sets the scale for the bremsstrahlung emission region in units of the ns radius @xmath36 .
requiring @xmath124 or equivalently @xmath236 , we find that @xmath237 . alternatively ,
if we require that the cloud lies inside the riaf transition radius ( @xmath238 or @xmath239 ) , then we find that @xmath240 . in either case , only a small fraction @xmath241 of the mass transferred reaches the ns in quiescence ; the rest accumulates at large @xmath112 or is expelled .
in particular , we note that much of the bremsstrahlung emission is coming from electrons at large @xmath112 . for @xmath242 ,
the electron density @xmath243 ranges from @xmath244 @xmath93 near the ns to @xmath245 @xmath93 near @xmath88 .
this scenario is able to self - consistently account for the observed bremsstrahlung luminosity .
however , it is challenging to explain the short ( @xmath3130 s ) lag time observed between the soft and hard flux variability in the x - ray light curve ( see 3.1 ) .
presumably , the hard flares would arise from mass fluctuations in the riaf flow , while the soft flares would arise when those fluctuations reach the ns surface .
the shortest possible time scale between the soft and hard flares is then the free - fall time scale from @xmath88 , @xmath246 this is only marginally consistent with our upper limit on the lag time .
we conclude that bremsstrahlung emission from the riaf flow is consistent with our observed spectral cutoff and luminosity , but that placing this emission far from the ns is difficult to reconcile with the fact that the soft and hard emission vary together on short time scales .
we have measured a cutoff in the hard x - ray power - law spectrum of cen x-4 which can be fit with an exponential cutoff at around 10 kev or a bremsstrahlung spectrum with @xmath247 kev .
this is the first detection of a power - law cutoff in this source class , and it finally permits a more detailed investigation of the origin of the hard component in quiescent ns / lmxbs .
we were able to rule out both thermal comptonization and synchrotron shock emission as the origin of the spectral cutoff . instead
, the hard x - ray spectrum can be understood as bremsstrahlung emission , arising either from a hot , optically thin corona above the ns atmosphere or from hot electrons in an optically thin riaf .
the ns atmosphere scenario has the advantage that it can easily explain why the soft and hard emission varies together on short time scales , while the riaf scenario has the advantage that it can self - consistently account for the observed luminosity .
the 18 kev electron temperature for the bremss model is much lower than either the @xmath1250 kev electron temperature predicted for the hot layer above a ns atmosphere @xcite or the @xmath140100 kev electron temperature expected in a riaf flow around a black hole @xcite
. this may be due to compton cooling of the bremsstrahlung electrons by the soft x - ray photons from the ns atmosphere , in which case we would expect @xmath248 to depend upon the soft x - ray luminosity @xmath249 .
the absence of a detectable compton emission component is not problematic .
the compton radiative power density is @xmath250 the resulting inverse compton luminosity is @xmath251 for cooling above the ns atmosphere , or @xmath252 for cooling in the accretion flow . in either case , the compton luminosity is no more than a few percent of the overall source luminosity , and thus essentially undetectable in our spectrum .
it is interesting to consider whether the flares observed in the x - ray light curve might be expected to affect the electron temperature ( and thus the cutoff energy ) .
the compton cooling time scale is @xmath253 . for emission above the ns atmosphere , this yields @xmath254 so the cooling is nearly instantaneous . for this scenario
, we might expect to measure spectral changes in the _ nustar _ band ( @xmath14010 kev ) during the flares , although we did not have sufficient signal - to - noise to do this with our observation .
by contrast , emission in the riaf gives @xmath255 so that in this case short - term flaring behavior will not result in significant compton cooling of the bremsstrahlung electrons in the riaf flow .
this might provide an avenue for discriminating between the two scenarios . on longer time scales ( months to years ) , the fact that the slope of the hard x - ray power - law spectrum in cen x-4 was observed to vary from epoch to epoch over the course of two decades @xcite makes it unlikely that the 18 kev bremsstrahlung spectrum we have measured is a constant feature of the source in quiescence .
in fact , we can demonstrate that the hard x - ray cutoff energy in cen x-4 is likely variable by noting that the departure of the hard x - ray spectrum from an unbroken power - law above @xmath36 kev is evident in our 2013 _ xmm - newton _ spectrum alone , even without including the _ nustar _ data .
the shape of an 18 kev bremsstrahlung spectrum will show noticeable curvature below 10 kev . on the other hand ,
all previous deep observations of cen x-4 in the 0.510 kev band are consistent with an unbroken hard x - ray power - law spectrum , indicating a higher cutoff energy for those observations .
all of these observations occurred at significantly lower luminosity ( see figure [ fig_gamma ] ) .
we would expect a lower thermal luminosity to result in reduced compton cooling of the bremsstrahlung electrons and hence a higher electron temperature , consistent with a higher cutoff energy .
llcccc 1994 feb 27 & _ asca _ & 39 & 1.19(11 ) & 1.24(17 ) & 1 + 2001 aug 20 & _ xmm _ & 53 & 1.50(5 ) & 1.41(5 ) & 1 + 2003 mar 1 & _ xmm _ & 78 & 1.07(2 ) & 1.26(8 ) & 1 + 2009 jan 16 & _ suzaku _ & 147 & 0.29(2 ) & 1.69(17 ) & 1 + 2010 aug 25 & _ xmm _ & 21 & 0.63(6 ) & 1.77(21 ) & 2 + 2010 sep 4 & _ xmm _ & 23 & 0.67(1 ) & 1.62(10 ) & 2 + 2011 jan 24 & _ xmm _ & 15 & 0.97(2 ) & 1.38(10 ) & 2 + 2011 jan 31 & _ xmm _ & 14 & 0.31(1 ) & 1.94(19 ) & 2 + 2013 jan 20 & _
xmm_+_nustar _ & 27/114 & 3.8(1 ) & 1.56(5 ) & 3 ) .
the _ xmm_/_nustar _
point is from a tbabs*(nsatmos+powerlaw ) fit to the 0.310 kev data only .
the thermal luminosity @xmath256 is computed for a distance of 1 kpc . for @xmath257 erg s@xmath2 ,
the data are roughly consistent with a trend of steeper power - law slope for higher @xmath256 as expected for bremsstrahlung emission , if we assume that @xmath4 is reduced from 50100 kev via compton cooling by thermal photons . in energy bandpasses well below @xmath4 ,
a bremsstrahlung spectrum is a @xmath258 power law .
the large @xmath59 measured at the lowest thermal luminosities may indicate a transition to synchrotron shock emission at extremely low @xmath11 .
[ fig_gamma],scaledwidth=47.0% ] moreover , we would expect the 210 kev power - law slope to vary systematically with the thermal luminosity @xmath256 . for energy bandpasses far below @xmath4 ,
a bremsstrahlung spectrum is a @xmath258 power - law ( see , e.g. * ? ? ?
* ) ; as one approaches the cutoff at @xmath4 from below , the effective @xmath59 over a fixed bandpass increases as the spectrum begins to gradually roll over .
thus , if we assume that @xmath259 50100 kev in the absence of compton cooling , then a compton - cooled bremsstrahlung model predicts that @xmath59 should be close to 1 at low @xmath256 and should increase as @xmath256 rises and @xmath4 falls .
this relationship will eventually break down when @xmath4 gets sufficiently low , because an unbroken power - law no longer provides even a rough fit to a sharp spectral cutoff .
the archival data roughly support this picture for @xmath257 erg s@xmath2 . in figure [ fig_gamma
] , we plot @xmath59 versus @xmath256 for deep quiescent observations of cen x-4 made since 1994 , including our observation values found by @xcite for these observations are much smaller than for any other observations , as would be expected if there is significant pileup .
we will reexamine these data elsewhere . ] .
these observations are listed in table [ tbl-3 ] ; the archival spectra are collected from @xcite and @xcite .
the trend of the observations in figure [ fig_gamma ] with @xmath257 erg s@xmath2 is roughly consistent with our expectation for a compton - cooled bremsstrahlung model .
at the lowest thermal lumnosities , however , @xmath59 jumps to higher values .
the abrupt change is suggestive of a spectral transition to a different emission mechanism .
given the extremely low luminosities , one might consider coronal x - ray emission from the companion star @xcite , but the observed spectral shapes for these observations were not consistent with coronal emission @xcite .
instead , we suggest that this may indicate a transition to synchrotron shock emission at the lowest luminosities .
we discuss this further in [ sec - comps ] . as in most quiescent transient ns / lmxbs ,
a basic requirement for cen x-4 is that most of the accretion flow does not reach the ns , since the inferred @xmath81 is substantially smaller than the binary mass transfer rate @xmath260 expected for a roche - lobe filing main sequence donor in a 15.1 hr binary ( e.g. , * ? ? ?
. there may be several mechanisms that contribute to this .
first of all , the disk instability model for lmxb transients predicts that most of the accretion flow during x - ray quiescence builds up in the outer accretion disk until a thermal instability ensues , causing an outburst ( see * ? ? ? * ) . at low @xmath11 , the outer disk will transition into a quasi - spherical riaf flow at @xmath261 .
it has previously been noted that riaf models for quiescent ns transients require that most of the riaf flow is somehow prevented from reaching the ns @xcite .
one way of achieving this is the adios - like outflow that we discussed in [ sec - bremss2 ] .
another possibility is that most of the flow reaching the ns magnetosphere is centrifugally inhibited by the magnetic `` propeller effect '' @xcite .
this occurs when the magnetosphere extends beyond the corotation radius ( see equation [ [ eq_corot ] ] ) we define the magnetospheric radius @xmath101 as the location where the magnetic and material stresses are equal , @xmath262 where @xmath263 is the magnetic dipole moment of the ns , @xmath264 is an order unity constant that depends upon the details of the accretion flow near the magnetosphere ( see , e.g. , * ? ? ?
* ) , and the usual @xmath265 scaling is modified by the @xmath36-dependence of @xmath266 . in the propeller regime , @xmath267 . for ordinary thin - disk magnetic accretion , the disk extends to the magnetosphere , and the accretion is entirely shut off in this regime .
however for a riaf flow onto a millisecond pulsar , @xmath101 will generally lie inside the transition radius @xmath261 , so that the flow onto the magnetosphere will be quasi - spherical . in this geometry ,
a small fraction of the flow is able to reach the ns despite the centrifugal barrier present in the propeller regime @xcite . whether material is expelled in a strong or weak outflow , or else accumulates outside @xmath104 ( e.g. , a `` dead '' disk ) ,
depends upon details of the disk - magnetosphere interaction @xcite .
however , observationally , @xcite have shown that a strong propeller outflow can likely be ruled out in cen x-4 .
our observations support the conclusion of @xcite that low - level accretion is occurring during x - ray quiescence in cen x-4 , indicating that a small fraction of the accretion flow must eventually reach the ns .
however , it remains unclear what combination of the above mechanisms ultimately controls what that fraction is .
after cen x-4 , the next brightest well - studied quiescent ns / lmxb transient is aql x-1 . unlike cen x-4 , aql x-1 has a relatively short recurrence time scale of 12 yr . there have been several recent studies of its quiescent emission @xcite , all observing a soft thermal component and hard power - law component with no cutoff .
@xcite argue that , for their 2007 _ suzaku _ observations ( @xmath268[39]@xmath269 for an assumed distance of 5.2 kpc ) , the most appropriate model for the hard component is comptonization in either an optically thin ( @xmath270 ) , very hot ( @xmath271 kev ) corona or an optically thick ( @xmath272 ) , somewhat cooler ( @xmath273 kev ) corona .
they do not find the same inconsistency between their measured @xmath64 and radially uniform accretion that we found in equation ( [ eq_tau ] ) .
this is a consequence of their @xmath11 being higher and their @xmath64 being lower than in our cen x-4 observation .
however , we noted in [ sec - compt ] that @xmath274 and @xmath4 vary inversely , and the high @xmath4 values ( and corresponding cutoff energies ) that @xcite fit lie above their observation bandpass .
the _ suzaku _
data are thus unable to rule out the presence of a cutoff below 50100 kev ( but still above their bandpass ) ; this would introduce the same difficulties for a comptonization model that we found in cen x-4 , although it would be somewhat mitigated by the higher @xmath11 .
we note that our bremsstrahlung model could explain the hard component in aql x-1 for @xmath275 kev .
it is interesting to also compare the behavior of cen x-4 with systems that have been observed to transition between lmxb and radio pulsar states during x - ray quiescence .
the theoretical expectation is that such transitions are controlled by location of the ns magnetospheric boundary @xcite .
we can compare @xmath101 to both @xmath104 and the light - cylinder radius , @xmath276 for sufficiently low @xmath11 , we have @xmath277 and the system will be in the propeller regime , with accretion onto the ns ( mostly ) shut off @xcite . for even lower @xmath278 , we will have @xmath279 . in this case
, the radio pulsar mechanism can turn on , with the radiation pressure of a radio pulsar wind clearing the magnetosphere and an intrabinary shock giving rise to synchrotron x - ray emission @xcite .
the @xmath11 implied for transition to a millisecond radio pulsar state corresponds to @xmath280 erg s , where the exact value depends on details of the system and the disk - magnetosphere interaction .
indirect evidence for such transitions in ns / lmxbs during x - ray quiescence was previously reported in aql x-1 ( @xmath281 erg s@xmath2 ; * ? ? ?
* ) and sax j1808.4@xmath53658 ( @xmath282 * ? ? ?
* ; * ? ? ?
however , more direct evidence has been reported more recently in at least three systems . the 1.7 ms radio pulsar psr j1023
+ 0038 ( hereafter j1023 ) has made two transitions between the lmxb and radio pulsar states .
it is now understood to have been in an lmxb state during 20002001 , with direct optical evidence for the presence of an accretion disk @xcite .
however , a state transition then occurred , with subsequent observations establishing the absence of an accretion disk during 20022013 @xcite as well as the presence of a millisecond eclipsing radio pulsar during 20072013 @xcite , along with a low x - ray luminosity associated with intrabinary synchrotron shock emission @xcite .
a second state transition was observed more recently , with the radio pulsar turning off and an accretion disk reemerging @xcite . in all of these observations
, j1023 has remained in x - ray quiescence in the sense that a high - luminosity ( @xmath283 erg s@xmath2 ) transient x - ray outburst was not observed .
however , two distinct sub - states are evident : a faint x - ray quiescent state ( @xmath284 erg s@xmath2 ) during which radio pulsations are seen , and a bright x - ray quiescent state ( @xmath285 erg s@xmath2 ) during which the radio pulsar is off .
this suggests that @xmath101 has moved outside the light cylinder in the fainter state . in both sub - states ,
the 0.310 kev x - ray emission has a power - law spectrum with little or no thermal component , with @xmath286 in the faint state @xcite and @xmath287 in the bright state @xcite .
_ nustar _ observations show that these power - law spectra remain unbroken up to at least 79 kev @xcite . during faint x - ray quiescence
, the x - ray flux shows high - amplitude modulation at the orbital period @xcite , similar to what is seen in the x - ray emission from some ( but not all ) eclipsing millisecond radio pulsars ( the so - called black widow and redback systems ; * ? ? ?
* ; * ? ? ?
* ) . during bright x - ray quiescence ,
the x - ray flux shows strong , rapid flickering , with the intensity varying by an order of magnitude on time scales @xmath0100 s @xcite .
a second object in which two lmxb / radio pulsar state transitions were seen is the m28 globular cluster source psr j1824@xmath52452i ( hereafter m28i ) , a 3.9 ms radio pulsar .
the radio pulsar underwent a bright transient x - ray outburst ( @xmath288 erg s@xmath2 ) in 2013 march during which accretion - powered millisecond pulsations and a thermonuclear x - ray burst were observed , establishing the system as an ns / lmxb .
the system returned to x - ray quiescence within a month , at which point radio pulsations were again detected @xcite .
these observations demonstrate that lmxb / radio pulsar state transitions can occur on time scales as short as days . during x - ray quiescence ,
rapid ( @xmath31500 s ) intensity variations of nearly an order of magnitude are again seen [ ( 0.64)@xmath289 erg s@xmath2 ] , with no change in the 0.310 kev x - ray spectrum : an absorbed power - law with @xmath291 and no detectable thermal component @xcite .
remarkably , this rapid variability seems to toggle between two stable flux levels ; @xcite suggest that this represents fast transitions between synchrotron shock emission and magnetospheric accretion .
no orbital variability of the x - ray flux is reported in m28i .
the third object in which an lmxb / radio pulsar transition was seen is xss j12270@xmath54859 ( hereafter j12270 ) , a faint hard x - ray source associated with a relatively bright _ fermi _ @xmath292-ray source . during 20032012 ,
the source was in a quiescent lmxb state with an absorbed power - law x - ray spectrum with @xmath293 , no evidence for a thermal spectral component , a 0.110 kev luminosity of @xmath294 erg s@xmath2 , highly variable x - ray flaring , and multiwavelength evidence for the presence of an accretion disk @xcite . in late 2012 , the source made a transition to a lower ( @xmath295 erg s@xmath2 ) luminosity state with a power - law x - ray spectrum with @xmath291 , a thermal fraction @xmath09% , and a large - amplitude orbital modulation of the x - ray flux @xcite .
after this transition , 1.69 ms radio pulsations were also detected @xcite .
cen x-4 has comparable luminosity to j1023 , m28i , and j12270 , and so it is presumably at least close to the regime where lmxb / radio transitions could occur . the rapid x - ray variability we observe ( see 3.1 ) is quite different from the large - amplitude orbital modulation seen in the low - quiescent state of j1023 , but it is qualitatively similar to ( although somewhat weaker than ) the flickering seen in the high - quiescent states of j1023 and j12270 as well as the m28i quiescent variability . on the other hand , unlike those three sources , the x - ray spectrum of cen x-4 has a substantial thermal fraction ( @xmath360% )
this may indicate that more of the accretion flow reaches the surface of cen x-4 than in the other systems .
if we apply our cen x-4 compton - cooled bremsstrahlung model to the hard x - ray emission in j1023 , m28i , and j12270 during their radio - quiet / x - ray quiescent states , then we would expect a high ( 50100 kev ) electron temperature and an unbroken 210 kev power - law x - ray spectrum , consistent with what was observed .
of course , synchrotron shock emission can also produce an unbroken power - law in the x - ray band , and we expect this mechanism to dominate in the radio pulsar state . at sufficiently low luminosity , cen x-4 should transition into a radio pulsar state ; we suggest that this may be what occurs at @xmath296 erg s@xmath2 in figure [ fig_gamma ] , and that the jump in @xmath59 might reflect a sharp transition from high - temperature bremsstrahlung emission to synchrotron shock emission .
this is consistent with the suggestion by @xcite that the power - law component in quiescent ns / lmxbs arises from accretion at higher @xmath11 and from some a different , non - accretion mechanism ( e.g. , synchroton shock emission ) at lower @xmath11 ; they used this to explain how the fractional power - law contribution to the quiescent luminosity varies with @xmath278 in quiescent ns / lmxbs .
it would be interesting to search for millisecond radio pulsations from cen x-4 when its x - ray luminosity next drops to @xmath297 erg s@xmath2 .
we thank the referee , craig heinke , for several suggestions that greatly improved our paper .
thanks herman marshall , sera markoff , caroline dangelo , stephen reynolds , federico bernardini , phil charles , and chris done for useful discussions , and luca zampieri and roberto soria for sharing their xspec additive table model zamp .
we also thank thorsten brand for help with evaluating the level of photon pileup in the _ xmm - newton _ data .
this work was supported in part under nasa contract nng08fd60c and made use of data from the _ nustar _ mission , a project led by the california institute of technology , managed by the jet propulsion laboratory , and funded by nasa .
we thank the _ nustar _ operations , software and calibration teams for support with the execution and analysis of these observations .
this research has made use of the _ nustar _ data analysis software ( nustardas ) , jointly developed by the asi science data center ( asdc , italy ) and the california institute of technology ( usa ) .
jat acknowledges partial support from the _ xmm - newton _ guest observer program through nasa grant nnx13ab47 g .
archibald , a. m. , kaspi , v. m. , bogdanov , s. , et al .
2010 , , 722 , 88 archibald , a. m. , stairs , i. h. , ransom , s. m. , et al .
2009 , science , 324 , 1411 arnaud , k. a. 1996 , in asp conf .
ser . 101 , astronomical data analysis software and systems v , ed
. g. h. jacoby & j. barnes ( san francisco , asp ) , 17 arons , j. 2002 , in asp conf . ser .
271 , neutron stars in supernova remnants , ed .
p. o. slane & b. m. gaensler ( san francisco , asp ) , 71 arons , j. , & tavani , m. 1993 , , 403 , 249 ( at93 ) asai , k. , dotani , t. , hoshi , r , et al .
1998 , , 50 , 611 bassa , c. g. , patruno , a. , hessels , j. w. t. , et al .
2014 , , 441 , 1825 belian , r. d. , conner , j. p. , & evans , w. d. 1972 , , 171 , l87 bernardini , f. , cackett , e. m. , brown , e. f. , et al . 2013 , , 436 , 2465 bildsten , l. , & rutledge , r. e 2000 , , 541 , 908 blair , w. p. , raymond , j. c. , dupree , a. k. , et al . 1984 , , 278 , 270 blandford , r. d. , & begelman , m. c. 1999 , , 303 , l1 bogdanov , s. , archibald , a. m. , hessels , j. w. t. , et al .
2011 , , 742 , 97 bogdanov , s. , patruno , a. , archibald , a. m. , et al .
2014 , , 789 , 40 brown , e. f. , bildsten , l. , & rutledge , r. e. 1998 , , 504 , l95 burderi , l. , di salvo , t. , dantona , f. , et al .
2003 , , 404 , l43 burderi , l. , possenti , a. , dantona , f. , et al .
2001 , , 560 , l71 cackett , e. m. , brown , e. f. , degenaar , n. , et al .
2013 , , 433 , 1362 cackett , e. m. , brown , e. f. , miller , j. m. , & wijnands , r. 2010 , , 720 , 1325 cackett , e. m. , fridriksson , j. k. , homan , j. , et al .
2011 , , 414 , 3006 cackett , e. m. , wijnands , r. , heinke , c. o. et al .
2005 , , 620 , 922 campana , s. , & stella , l. 2000 , , 541 , 849 campana , s. , & stella , l. 2003 , , 597 , 474 campana , s. , colpi , m. , mereghetti , s. , stella , l. , & tavani , m. 1998a , , 8 , 279 campana , s. , davanzo , p. , casares , j. , et al .
2004 , , 614 , l49 campana , s. , israel , g. l. , stella , l. , gastaldello , f. , & mereghetti , s. 2004 , , 601 , 474 campana , s. , stella , l. , gastaldello , f. , et al .
2002 , , 575 , l15 campana , s. , stella , l. , mereghetti , s. , et al . 1998b , , 499 , l65 canizares , c. r. , mcclintock , j. e. , & grindlay , j. e. 1980 , , 236 , l55 chevalier , c. , ilovaisky , s. a. , van paradijs , j. , pedersen , h. , & van der klis , m. 1989 , , 210 , 114 conner , j. p. , evans , w. d. , & belian , r. d. 1969 , , 157 , l157 coti zelati , f. , campana , s. , davanzo , p. , & melandri , a. 2014 , , 438 , 2634 dangelo , c. r. & spruit , h. c. 2010 , , 406 , 1208 dangelo , c. r. & spruit , h. c. 2012 , , 420 , 416 davanzo , p. , campana , s. , casares , j. , et al .
2005 , , 444 , 905 de martino , d. , belloni , t. , falanga , m. , et al .
2013 , , 550 , a89 de martino , d. , falanga , m. , bonnet - bidaud , j .-
m . , et al .
2010 , , 515 , a25 deloye , c. j. , heinke , c. o. , taam , r. e. , & jonker , p. g. 2008 , , 391 , 1619 deufel , b. , dullemond , c. p. , & spruit , h. c. 2001 , , 377 , 955 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 evans , w. d. , belian , r. d. , & conner , j. p. 1970
, , 159 , l57 frank , j. , king , a. , & raine , d. 2002 , accretion power in astrophysics , 3rd ed .
( cambridge u. press ) gaensler , b. m. , & slane , p. o. 2006 , , 44 , 17 guillot , s. , servillat , m. , webb , n. a. , & rutledge , r. e. 2013 , , 772 , 7 harrison , f. a. , craig , w w. , christensen , f. e. , et al .
2013 , , 770 , 103 heinke , c. o. , rybicki , g. b. , narayan , r. , & grindlay , j. e. 2006 , , 644 , 1090 hjellming , r. m. 1979 , iau circ . , 3369 hjellming , r. m. , calovini , t. a. , han , x. h. , et al .
1988 , , 335 , l75 hua , x .-
titarchuk , l. 1995 , , 449 , 188 illarionov , a. f. , & sunyaev , r. a. 1975 , , 39 , 185 jansen , f. , lumb , d. , altieri , b. , et al .
2001 , , 365 , l1 jonker , p. g. , galloway , d. k. , mcclintock , j. e. , et al .
2004 , , 354 , 666 joss , p. c. , & li , f. k. 1980 , , 238 , 287 kalberla , p. m. , burton , w. b. , hartmann , d. , et al .
2005 , , 440 , 775 kaluzienski , l. j. , holt , s. s. , & swank , j. h. 1980 , , 241 , 779 kennel , c. f. , & coroniti , f v. 1984a , , 283 , 694 kennel , c. f. , & coroniti , f v. 1984b , , 283 , 710 king , a. r , kolb , u. , & burderi , l. 1996 , , 464 , l127 kulkarni , s. r. , & hester , j. j. 1988 , , 335 , 801 kuulkers , e. , int zand , j. j. m. , & lasota , j .- p .
2009 , , 503 , 889 lasota , j .- p .
2001 , new astron .
, 45 , 449 lii , p. s. , romanova , m. m. , ustyugova , g. v. , et al .
2014 , , 441 , 86 linares , m. , bahramian , a. , heinke , c. , et al .
2014 , , 438 , 251 mahadevan , r. , & quataert , e. 1997 , , 490 , 605 matsuoka , m. , inoue , h. , koyama , k. , et al .
1980 , , 240 , l137 mcclintock , j. e. , & remillard , r. a. 2000 , , 531 , 956 menou , k. , & mcclintock , j. e. 2001 , , 557 , 304 menou , k. , esin , a. a. , narayan , r. , et al .
1999 , , 520 , 276 narayan , r. , & yi , i. 1995 , , 452 , 710 narayan , r. , mahadevan , r. , & quataert , e. 1998 , in the theory of black hole accretion disks , ed . m. a. abramowicz , g. bjornsson , & j. e. pringle ( cambridge u. press ) , 148 ( arxiv : astro - ph/9803141 ) papitto , a. , ferrigno , c. , bozzo , e. , et al .
2013 , , 501 , 517 patruno , a. , archibald , a. m. , hessels , j. w. t. , et al .
2014 , , 781 , l3 park , s. , & garcia , m. r. 2011 , , 43 , 144.25 psaltis , d. , & chakrabarty , d. 1999 , , 521 , 332 rutledge , r. e. , bildsten , l. , brown , e. f. , pavlov , g. g. , & zavlin , v. e. 1999 , , 514 , 945 roberts , m. s. e. 2012 , proc .
iau symp . , 291 , 127 roberts , m. s. e. , mclaughlin , m. a. , gentile , p. , et al .
2014 , astron .
nachr . , 335 , 313 roy , j. , bhattacharyya , b. , & ray , p. s. 2014 , atel , 5890 rutledge , r. e. , bildsten , l. , brown , e. f. , pavlov , g. g. , & zavlin , v. e. 2002 , , 577 , 346 rybicki , g. r. , & lightman , a. p. 1979
, radiative processes in astrophysics ( new york , wiley ) sakurai , s. , torii , s. , noda , h. , et al 2014 , , 66 , 10 schlafly , e. f. , & finkbeiner , d. p. 2011
, , 737 , 103 shahbaz , t. , watson , c. a. , & dhillon , v. s. 2014 , , 440 , 594 soria , r. , zampieri , l. , zane , s. , & wu , k. 2011 , , 410 , 1886 spruit , h. c. & taam , r. e. 1993 , , 402 , 593 stappers , b. w. , archibald , a. m. , hessels , j. w. t. , et al .
2014 , , 790 , 39 stappers , b. w. , gaensler , b. m. , kaspi , v. m. , et al .
2003 , science , 299 , 1372 stella , l. , campana , s. , colpi , m. , mereghetti , s. , & tavani , m. 1994 , , 423 , l47 strder , l. , briel , u. , dennerl , k. , et al .
2001 , , 365 , l18 tendulkar , s. p. , yang , c. , an , h. , et al .
2014 , , 791 , 77 thorstensen , j. r. , & armstrong , e. 2005 , , 130 , 759 titarchuk , l. 1994 , , 434 , 313 titarchuk , l. , & lyubarskij 1995 , , 450 , 876 torres , m. a. p. , casares , j. , martinez - pais , i. g. , & charles , p. a. 2002 , , 334 , 233 turner , m. j. l. , abbey , a. , arnaud , m. , et al .
2001 , , 365 , l27 ustyugova , g. v. , koldoba , a. v. , romanova , m. m. , et al .
2006 , , 646 , 304 van paradijs , j. 1996 , , 464 , l139 vaughan , s. , edelson , r. , warwick , r. s. , & uttley , p. 2003
, , 345 , 1271 verner , d. a. , ferland , g. j. , korista , k. t. , & yakovlev , d. g. 1996 , , 465 , 487 wang , z. , archibald , a. m. , thorstensen , j. r. , et al .
2009 , , 703 , 2017 wijnands , r. , degenaar , n. , & page , d. 2013 , , 432 , 2366 wik , d. r. , hornstrup , a. , molendi , s. , et al .
2014 , , 792 , 48 wilms , j. , allen , a. , & mccray , r. 2000 , , 542 , 914 zampieri , l. , campana , s. , turolla , r. , et al .
2001 , , 378 , l5 zampieri , l. , turolla , r. , zane , s. , & treves , a. 1995 , , 439 , 849 | the low - mass x - ray binary cen x-4 is the brightest and closest ( @xmath01.2 kpc ) quiescent neutron star transient .
previous 0.510 kev x - ray observations of cen x-4 in quiescence identified two spectral components : soft thermal emission from the neutron star atmosphere and a hard power - law tail of unknown origin .
we report here on a simultaneous observation of cen x-4 with _ nustar _
( 379 kev ) and _ xmm - newton _ ( 0.310 kev ) in 2013 january , providing the first sensitive hard x - ray spectrum of a quiescent neutron star transient .
the 0.379 kev luminosity was @xmath1 erg s@xmath2 , with @xmath360% in the thermal component .
we clearly detect a cutoff of the hard spectral tail above 10 kev , the first time such a feature has been detected in this source class .
we show that thermal comptonization and synchrotron shock origins for the hard x - ray emission are ruled out on physical grounds .
however , the hard x - ray spectrum is well fit by a thermal bremsstrahlung model with @xmath4=18 kev , which can be understood as arising either in a hot layer above the neutron star atmosphere or in a radiatively - inefficient accretion flow ( riaf ) .
the power - law cutoff energy may be set by the degree of compton cooling of the bremsstrahlung electrons by thermal seed photons from the neutron star surface .
lower thermal luminosities should lead to higher ( possibly undetectable ) cutoff energies .
we compare cen x-4 s behavior with the psr j1023 + 0038 , igr j18245@xmath52452 , and xss j12270@xmath54859 , which have shown transitions between lmxb and radio pulsar modes at a similar x - ray luminosity . |
ngc383 , of type s0 ( arp 1968 ) , is a member of a bright chain of galaxies ( arp 1966 , arp & bertola 1971 , kormendy & bahcall 1974 ) , which itself belongs to a group of galaxies ( zwicky et al .
1961 ) located within the perseus - pisces filament .
ngc383 is at redshift @xmath6=0.017 and has a companion galaxy in 33 distance .
ngc383 is a moderately bright radio galaxy ( 3c31 ) and has been extensively studied at radio wavelengths in the past ( e.g. , macdonald et al .
1968 , burch 1977,1979 , klein & wielebinski 1979 , fomalont et al . 1980 ,
ekers et al .
1981 , van breugel 1982 , strom et al . 1983 , condon et al .
1991 , andernach et al .
1992 , artyukh et al .
1994 , lara et al .
1997 , henkel et al .
it shows a symmetric edge darkened double - source structure with two strong jets .
the origin of the structures seen in the radio morphology of dominant cluster / group members is still not well understood .
the structures of the jets of ngc383 were interpreted by blandford & icke ( 1978 ) as due to tidal interaction with the companion galaxy ngc382 . however , fraix - burnet et al .
( 1991 ) found no evidence for interaction between the two galaxies .
further jet models were presented by bridle et al .
( 1980 ) and bicknell ( 1984 ) . whereas butcher et al .
( 1980 ) reported evidence for a detection of the jet at optical wavelengths , this was not confirmed by later studies ( keel 1988 , owen et al .
1990 , fraix - burnet et al .
1991 ) .
the optical spectrum of the nucleus is characterized by emission lines .
the optical emission was found to be extended , and in form of a rotating disk ( owen et al .
1990 ) or ring ( fraix - burnet et al .
1991 ) that coincides with a dust - ring reported by butcher et al .
the emission line ratios in the disk were found to be similar to those at the nucleus and owen et al . concluded that the disk ionization is probably driven by the nucleus .
ngc383 is the brightest member of a rich group of galaxies .
the group is included in the zwicky catalogue ; membership studies based on velocity measurements were performed by moss & dickens ( 1977 ) , garcia ( 1993 ) , sakai et al .
( 1994 ) , and ledlow et al .
( 1996 ) , who included 25 , 22 , 16 , 32 member galaxies , respectively . among
the 27 groups studied , ledlow et al . found the galaxy velocity distribution to significantly deviate from a gaussian in 55% of them ; the ngc383 group was not among those .
sakai et al . , using a standard estimator as in heisler et al . ( 1985 ) , derived a virial mass of @xmath7 = 0.210@xmath8 @xmath9 , and a mass - light ratio @xmath10 = 240 . extended x - ray emission from the group was first detected by ( fabbiano et al .
the data were analyzed by morganti et al .
( 1988 ) in the course of a large sample study of the effect of gas pressure on radio sources . to perform a detailed spatial and spectral analysis of the x - ray emission from ngc383 , the intra - group medium ( igm ) , and an investigation of the relation with the radio jet , a pspc observation was applied for .
these data presented here have been previously partly analyzed by trussoni et al .
( 1997 ; t97 hereafter ) in a study of hot coronae in nearby radio galaxies . here , we extent their analysis and focus on the properties of the igm and the ngc383 group as a whole , in particular , the determination of gas and total mass ; and the nature of the x - ray emission from ngc383 .
we also briefly discuss new hri data which were retrieved from the archive .
also located in the field of view is the high - redshift ( @xmath6=2.027 ) bal quasar qso 0104 + 3153 .
x - ray emission from the direction of this quasar was discovered in the medium sensitivity survey ( stocke et al .
the qso , and a nearby ( 10 ) giant elliptical galaxy , member of a small group , were identified as possible counterparts of the x - ray emission .
due to its proximity to the elliptical , the quasar is considered as prime candidate for lensing by halo stars .
optical variability of @xmath110.5@xmath12 was detected by gioia et al .
the pspc spectrum was briefly discussed in a large sample study by ciliegi & maccacaro ( 1996 ) .
the identification of the source ( qso , elliptical galaxy or igm of the small group ) remained unclear . here
, we perform a detailed study of the properties of this x - ray source and address the counterpart question on the basis of ( i ) long- and short - term variability behaviour , ( ii ) x - ray spectral shape , ( iii ) improved spatial position obtained with the hri .
physical parameters are calculated for @xmath13 = 50 km / s / mpc , @xmath14 = 0.5 and assuming the galaxies / group to follow the hubble flow . for the distance of the ngc383 group
, 1 corresponds to a scale of 0.5 kpc .
the group was the target of a pointed pspc ( trmper 1983 ; pfeffermann et al .
1987 ) observation with ngc383 in the center of the field of view .
our observation was performed from july 28 29 , 1991 with a duration of 27.5 ksec .
the satellite was not in wobble mode during this observation .
the background was usually determined in a source - free ring around the target source ( for details see below ) .
the data were corrected for vignetting and dead - time , using the exsas software package ( zimmermann et al .
widely extended x - ray emission is present ( fig .
1 ) . the central region appears roughly circularly symmetric .
emission from the group can be traced out to a distance of @xmath1133 radius ( @xmath11 1 mpc ) .
the total countrate ( channels 52201 ) within this region is 0.58 @xmath15 cts / s .
several group ellipticals are individually detected , the brightest with countrates of 0.024@xmath16 cts / s ( ngc383 ) , 0.0058@xmath17 cts / s ( ngc379 ) , 0.0124@xmath18 cts / s ( ngc380 ) , 0.0041@xmath19 cts / s ( ngc385 ) , and 0.0021@xmath17 cts / s ( ngc384 ) as summarized in table [ fitres2 ] .
x - ray emission from the direction of 1e0104 is detected with 0.028@xmath16 cts / s .
to carry out the spectral analysis of the individual sources , photons in the amplitude channels 11 - 240 were binned according to a constant signal / noise ratio of @xmath20 5@xmath21 .
results given below refer to the pspc data if not mentioned otherwise . an hri observation centered on ngc383 was performed between july 11 and 25 , 1994 with a duration of 25.0 ksec .
the data were retrieved from the archive .
( an earlier hri observation , of jan .
1992 with 2.8 ksec duration , is briefly discussed in t97 . ) due to the lower sensitivity of the hri , essentially only the bright point - like sources are detected .
these data are analyzed mainly to study the spatial structure of the x - ray emission from the bright group galaxies , and to search for variability in individual sources .
a raymond - smith ( rs ) model was fit to the extended x - ray emission . if not noted otherwise , abundances were fixed to 0.35@xmath22solar ( anders & grevesse 1989 ) . in a first step ,
the group photons within a circular region of radius 1000 were selected ; in a second step , this region was split into an inner circle ( radius 200 ) , an intermediate ring ( 200 out to 500 ) and an outer ring ( 500 to 1000 ) .
point - like sources were removed . to this end
, we first carried out a source detection using the exsas software .
this resulted in about 35 sources detected within the total field of view .
for the brightest sources , extents were then determined by inspecting the radial source profiles ( the radius where the source profile merges with the local background taken as source extent ) and by comparison with the profile expected from the psf of the detector ( for a more detailed discussion of ngc383 see sect .
this resulted in source extraction radii of @xmath11100 for the brightest central sources .
the detected sources were then removed within circular regions centered on their x - ray positions . as background we selected a source - free circular region in the outer part of the fov . since the spectrum below 0.5 kev is strongly dominated by background , we usually excluded photons below this energy from the spectral fit . treating the cold absorption as free parameter always resulted in a value ( nearly ) consistent with the galactic absorption towards ngc383 ( @xmath23 @xmath24 , dickey & lockman 1990 ) within the errors .
we therefore fixed @xmath25 = @xmath26 .
for the total emission within 1000 we find a temperature @xmath27 = 1.5@xmath28 kev .
the temperature values derived for the three separate regions are consistent with this value and with each other within the errors ( table [ fitres ] ) .
this value of @xmath29 is somewhat lower than the one given in morganti et al .
( 1988 ; their tab .
2 ) on the basis of an ipc observation ; they estimate @xmath30 3 kev .
to check the robustness of the obtained temperature , we performed a few tests : if we do _ not _ remove the point sources and re - fit the total spectrum we get @xmath27 = 1.4@xmath28 kev ; if we fit a thermal bremsstrahlung model we obtain @xmath27 = 1.4@xmath31 kev ; if an rs plasma with solar instead of depleted abundances is assumed , the quality of the fit slightly improves without affecting the value of the temperature within the errors and we derive @xmath27 = 1.6@xmath28 kev .
t97 obtain @xmath27 = 1.5 kev for abundances of 0.35@xmath22solar .
results of our spectral fits are summarized in table [ fitres ] .
using @xmath27 = 1.5 kev and @xmath32 , the total ( 0.12.4 kev ) luminosity within 33 is @xmath33 erg / s (= 1.310@xmath34 erg / s if the abundances are fixed to the solar value ) .
all bright ellipticals of the optical chain arp331 are individually detected in x - rays .
the center of the x - ray emission maximum from ngc383 at @xmath35 = 1@[email protected] , @xmath37 = 3224445 ( j2000 ) coincides well with the position of the optical nucleus at @xmath35 = 1@[email protected] , @xmath37 = 3224448 . for an overlay of the pspc x - ray contours on an optical image of the ngc383 group
see t97 ( their fig .
1 ) . to derive physical properties of the group ,
we first assume spherical symmetry of the extended x - ray emission , the emission to be centered on ngc383 , and rough isothermality to hold . for critical comments on and justification of `
standard ' assumptions see , e.g. , bhringer et al .
1998 ( their sects . 2 , 3 ) .
a @xmath38-model ( e.g. , cavaliere & fusco - femiano 1976 , gorenstein et al .
1978 , jones & forman 1984 ) of the form @xmath39 was fit to the azimuthally averaged surface brightness profile of the pspc observation ( detected point sources were , again , removed except emission from ngc383 , inner bins were then excluded from the fit since they are dominated by emission from ngc383 itself ) .
this yields a central surface brightness @xmath40 cts / s / arcmin@xmath41 , a slope parameter @xmath38 = 0.34 and a core radius @xmath42 = 64 kpc . the gas mass enclosed inside 1 mpc amounts to @xmath43 = 1.510@xmath44 m@xmath2 .
inspection of fig .
1 shows that ngc383 is not located perfectly in the center of the large scale x - ray emission , but offset by about 1 to the north - east .
in fact , using the best - fit beta - model to construct a synthetic model image and subtracting the model image from the observed one leaves some residual extended emission to the sw and some ` negative ' emission levels to the ne .
therefore , in a second step , the surface brightness profile was centered at @xmath35 = 1@[email protected] , @xmath37 = 3223457 and re - fit after emission from ngc383 was removed within a segment . in this case
, we obtain @xmath45 cts / s / arcmin@xmath41 , @xmath46 , and @xmath47 kpc ( errors are 1@xmath21 ) . the only slight change of the fit parameters underlines the robustness of the fit .
the beta - model fit to the surface brightness profile is shown in fig .
[ profile ] . again constructing a model image and subtracting this from the observed one
now does not show any residual emission except the point - like sources .
this model is also used for the gas and total mass estimate below .
these results differ from the previous analysis of t97 in yielding a smaller core radius ( they derived @xmath42 = 230 kpc ) and shallower slope ( they found @xmath38 = 0.6 ) for the data presented here . we have checked that our results are robust ( @xmath38 varying between @xmath11 0.3 and 0.45 in extreme cases ) against non - removal of individual sources ( up to all ) , shifts of the central position , and to systematic exclusion of individual bins ( e.g. , the inner or outer ones ) before fitting the surface brightness profile .
our best fit leads to a larger ratio of gas to total mass at large radii .
it also has consequences when comparing the pressure of the igm with the non - thermal pressure of the radio jet ( sect .
5.1.3 ) . for the beta - modell applied in the previous section ,
the gas density distribution is given by @xmath48 this implies a central density @xmath49 = 1.310@xmath50 @xmath24 and a gas mass within 1mpc of @xmath43 = 1.310@xmath44 m@xmath2 . assuming spherical symmetry and the group to be approximately in hydrostatic equilibrium , the total gravitating mass follows the relation @xmath51 with the observed parameters this results in an integrated total mass of @xmath52 = 6.310@xmath44 m@xmath2 within 1 mpc radius and a gas mass fraction of 21% . using instead the galaxy velocity dispersion as derived from optical observations , @xmath21 = 466 km / s ( ledlow et al .
1996 ; see also sakai et al .
1994 ) , and a core radius of 73 kpc we get a mass of @xmath11710@xmath44 m@xmath2 within 1 mpc , which is well consistent with the value obtained purely from the x - ray data .
the profile of total and gas mass is displayed in fig .
errors on the mass @xmath52 are obtained from the temperature range allowed by the x - ray spectral analysis , @xmath27 = 1.5@xmath31 kev , and a temperature profile for a family of @xmath53 models with polytropic index @xmath53 in the range 0.9 1.3 . in the polytropic models
the nominal temperature is fixed at the core radius .
there are some spectacular examples of pressure interaction between the radio and x - ray gas in clusters of galaxies ( e.g. , bhringer et al .
1993 , 1995 , harris et al .
1994 , clarke et al .
1997 , otani et al .
1998 ) . in the present case
, we do not find conspicuous morphological correlations between radio- and x - ray emission ( fig .
[ over_rx ] )
. this may be partly due to the narrowness of the jet , the still limited spatial resolution of the pspc , and the 2d view of the 3d source structure .
changes of the jet orientation angle near the locations of some optical chain galaxies , now also detected as strong x - ray sources , were already noted by strom et al .
as background we chose ( i ) a source - free ring around the target source , and ( ii ) a source - free circular region near the target source .
this allows to check for sensitivity against the background correction .
all majour fits were repeated for both background geometries .
further , we note that the source emission in the first bin ( below 0.4 kev ) is weak .
we repeated all fits after having removed the first bin from the spectrum . in all cases ,
the results presented below are found to be robust .
first , several single component models were fit to the x - ray spectrum , starting with a powerlaw ( pl ) .
although this model fits the x - ray spectrum ( @xmath54 = 1.3 ) , the derived parameters are unusual .
the slope is extremely steep @xmath55 5 , and there is evidence for strong excess absorption ( about 8 times the galactic value ) . a single steep pl may be mimicked by a flat pl plus soft excess . parameterizing the excess as black body and fixing = 1.9 , we do not find a successful fit .
this also holds for a single pl in which absorption is fixed to the galactic value ( @xmath54 = 3.3 )
. a single rs model with metal abundances of 0.35 @xmath22 solar does not give an acceptable fit , either ( @xmath54 = 2.4 ) .
lowering the abundances up to @xmath56 0.1 @xmath22 solar yields an acceptable fit ( see also t97 ) , but such low abundances are unexpected for dominant group galaxies yr ; see also sarazin ( 1997 ) .
+ for a recent thorough discussion of the issue of single-@xmath29 models of very subsolar abundances vs. two - component models of @xmath11solar abundances see also buote & fabian ( 1998 ) . ] . in a second step ,
two - component models consisting of contributions from both , a rs plasma and a pl source , or two rs sources were applied . in the rs+pl description ,
the pl index was fixed to = 1.9 ( the value typically observed in sy ; since we want to specifically test for the presence of an agn ) .
this provides a good fit ( @xmath54 = 1.1 ) .
if @xmath25 is treated as additional free parameter , it is found to be of the order of the galactic value .
for the rs component , we obtain @xmath4 0.6 kev ( error contours are displayed in fig .
[ chi2 ] ) .
the pl component contributes with a ( 0.12.4 kev ) luminosity of @xmath57 erg / s , the total ( 0.12.4 kev ) luminosity is @xmath58 erg / s ( for @xmath25 fixed to @xmath26 ) .
alternatively , the spectrum can be fit with a double rs model . in this case
we find the hotter component to be ill - constrained .
if we fix @xmath59 = 1.5 kev , we get @xmath60 = 0.4@xmath31 kev ( @xmath54 = 1.1 ; we used @xmath32 ) and luminosities of @xmath61 erg / s , @xmath62 erg / s . for comparison , if @xmath59 = 5 kev is chosen , @xmath60 = 0.5@xmath31 kev ( @xmath54 = 1.0 ) and @xmath63 erg / s , @xmath64 erg / s
the results of the spectral fits are summarized in table [ fitres ] .
lllllcc & model & @xmath25 & @xmath65 & @xmath27 & @xmath66 & @xmath67 + & & 10@xmath68 @xmath69 & & kev & x solar & + ngc383 & pl & 4.4 & @xmath70 & & & 1.38 ( 11 ) + & pl & 0.523@xmath71 & @xmath72 & & & 3.25 ( 12 ) + & rs & 0.43 & & 1.14 & 0.35@xmath73 & 2.42 ( 11 ) + & rs & 0.59 & & 0.98 & 0.10@xmath73 & 1.50 ( 11 ) + & rs+pl & 0.69 & @xmath74 & 0.55 & 0.35@xmath73 & 1.12 ( 10 ) + & rs+pl & 0.523@xmath71 & @xmath74 & 0.67 & 0.35@xmath73 & 1.18 ( 11 ) + & rs+rs & 0.523@xmath71 & & 0.5/5@xmath75 & 0.35@xmath73 & 1.02 ( 11 ) + & rs+rs & 0.523@xmath71 & & 0.4/5@xmath75 & 1.00@xmath73 & 1.01 ( 11 ) + ngc380 & pl & 0.523@xmath71 & @xmath76 & & & 5.58 ( 7 ) + & rs & 0.523@xmath71 & & 0.92@xmath77 & 0.35@xmath73 & 0.84 ( 7 ) + ngc379 & pl & 0.523@xmath71 & @xmath78 & & & 3.38 ( 4 ) + & rs & 0.523@xmath71 & & 0.50@xmath79 & 0.35@xmath73 & 1.03 ( 4 ) + igm & rs ; total & 0.523@xmath71 & & 1.46@xmath80 & 0.35@xmath73 & 1.14 ( 22 ) + & rs ; total & 0.523@xmath71 & & 1.56@xmath81 & 1.00@xmath73 & 0.73 ( 22 ) + & rs ; inn . &
0.523@xmath71 & & 1.48@xmath81 & 0.35@xmath73 & 1.43 ( 18 ) + & rs ; mid . & 0.523@xmath71 & & 1.56@xmath82 & 0.35@xmath73 & 1.79 ( 19 ) + & rs ; out . & 0.523@xmath71 & & 1.37@xmath83 & 0.35@xmath73 & 1.42 ( 14 ) + @xmath84 fixed to the galactic value ; @xmath85 fixed ; @xmath86 temperature of second rs component , fixed + the x - ray emission from the direction of ngc383 is strongly peaked . to check
how much of the emission might arise from a point source we compared the radial source profile with the instrumental point - spread function ( psf ) .
we find that the emission from ngc383 is not significantly extended beyond the psf of the pspc .
thus , the data are consistent with the bulk of the x - ray emission arising from a point source . performing a similar analysis for the hri observation
, we find a deviation of the source profile from the hri psf .
however , similar deviations are found for the ( presumably pointlike ) f star which is located within the field of view ( after taking into account the appropriate off - axis psf for the star ) .
we conclude that in the present data there is no evidence for source extent .
the x - ray lightcurve of ngc383 is displayed in fig .
[ light ] .
an agn / point - source might reveal itself by variability ( but not necessarily ) .
we find a constant source flux within the errors .
a spectral analysis was performed for the two x - ray brightest galaxies , ngc379 and ngc380 .
a pl model does not provide an acceptable fit ( @xmath54 = 3.4 and 5.6 ) with residuals strongly indicative of the presence of an rs component .
such a model indeed gives an excellent fit ( @xmath54 = 1.0 and 0.8 ) .
we find temperatures of 0.5 kev ( ngc379 ) and 0.9 kev ( ngc380 ) .
the absorption - corrected fluxes for this model description in the ( 0.12.4 ) kev band are @xmath87 erg / cm@xmath41/s ( ngc379 ) and @xmath89 erg / cm@xmath41/s ( ngc380 ) , and the corresponding luminosities @xmath90 erg / s ( ngc379 ) and @xmath91 erg / s ( ngc380 ) .
to estimate luminosities also for ngc384 and 385 , which are too weak to allow direct spectral fits , we adopted a rs spectrum of 0.5 kev .
the derived luminosities are given in table 2 .
ccccc galaxy & @xmath92 & @xmath93 & @xmath94 & @xmath95 + & 10@xmath96 cts / s & kev & erg / s & erg / s + ngc379 & 0.58 & 0.5 & 1.110@xmath97 & 3.110@xmath34 + ngc380 & 1.24 & 0.9 & 2.310@xmath97 & 4.010@xmath34 + ngc383 & 2.40 & 0.6+pl & 4.710@xmath97 & 5.110@xmath34 + ngc384 & 0.21 & 0.5@xmath98 & 5.010@xmath99 & 1.810@xmath34 + ngc385 & 0.41 & 0.5@xmath98 & 7.610@xmath99 & 2.610@xmath34 + @xmath100 either determined directly from spectral fit ( sect . 4.1.1 , 4.2 ) or fixed to 0.5 kev ( marked with ` @xmath98 ' ) in cases where the spectrum could not be fit directly . to compare the derived x - ray luminosities with blue luminosities we used the observed blue magnitudes of de vaucouleur et al .
( 1991 via ned ; see also smith et al .
for the extinction correction we converted the galactic @xmath25 as given in dickey & lockman ( 1990 ) into @xmath102 assuming a standard gas / dust ratio , the relation of bohlin et al .
( 1978 ; see also predehl & schmitt 1995 ) , and the extinction curve as given in osterbrock et al .
( 1989 , his tab .
this yields @xmath102 = 0.38@xmath12 .
@xmath95 was then calculated using @xmath103 and assuming the same @xmath6=0.017 for all galaxies .
results are listed in table 2 .
the ngc383 group turns out to be the brightest group in x - ray luminosity when compared to the samples of ponman et al .
( 1996 ) and mulchaey & zabludoff ( 1998 ) .
how does it fit into the known @xmath104 relation for groups and clusters of galaxies ( e.g. , fabian et al .
1994 , white 1996 , ponman et al .
1996 , arnaud & evrard 1998 , reiprich 1998 ) ?
since @xmath29 is outside the range for which ponman et al . find a very steep @xmath104 dependence , we use the relation of markevitch ( 1998 ) which predicts a ( 0.1 - 2.4 kev ) x - ray luminosity @xmath5 erg / s .
this agrees well with the observed value of @xmath105 erg / s . for the given @xmath27 = 1.5 kev , and a galaxy velocity dispersion of @xmath21 = 466 km / s ( ledlow et al .
1996 ; see also sakai et al .
1994 ) we derive @xmath106 .
the low value of the slope parameter @xmath38 derived from the x - ray spatial analysis is in line with earlier findings of a trend of decreasing @xmath38 toward lower @xmath29 ( e.g. , david et al .
1990 , white 1991 , arnaud & evrard 1998 ) which is reproduced by models of cluster formation that incorporate galactic winds ( e.g. , metzler & evrard 1997 ) . assuming spherical symmetry and isothermality , the total gravitating mass within 1 mpc amounts to @xmath52 = 0.610@xmath8 m@xmath2 . we find a gas mass fraction of 21% which is at the upper end of the values typically observed in groups ( 325% , e.g. , david et al .
1995 , bhringer 1995 ) but not inconsistent with similar results for such x - ray luminous groups as for example ngc533 or ngc4104 ( mulchaey et al .
the gas mass fraction slightly decreases with decreasing radius .
[ gprop ] l spectral fits : + @xmath27=1.5 kev , @xmath107 = 1.510@xmath34 erg / s + beta - model results : + @xmath108 cts / s / arcmin@xmath41 , @xmath109 , @xmath110 kpc + central density , mass : + @xmath49 = 1.310@xmath50 @xmath24 ; @xmath52 = 0.610@xmath8 m@xmath2 , + gas mass fraction 21% ( at @xmath111 = 1mpc ) + concerning the morphology of the igm , we note that the rather spherically symmetric shape of the extended emission , as compared to the completely different morphology defined by the bright ellipticals aligned in a chain , argues against an origin of the gas in terms of halos of the chain galaxies , but rather for an association with the global group potential .
the x - ray emission of the igm can be traced out to about 1 mpc , which turns out to be about the virial radius of the group if we use the total mass value determined from the x - ray observations and the assumption that the virial radius is approximately characterized by the region inside which the mean overdensity is a factor of 200 above the critical density of the universe ( e.g. , evrard et al .
1996 ) . the fairly spherically symmetric appearance of the group s x - ray halo ( except possibly for some faint outer extensions ) taken together with the perfect consistency of the mass determined from the velocity dispersion and the x - ray properties , and the findings of ledlow et al .
( 1996 ) that the galaxy velocity distribution does not significantly deviate from a gaussian , implies that the matter in the group inside a radius of about 1 mpc is most probably quite relaxed .
the position of the central galaxy ngc383 is found to be slightly off - set from the center of the extended x - ray emission , and thus presumably from the center of the dark matter potential .
such off - sets of cd galaxies have also been observed in a number of other poor systems ( e.g. , awm7 , neumann & bhringer 1995 ; fornax cluster , ikebe et al .
1996 ) and some abell clusters ( e.g. , lazzati & chincarini 1998 ) . in lazzati & chincarini ( 1998 ) ,
this is traced back to a small - amplitude oscillation of the cd galaxy around the bottom of the cluster potential .
the cooling time in the center is @xmath112 yr , i.e. no ` large - scale ' cooling flow is expected to have developed .
further , the enhanced x - ray emission from the direction of ngc383 is consistent with originating from a point source .
we also note that although strong low - ionization optical emission lines have been reported for some central galaxies in cluster cooling flows ( e.g. , cowie et al .
1983 , heckman 1989 , crawford & fabian 1992 ) , the morphology of the ring- or disk - like emission line region in ngc383 ( owen et al .
1990 , fraix - burnet et al .
1991 ) does not argue for a connection to a cooling flow .
further , we find the locus of ngc383 in the emission line - ratio diagram [ sii ] vs. [ nii ] to ly outside of the ` class i ' and ` class ii ' cooling flow nebulae of heckman et al .
( 1989 ; see their fig .
( this alone does not exclude the presence of a cooling flow , though , since not all of them are associated with emission line nebulae . )
although 3c31 is only a moderately bright radio galaxy , its jets could be studied in detail due to its proximity .
concerning the origin , morphology and confinement of the jets , several models were explored ( e.g. , blandford & icke 1978 , bridle et al .
1994 , bicknell 1994 ) .
the gas density and temperature derived for the x - ray gas allow a comparison with the pressure of the radio gas , and an assessment of the confinement of the jet material . in fig .
[ pressure ] we compare the pressure of the radio emitting region as given in strom et al .
( 1983 ) and morganti et al .
( 1988 ) with the thermal pressure of the x - ray gas derived from the run of density ( sect .
3.3 ) and a temperature of @xmath27=1.5 kev .
whereas in the central region ( i.e. within ngc383 ) , there seems to be a strong overpressure of the radio gas ( but note that there certainly is an additional contribution to thermal pressure from higher - density gas within the galaxy ) , pressure equilibrium is reached at about 35 kpc ( projected distance from the center ) .
further out the thermal pressure increasingly exceeds the nonthermal pressure .
it is interesting to note that bridle et al .
( 1980 ) find the expansion rate of the jets of 3c31 transverse to their length to decrease with increasing distance from the radio core . this may be related to the relative increase of the thermal pressure of the ambient medium with increasing radius .
while bridle s trend refers mainly to the presently in x - rays barely resolved core region it may be interesting to explore this relation further with higher - resolution x - ray data .
a similar comparison of thermal vs. non - thermal pressure was performed in t97 .
they derived a somewhat different surface brightness profile and thus change in thermal pressure with the consequence that the radius where both pressure values are of the same order shifts further out which led them to suggest that ngc383 might have a giant halo that escaped detection in the ( short ) hri exposure .
the galaxies of the chain are among those with high @xmath114 and show a larger spread in @xmath115 than in @xmath95 as often observed ( e.g. , eskridge et al .
1995 ) . the high @xmath115 in cluster / group ellipticals ( e.g. , fig . 1 of brown & bregman 1998 , fig . 2 of beuing et al . 1998 , fig . 1 of irwin & sarazin 1998 )
are usually traced back to the influence of the surroundings , via accretion of gas from the group environment ( e.g. , beuing et al .
1998 ) , or stiffling of winds by the ambient medium ( brown & bregman 1998 ) ; see also mathews & brighenti ( 1998 ) .
the low @xmath116 systems seem to be dominated by discrete sources , mainly lmxbs ( e.g. , canizares et al .
1987 , irwin & sarazin 1998 ) .
the present galaxies show a wide range in @xmath115 reaching up to the high end of the @xmath115/@xmath95 relation .
[ lxlb ] ; note that the objects of very high @xmath115 of the sample of beuing et al .
1998 are all ellipticals in the _ cluster _ environment ) .
the x - ray luminosities generally exceed those expected from a discrete source contribution and the x - ray properties of most galaxies are consistent with the bulk of the x - ray emission arising from the ism of the individual galaxies .
ngc383 is separately discussed in the next section .
the x - ray analysis suggests the presence of two components in the x - ray spectrum of ngc383 : emission from a rs plasma of @xmath4 0.40.7 kev and a second component that was parameterized as powerlaw or second rs contribution .
compact pl emission could originate from an agn or via ssc at the base of the radio jet ; e.g. , jones et al .
( 1974 ) , marscher ( 1987 ) , birkinshaw & worrall ( 1993 ) . using the correlation of 1kev x
- ray core flux with 5ghz radio core flux of worrall et al .
( 1994 , their fig .
3 ; see also fabbiano et al . 1984 )
we find 3c31 above this relation ( a factor @xmath116 in x - ray flux for the given radio core flux for which we derive a 1kev x - ray flux in the pl component of 0.058 @xmath117jy ] ) , and within the region populated by lobe - dominated quasars .
is there evidence for a relatively _ unobscured _ view on an agn ?
optical emission line ratios were presented by owen et al .
( 1990 ) , who found the line ratios in the disk to be similar to those at the nucleus ( [ nii]/h@xmath35=1.5 , [ sii]/h@xmath35=0.5 , [ oiii]/h@xmath35=0.1 ) and concluded that the disk ionization is probably driven by the nucleus . plotting these line ratios in the diagnostic diagrams of veilleux & osterbrock ( 1987 ) we find them to be liner - like .
the dominant excitation mechanism in liners is still under discussion , but there is growing evidence that the bulk of them is accretion - powered ( e.g. , ho et al . 1998 ) .
owen et al . also briefly mention the presence of a broad emission line component .
alternatively , two - component rs models fit the x - ray spectrum ; a second component may originate from non - perfect correction of the cluster contribution or a temperature range in emission from the ism of the galaxy .
in fact , more than one temperature component is favoured for nearly every early - type galaxy examined in , e.g. , matsushita et al .
( 1994 ) , matsumoto et al .
( 1997 ) and buote & fabian ( 1998 ) . whereas the hot component is occasionally interpreted in terms of emission from discrete sources ( e.g. , matsushita et al .
1994 ) , we find its luminosity ( @xmath57 erg / s , @xmath118 erg / s if parameterized by a pl or rs model , respectively ) to be higher than expected from discrete sources ( e.g. , canizares et al .
1987 ) for the given @xmath95 . in conclusion ,
although the exact shape and strength of the hard x - ray component can not be determined with present data , its presence and high luminosity suggest as origin emission from an active nucleus in ngc383 .
the source 1e0104 was serendipitously detected as bright x - ray emitter by and was found to be close to a bal quasar with @xmath6=2.027 ( stocke et al .
since the qso is also located very near ( 10 ) to a giant elliptical galaxy at @xmath6=0.111 , which is part of a small group , it remained unclear from which of the three the x - ray emission originated : the qso , the giant elliptical , or the igm of the small group ( stocke et al .
gioia et al .
( 1986 ) , in a deep observation , could not detect x - ray emission in the 0.052 kev band , which they traced back to either variability ( intrinsic to the source , or caused by a microlensing event during the observation ; which would favour the quasar identification of the x - ray source ) , or a strongly absorbed x - ray spectrum .
the incidental pspc observation of the source was briefly discussed by ciliegi & maccacaro ( 1996 ) in a large sample study of spectra of emss agn .
they applied a powerlaw to the spectrum and found excess absorption .
below , we analyze the x - ray spectral properties in more detail and also perform a timing analysis , in order to get clues on the origin of the x - ray emission ( elliptical galaxy , group or quasar ) .
the hri data improve the position of 1e0104 , but not much , since the source is weak and located near the border of the fov . in fig .
[ pos ] we compare the hri position with the optical positions of the quasar and the galaxies ( as given in stocke et al .
1984 ; all coordinates were converted to j2000 ) and the ipc position .
both sources , the quasar and the elliptical , remain within the error circle .
we note , however , that the x - ray source appears to be extended .
although the analysis is complicated by a newly detected closeby second source , a comparison with the x - ray emission from the f star at nearly the same off - axis angle , clearly suggests that the x - ray emission from 1e0104 does not originate from a point source ( see fig . 1 ; the f star is the brightest source at the right border of the image , 1e0104 is the one at the lower border ) .
we do not detect short - time variability during the pspc observation ( fig .
[ light ] ) . to examine the long - term trend , we converted the pspc countrate to hri countrate assuming constant spectral shape .
again , the source emission is found to be constant .
we then compared with the ipc flux ( this obs .
was performed 11 yrs prior to the pspc observation ) given in stocke et al .
( 1984 ) . to this end
, we first re - fit the data with the same spectral model as in stocke et al .
( they used = 1.5 and @xmath119 @xmath69 ; see gioia et al .
1986 for similar parameters ) .
we then derive an observed flux ( converted to the 0.3 3.5 kev band ) of @xmath120 erg / cm@xmath41/s which is in excellent agreement with the flux of @xmath121 erg / cm@xmath41/s
. a powerlaw with absorption fixed to the galactic value does not give a good fit with = 1.7 and @xmath54 = 1.5 . the fit
improves after allowing for excess cold absorption .
this yields @xmath122 @xmath69 , = 3.2 and @xmath54 = 1.1 .
alternatively , the spectrum can be successfully described in terms of a rs model . in this case
we used @xmath6=0.111 since rs emission is more likely to originate from the elliptical and/or igm .
if @xmath25 is treated as free parameter it is consistent with @xmath26 but comes with large errors .
we obtain @xmath123 @xmath69 and @xmath124 kev ( @xmath54 = 1.0 ) . for @xmath25=@xmath26 , @xmath27 = 1.9 kev and @xmath125 erg / s . plugging this temperature into the @xmath126 relation for groups / clusters ( markevitch 1998 ) we predict a ( 0.12.4 kev ) x - ray luminosity of 510@xmath34 erg / s which is of the order of the observed value and thus consistent with an identification of 1e0104 with the igm of the small group .
the x - ray emission of 1e0104 turned out to be constant . in particular
, there is excellent agreement with the flux derived for the observation ( adopting the same spectral model ) .
this renders one of several possibilities discussed in stocke et al .
( 1984 ) the one that the quasar was temporarily brightened via microlensing during the observation unlikely .
we confirm the excess cold absorption in case of the pl description of the x - ray spectrum , not unexpected for a bal quasar , but find an rs model to be similarly successful .
the best - fit parameters of the latter model , which due to its shape would argue for an origin from the group s igm or brightest group elliptical , are well consistent with the known @xmath115-@xmath29 relation of clusters and groups of galaxies , i.e. , with the igm as counterpart .
further , the derived @xmath5 erg / s would be rather high for an elliptical outside a rich cluster environment ( particularly for the observed ( stocke et al .
1984 ) r - magnitude of @xmath127 = 15.1 ) , but usual for the igm .
the hri position is intermediate between the elliptical and the quasar , and consistent with all three identifications .
however , there is evidence that the source emission is extended . in summary ,
if only one emitter dominates the x - ray flux ( instead of a contribution from _ all three _ potential counterparts , which would strongly complicate matters but can not be excluded ) it seems that the available data ( constancy in source flux , high @xmath115 , fit into @xmath115-@xmath29 relation , evidence for source extent ) favour an identification with the igm of the nearby group at @xmath6=0.111 .
a deep high spatial resolution observation _ centered _ on 1e0104 should finally resolve the counterpart question .
we have presented a study of the x - ray properties of the ngc383 group of galaxies , extending the work of t97 , and of the source 1e0104 + 3135 .
the properties of the intra - group medium derived from the pspc observation are summarized in tab .
3 . the x - ray emission of the igm can be traced out to about @xmath128 mpc , which turns out to be about the virial radius of the group .
several lines of evidence were presented that the group inside this radius is quite relaxed . with the given depth of the pspc observations we can therefore characterize the entire galaxy system as far as it has approached a dynamical equilibrium state . for this part of the group we find a total mass of @xmath1 m@xmath2 and a gas mass fraction of 21% .
the latter value is at the upper end of the distribution for groups .
the result implies that the gravitational potential of the group is deep enough to prevent a majour gas loss , in contrast to less massive groups ( davis et al .
1998 ) .
the surface brightness profile is characterized by a slope parameter @xmath129 which is shallower than for most of the galaxy clusters but a quite common value for groups .
the temperature of @xmath3 kev found for the igm is well consistent with the @xmath130 relation giving further support to the picture that the group is a well relaxed and normal system . with an estimated central cooling time larger than the hubble time
no central cooling flow is expected and no signature for it is found neither in the spatial and spectral x - ray data nor in the optical spectrum .
we also discussed the x - ray properties of the well - studied radio galaxy 3c31 which is located near the center of the extended x - ray emission .
if one wishes to avoid excessively depleted metal abundances , the spectrum of 3c31 is best described by a two - component model , consisting of a low - temperature rs component and a hard tail ( pl or second rs ) , confirming t97 .
the soft component contributes with @xmath131 erg / s to the total x - ray luminosity of @xmath132 erg / s ( assuming metal abundances of 0.35@xmath22solar ) .
no temporal variability in the x - ray flux is detected .
x - ray emission from the direction of the interesting source 1e0104 + 3135 , by chance located in the field of view , was analyzed .
one scenario ( transient brightening of the @xmath6=2 bal qso0104 + 3135 due to lensing during the earlier observation ) turned out to be unlikely . besides an absorbed powerlaw , the spectrum of 1e0104 can be described by a raymond - smith model with @xmath4 2 kev resulting in an intrinsic luminosity of @xmath5 erg / s at @xmath6=0.111 .
although no potential counterpart ( qso , nearby elliptical galaxy or igm of the small group to which the elliptical belongs ) can be safely ruled out at present , there are several hints ( constancy in source flux , high @xmath115 , consistency with @xmath115-@xmath29 relation for groups / clusters , evidence for source extent ) for an identification of the x - ray source with the igm of the nearby group of galaxies at @xmath6=0.111 .
we acknowledge support from the verbundforschung under grant no .
it is a pleasure to thank vadim burwitz for help with figure 4 .
the project is supported by the german bundesministerium fr bildung , wissenschaft , forschung und technologie ( bmbf / dlr ) and the max - planck - society .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , caltech , under contract with the national aeronautics and space administration . | we present results from an analysis of the x - ray properties of the ngc383 galaxy group based on pspc and hri data .
x - ray emission can be traced out to @xmath0 mpc , the estimated virial radius of the system .
we determine a total mass of @xmath1 m@xmath2 for the group inside this radius with a gas mass fraction of 21% .
the intragroup gas temperature of @xmath3 kev is both consistent with the galaxy velocity dispersion and the x - ray luminosity - temperature relation of groups and clusters suggesting that the group is fairly relaxed .
this is also indicated by the almost spherically symmetric appearance of the group s x - ray halo .
the x - ray properties of the radio galaxy ngc383 ( 3c31 ) which is located near the center of the group are discussed .
its spectrum is best described by a two - component model , consisting of emission from a low - temperature raymond - smith plasma , and a hard tail . the emission from ngc383
is not resolved by the hri .
the possible interaction of the radio jets of 3c31 with the igm is studied .
a spatial , spectral and temporal analysis of the source 1e0104 + 3153 located within the field of view is performed , one goal being the identification of the optical counterpart ( with both , a high - redshift bal quasar and a nearby elliptical galaxy , member of a small group , located within the x - ray error circle ) .
we find evidence that the igm of the small group contributes significantly to the x - ray emission of 1e0104 , which can be described by a raymond - smith model of @xmath4 2 kev and a soft x - ray luminosity of @xmath5 erg / s . |
the problem of the existence of dark matter in our universe dates back to the astrophysical observations at the beginning of past century @xcite , but the presence of dark matter in our universe has been definitively accepted by the scientific community only about 40 years later , when two groups performed systematic measurements of the rotational velocities of celestial bodies in spiral galaxies @xcite . after the 70 many other observations have further confirmed the presence of dark matter in the universe and , at present , the measurements are mainly devoted to the investigation of the quantity , of the distribution ( from the cosmological scale down to the galactic one ) and of the nature of the dark matter in the universe .
recent measurements of the cmb temperature anisotropy by wmap @xcite , analysed in the framework of the _ big bang _ cosmological scenario , support a density of the universe : @xmath3 = 1 , further crediting that most of the universe is dark .
recently , it has been suggested from observations on the supernovae ia at high red - shift as standard candles that about 73% of @xmath3 might be in form of a _ dark energy _
@xcite ; the argument is still under investigation and presents some problems on possible theoretical interpretations . however , even in this scenario where the matter density in the universe would be @xmath4 @xcite large space for dark matter particles in the universe exists . in fact
, the luminous matter can only account for a density @xmath5 and the baryonic dark matter for @xmath6 . on the other hand , the contribution of dark matter particles relativistic at the decoupling time
is also restricted to be @xmath7 by considerations on large scale structure formations @xcite .
thus , most of the dark matter particles in the universe , relics from the big bang , were non relativistic at decoupling time ; they are named cold dark matter particles ( cdm ) .
the cdm candidates have to be neutral , stable or quasi - stable ( e.g. with a time decay of order of the age of the universe ) and have to weakly interact with ordinary matter .
these features are respected by the axions ( also investigated by dama / nai @xcite ) and by a class of candidates named wimps ( weakly interacting massive particles ) . in particular , in the standard model of particle physics no particle can be a suitable candidate as cdm ; thus , a new window beyond the standard model can be investigated .
finally , several observations have suggested that our galaxy should also be embedded in a dark halo with mass at least 10 times larger than that of the luminous matter .
the verification of the presence of a dark matter particle component in the galactic halo has been the main goal of the dama / nai experiment .
the dama / nai experiment was proposed in 1990 @xcite , designed and realized having the main aim to investigate in a model independent way the presence of a dark matter particle component in the galactic halo @xcite . for this purpose
, we planned to exploit the effect of the earth revolution around the sun on the dark matter particles interactions on the target - nuclei of suitable underground detectors .
in fact , as a consequence of its annual revolution , the earth should be crossed by a larger flux of dark matter particles in june ( when its rotational velocity is summed to the one of the solar system with respect to the galaxy ) and by a smaller one in december ( when the two velocities are subtracted ) ( see fig .
[ fg : annmod1 ] ) .
this offers an efficient model independent signature , able to test a large interval of cross sections and of halo densities ; it is named _ annual modulation signature _ and was originally suggested in the middle of 80 by @xcite . in fact , the expected differential rate as a function of the recoil energy , @xmath8 ( see ref .
@xcite for detailed discussion ) , depends on the wimp velocity distribution and on the earth s velocity in the galactic frame , @xmath9 . projecting @xmath9 on the galactic plane , one can write : @xmath10 . here
@xmath11 is the sun s velocity with the respect to the galactic halo ( @xmath12 km / s and @xmath13 is the local velocity whose value is in the range 170 - 270 km / s @xcite ) ; @xmath14 = 30 km / s is the earth s orbital velocity around the sun on a plane with inclination @xmath15 = 60@xmath16 with the respect to the galactic plane .
furthermore , @xmath17= 2@xmath18/t with t=1 year and roughly t@xmath19 @xmath0 2@xmath20 june ( when the earth s speed is at maximum ) .
the earth s velocity can be conveniently expressed in unit of @xmath13 : @xmath21 , where depending on the assumed value of the local velocity @xmath22=1.04 - 1.07 is the yearly average of @xmath23 and @xmath24 = 0.05 - 0.09 . since @xmath25
, the expected counting rate can be expressed by the first order taylor approximation : @xmath26 = \frac{dr}{de_r}[\eta_0 ] + \frac{\partial}{\partial \eta } \left ( \frac{dr}{de_r } \right)_{\eta = \eta_0 } \delta \eta \cos\omega(t - t_0 ) .\ ] ] averaging this expression in a @xmath27-th energy interval one obtains : @xmath28 with the contribution from the highest order terms less than 0.1@xmath29 .
thus , the annual modulation signature is very distinctive since a wimp - induced seasonal effect must simultaneously satisfy all the following requirements : ( i ) the rate must contain a component modulated according to a cosine function ; ( ii ) with one year period ; ( iii ) a phase that peaks roughly around @xmath0 2@xmath20 june ; ( iv ) this modulation must only be found in a well - defined low energy range , where wimp induced recoils can be present ; ( v ) it must apply to those events in which just one detector of many actually `` fires '' , since the wimp multi - scattering probability is negligible ; ( vi ) the modulation amplitude in the region of maximal sensitivity must be @xmath307@xmath29 for usually adopted halo distributions , but it can be larger in case of some possible scenarios such as e.g. those in refs .
only systematic effects able to fulfil these 6 requirements and to account for the whole observed modulation amplitude could mimic this signature ; thus , no other effect investigated so far in the field of rare processes offers a so stringent and unambiguous signature .
with the present technology , the annual modulation signature remains the main signature of a wimp signal .
of course , the amount of the measured effect depends e.g. : i ) on the sensitivity of the experiment to the coupling and to the particle physics features of the wimp ; ii ) on the nuclear features of the used target - nucleus ; iii ) on the features of the dark halo and , in particular , on the spatial and velocity distributions of the dark matter particles .
the quality of the running conditions also plays a crucial role to detect such a rare process .
considering its main goal , dama / nai has been designed by employing and further developing all the necessary low - background techniques and procedures .
a detailed description of the set - up , of its radiopurity , of its performance , of the used hardware procedures , of the determination of the experimental quantities and of the data reduction has been given in refs .
@xcite . here
only few arguments are addressed .
the dama / nai experiment was located deep underground in the gran sasso national laboratory of i.n.f.n .
, 6.5 kg pure liquid xenon scintillator ) @xcite ; ii ) dama / r&d , set - up devoted to tests on prototypes and small scale experiments @xcite ; iii ) the new second generation large mass nai(tl ) radiopure set - up dama / libra ( see later ) ; iv ) dama / ge detector for sample measurements . ]
whose main features have been reported in @xcite .
in particular , the nai(tl ) scintillator was chosen as the best target material to investigate the process since it offers e.g. : * well known technology * reachable high radiopurity by material selections and protocols , by chemical / physical purifications , etc . * large mass * high duty cycle * feasible well controlled operational conditions and monitoring * routine calibrations feasible down to kev range in the same conditions as the production runs * high light response , that is kev threshold reachable * absence of the necessity of re - purification or cooling down / warming up procedures ( implying high reproducibility , high stability , etc . ) * absence of microphonic noise and an effective noise rejection at threshold ( time decay of nai(tl ) pulses is hundreds ns , while that of noise pulses is tens ns ) * sensitivity to spin - independent ( si ) , spin - dependent ( sd ) and mixed ( si&sd ) couplings as well as to several other existing scenarios * sensitivity to both high ( by iodine target ) and low ( by na target ) mass candidates * possibility to effectively investigate the annual modulation signature in all the needed aspects * pulse shape discrimination feasible at reasonable level * possibility to achieve significant results on several other rare processes * no safety problems * necessity of a relatively small underground space * the lowest cost with the respect to every other considered technique however , neither commercial low background nai(tl ) detectors nor nai(tl ) detectors grown with old technology ( even after `` revision '' ) can reach the needed sensitivity ; thus , a devoted special r&d was realized with the crismatec company . in this framework ,
all the needed materials were selected and through devoted and severe protocols the final dama / nai highly radiopure detectors were realized .
the performances of the nine 9.7 kg highly radiopure dama / nai detectors , including their measured radiopurity , are discussed in details in @xcite .
the bare nai(tl ) crystals are encapsulated in radiopure cu housings ; moreover , 10 cm long tetrasil - b light guides act as optical windows on the two end faces of each crystals and are coupled to specially developed low background photomultipliers ( pmt ) .
the measured light response is 5.5 7.5 photoelectrons / kev depending on the detector .
the two pmts of a detector work in coincidence with hardware thresholds at the single photoelectron level in order to assure high efficiency for the coincidence at few kev level .
the energy threshold of the experiment , 2 kev , is instead determined by means of x - rays sources and of kev range compton electrons on the basis also of the features of the noise rejection procedures and of the efficiencies when lowering the number of available photoelectrons @xcite .
the detectors are enclosed in a sealed copper box , continuously maintained in hp nitrogen atmosphere in slightly overpressure with respect to the external environment .
a suitable low background hard shield against electromagnetic and neutron background was realized using very high radiopure cu and pb bricks @xcite , cd foils and 10/40 cm polyethylene / paraffin ; the hard shield is also sealed in a plexiglas box and maintained in the high purity ( hp ) nitrogen atmosphere .
moreover , about 1 m concrete ( made from the gran sasso rock material ) almost fully surrounds the hard shield outside the barrack and at its bottom , acting as a further neutron moderator .
a three - level sealing system from environmental radon is effective .
in fact , the inner part of the barrack , where the set - up is allocated , has the floor ( above the concrete ) and all the walls sealed by supronyl ( permeability : 2 @xmath1 10@xmath31 @xmath32/s
@xcite ) plastic and the entrance door is air - tight .
a low level oxygen alarm informs the operator before entering the inner part of the barrack since the hp nitrogen which fills both the inner cu box and the external plexiglas box is released in this closed environment .
the radon level inside the ( sealed ) barrack is continuously monitored and recorded with the production data @xcite .
on the top of the shield a glove - box ( also maintained in the hp nitrogen atmosphere ) is directly connected to the inner cu box , housing the detectors , through cu pipes .
the pipes are filled with low radioactivity cu bars ( covered by 10 cm of low radioactive cu and 15 cm of low radioactive pb ) which can be removed to allow the insertion of radioactive sources for calibrating the detectors in the same running condition , without any contact with external environment @xcite .
the whole installation is air - conditioned and the operating temperature as well as many other parameters are continuously monitored and acquired with the production data .
moreover , self - controlled computer processes automatically monitor several parameters and manage alarms @xcite .
the electronic chain and the data acquisition system operative up to summer 2000 have been described in ref .
@xcite , while the new electronics and daq installed in summer 2000 have been described in ref .
@xcite .
as regards other aspects , we recall that the linearity and the energy resolution of the detectors have been investigated using several radioactive sources @xcite such as , for the low energy region , @xmath33fe ( 5.9 kev x - rays ) , @xmath34cd ( 22 kev x - rays and 88 kev @xmath15 line ) and @xmath35am ( 59.5 kev @xmath15 line ) sources . in particular , in the production runs , the knowledge of the energy scale is assured by periodical calibrations with @xmath35am source and by monitoring ( in the production data themselves summed every @xmath0 7 days ) the position and energy resolution of the 46.5 kev @xmath15 line of the @xmath36pb @xcite .
the latter peak was present at level of few counts per day per kg ( cpd / kg ) in the measured energy distributions mainly because of a contamination ( by environmental radon ) of the external surface of the crystals cu housings , occurred during the first period of the underground storage of the detectors .
the only procedure applied to the data regards the noise rejection , which is particularly efficient since the difference in time decay of the nai(tl ) scintillation pulses and of the noise pulses is of order of hundreds ns ; the procedure have been described e.g. in refs . @xcites veto time ; see above ) induced by high energy events or from any possible @xmath37erenkov pulse in the light guide or in the pmts ; in fact , they also have fast time decay ( of order of tens ns ) . ] .
all the periodical long calibration procedures @xcite and the time specifically allocated for maintenance and/or for improvements are the main components affecting the duty cycle of the experiment . moreover , in the first period data have been taken only in the two extreme conditions for the annual modulation signature .
the energy threshold , the pmt gain and the electronic line stability are continuously verified and monitored during the data taking by the routine calibrations , by the position and energy resolution of the @xmath36pb line ( see above ) and by the study of the hardware rate behaviours with time .
finally , we remind that the experiment took data up to mev energy region despite the optimization was done for the kev energy range .
the measured low energy distributions of interest for the wimp investigation have been given in refs .
@xcite , where the corrections for efficiencies and acquisition dead time have already been applied . we note that usually in dama / nai the low energy distributions refer to those events where only one detector of many actually fires ( that is , each detector has all the others in the same installation as veto ; this assures a further background reduction , which is of course impossible when a single detector is used ) , _ single - hit _ events .
moreover , the data taking of each annual cycle has been started before the expected minimum of the signal rate ( which is roughly around @xmath0 2@xmath20 december ) and concluded after the expected maximum ( which is roughly around @xmath0 2@xmath20 june ) .
the dama / nai set - up has exploited the wimp annual modulation signature over seven annual cycles @xcite and the following part of this paper will summarize the final model independent result , some of the corollary quests for the candidate particle , implications and pespectives . however , it is worth to remind that - thanks to its radiopurity and features - dama / nai has also investigated other approaches for wimps in ref .
@xcite and several other rare processes such as : possible processes violating the pauli exclusion principle @xcite , cnc processes in @xmath38na and @xmath39i @xcite , electron stability @xcite , searches for neutral simps @xcite , for neutral nuclearites @xcite , for q - balls @xcite and for solar axions @xcite ; moreover , dama / nai allowed also the study of nuclear transition to a possible superdense nuclear state @xcite and possible two - body cluster decay of @xmath39i @xcite .
as reported in ref .
@xcite , a model independent investigation of the annual modulation signature has been realized by exploiting the time behaviour of the residual rates of the _ single - hit _ events in the lowest energy regions over the seven annual cycles ( total exposure : 107731 kg @xmath1 day ) @xcite .
these experimental residual rates of the _ single - hit _ events are given by : @xmath40 , where @xmath41 is the rate in the considered @xmath42-th time interval for the @xmath43-th detector in the @xmath27-th considered energy bin and @xmath44 is the rate of the @xmath43-th detector in the @xmath27-th energy bin averaged over the cycles .
the average is made on all the detectors ( @xmath43 index ) and on all the energy bins in the considered energy interval .
this model independent approach offers an immediate evidence of the presence of an annual modulation of the rate of the _ single - hit _ events in the lowest energy region over the seven annual cycles as shown in fig .
[ fig2 ] , where the energy and time behaviours of the _ single - hit _ residual rates are depicted . in particular ,
the data favour the presence of a modulated cosine - like behaviour at 6.3 @xmath2 c.l . ; in fact , their fit for the ( 26 ) kev larger statistics energy interval offers a modulation amplitude equal to @xmath45 cpd / kg / kev , a phase : @xmath46 days and a period : @xmath47 year , all parameters kept free in the fit .
the period and phase agree with those expected in the case of a wimp induced effect ( @xmath48 = 1 year and @xmath49 roughly at @xmath0 152.5-th day of the year ) .
the @xmath50 test on the ( 26 ) kev residual rate in fig .
[ fig2 ] disfavours the hypothesis of unmodulated behaviour giving a probability of @xmath51 ( @xmath52 = 71/37 ) .
moreover , if the experimental residuals of fig .
[ fig2 ] are fitted fixing the period at 1 year and the phase at @xmath53 june , the following modulation amplitudes are obtained : @xmath54 cpd / kg / kev , @xmath55 cpd / kg / kev and @xmath56
cpd / kg / kev for the ( 24 ) , ( 25 ) and ( 26 ) kev energy intervals , respectively . in fig .
[ fig3 ] the _ single - hit _ residual rate in a single annual cycle from the total exposure of 107731 kg @xmath1 day is presented for two different energy intervals ; as it can be seen the modulation is clearly present in the ( 26 ) kev energy region , while it is absent just above .
the same conclusion is obtained by investigating the data by means of the fourier analysis ( performed according to ref .
@xcite including also the treatment of the experimental errors and of the time binning ) ; in fact , the results shown in fig . [ fig4 ] show a clear peak for a period of 1 year in the ( 26 ) kev energy interval , while it is absent in the energy interval just above .
a quantitative investigation of the whole energy spectrum up to mev energy region has not shown modulation in any other energy interval ( see e.g. @xcite and few arguments given later ) . finally , in order to show if the modulation amplitudes are statistically well distributed in all the crystals , in all the annual cycles and in the energy bins , the distributions of the variable @xmath57
are reported in fig .
[ gaus1 ] .
the @xmath58 are the experimental modulation amplitudes for each detector , for each annual cycle and for each considered energy bin ( taken as an example equal to 0.25 kev ) , @xmath2 are their errors and the @xmath59 represent the mean values of the modulation amplitudes over the detectors and the annual cycles for each energy bin .
the left panel of fig .
[ gaus1 ] shows the distribution referred to the region of interest for the observed modulation , ( 26 ) kev , while the right panel includes also the energy region just above , ( 214 ) kev .
since this variable is distributed as a gaussian with an unitary standard deviation , the modulation amplitudes are statistically well distributed in all the crystals , in all the data taking periods and in all considered energy bins . in conclusion ,
the data satisfy all the peculiar requirements , given above , for the wimp model independent annual modulation signature . as mentioned , to mimic this signature , systematic effects or side reactions
should not only be able to fully account for the measured modulation amplitude , but also to satisfy the peculiar requirements as for a wimp induced effect .
a careful investigation of all the known possible sources of systematics and side reactions has been regularly carried out by dama / nai and published at time of each data release @xcite .
no effect able to mimic the signature has been found ; thus , upper limits ( 90% c.l . ) on the possible contributions to the modulated amplitude have been calculated as summarized in table [ tb : sist ] ; for a detailed quantitative discussion see ref . @xcite .
& ( 90%c.l . ) + radon & @xmath60 + temperature & @xmath61 + noise & @xmath62 + energy scale & @xmath62 + efficiencies & @xmath62 + background & @xmath63 + side reactions & @xmath64 + + [ tb : sist ] in particular , they can not account for the measured modulation amplitude because quantitatively not relevant and unable to satisfy all the peculiar requirements of the signature . for the sake of completeness , we remind that possible diurnal effects correlated both with the sidereal and with the solar time have already been excluded by the analysis reported in @xcite . in particular ,
as mentioned above , no modulation has been observed in the background ; in fact , the whole energy spectrum up to mev energy region has been analyzed and the presence of a background modulation in the whole energy spectrum has been excluded at a level much lower than the effect found in the lowest energy region @xcite .
this result already accounts also for the background component due to the neutrons and the radon ; nevertheless , further additional independent and cautious analyses to estimate their possible contribution have been given in ref .
in fact , it has been demonstrated in @xcite that a modulation of neutron flux
possibly observed by the icarus coll .
as reported in the icarus internal report tm03 - 01 can not quantitatively contribute to the dama / nai observed modulation amplitude , even if the neutron flux would be assumed to be 100 times larger than measured at lngs by several authors with different techniques over more than 15 years .
moreover , in no case the neutrons can mimic the signature since some of the peculiar requirements of the signature would fail .
similarly , any possible effect of the muon flux modulation reported by the macro experiment @xcite is excluded both by quantitative investigation @xcite and by inability to fulfil all the peculiarities of the signature .
as regards the possibility of a contribution from the radon , we remind that the dama / nai has three levels of insulation from the environmental air ( see above ) .
moreover , the radonmeter which continuosly recorded the radon level inside the barrack ( that is external to the hard shield and to the detectors ) typically measured values at level of its sensitivity and no modulation has been observed @xcite . to be on the safest side ,
even the possible presence of radon trace in the hp nitrogen atmosphere inside the cu box , housing the detectors , has been investigated by searching for the double coincidences of the gamma - rays ( 609 and 1120 kev ) from @xmath65bi radon daughter , obtaining an upper limit of : @xmath66 bq / m@xmath67 ( 90% c.l . ) .
it gives rise to the upper limit reported in table [ tb : sist ] when assuming an hypothetical 10% , modulation of possible radon in the hp nitrogen atmosphere of the cu box .
anyhow , it is worth to remark that in every case even a sizeable quantity of radon nearby a detector can not mimic the wimp annual modulation signature since some of the peculiarities of the signature would fail . for more detailed discussion
@xcite . to perform a further relevant investigation , in 1999
we proposed to renew the electronic chain of dama / nai removing the multiplexer system and equipping each detector with its own transient digitizer .
this occurred in summer 2000 , thus in the last two annual cycles the _ multiple - hits _ events and the _ single - hit _ events have been acquired and analyzed using the same identical hardware and the same identical software procedures .
the _ multiple - hits _ events class on the contrary of the _ single - hit
_ one does not include events induced by wimps since the probability that a wimp scatters off more than one detector is negligible .
the obtained results are depicted in fig .
[ fig_mul ] .
the fitted modulation amplitudes are : @xmath68
cpd / kg / kev and @xmath69 cpd
/ kg / kev for _ single - hit _ and _ multiple - hits _ residual rates , respectively .
thus , evidence of annual modulation is present in the _ single - hit _ residuals ( events class to which the wimp - induced recoils belong ) , while it is absent in the _ multiple - hits _ residual rate ( event class to which only background events belong ) . since the same identical hardware and the same identical software procedures
have been used to analyse the two classes of events , the obtained result offers an additional strong support for the presence of a dark matter particle component in the galactic halo further excluding any side effect either from hardware or from software procedures or from background . in conclusion ,
the presence of a dark matter particle component in the galactic halo is supported by dama / nai at 6.3 @xmath2 c.l . and
the modulation amplitude measured over the 7 annual cycles in nai(tl ) at the location of the gran sasso laboratory for the ( 2 6 ) kev energy region is @xmath45 cpd / kg / kev .
this is the experimental result of dama / nai .
it is model independent ; no other experiment whose result can be directly compared with this one is available so far in the field of dark matter investigation .
on the basis of the obtained model independent result , corollary investigations can also be pursued on the nature and coupling of the wimp candidate .
this latter investigation is instead model dependent and considering the large uncertainties which exist on the astrophysical , nuclear and particle physics assumptions and on the parameters needed in the calculations has no general meaning ( as it is also the case of exclusion plots and of the wimp parameters evaluated in the indirect detection experiments ) .
thus , it should be handled in the most general way as we have preliminarily pointed out with time passing in the past @xcite and we have discussed in some specific details in ref .
@xcite .
it is worth to note that the results presented in the following are , of course , not exhaustive of the many possible scenarios which at present level of knowledge can not be disentangled . some of the open questions are : i ) which is the right nature for the wimp particle ; ii ) which is its right couplings with ordinary matter ( mixed si&sd , purely si , purely sd or _ preferred inelastic _ ) ; iii ) which are the right form factors and related parameters for each target nucleus ; iv ) which is the right spin factor for each target nucleus ; v ) which are the right scaling laws ( let us remind that even for the neutralino case in a mssm framework with purely si interaction the scenario could be drastically modified as pointed out in ref .
@xcite ) ; vi ) which is the right halo model and related parameters ; vii ) which are the right values of the experimental parameters within their uncertainties ; etc . as regards , in particular , the dark matter particle - nucleus elastic scattering
, the differential energy distribution of the recoil nuclei can be calculated @xcite by means of the differential cross section of the wimp - nucleus elastic processes given by the sum of the si and the sd contributions . in the si case ,
the nuclear parameters can be decoupled from the particle parameters and the nuclear cross sections are usually scaled to a defined point - like si dark matter particle - nucleon cross section , @xmath70 . in the sd case the notations @xcite : @xmath71 , can be used , where @xmath72 are the effective wimp - nucleon coupling strengths for sd interactions .
the mixing angle @xmath73 is defined in the @xmath74 interval ; in particular , @xmath73 values in the second sector account for @xmath75 and @xmath76 with different signs . also in the sd case all the nuclear cross sections are usually scaled to a defined point - like sd dark matter particle - nucleon cross section , @xmath77 @xcite .
thus , the energy distribution of the recoil rate can be written as a function of @xmath70 , @xmath77 and @xmath73 .
another important parameter is the local density of the dark matter particles , @xmath78 , where @xmath79 is the local halo density and @xmath80 ( @xmath81 ) is the fractional amount of local density of dark matter particles . among the ingredients entering in the model dependent analyses
there are the nuclear si and sd form factors , which generally depend on the nature of the involved particle . in ref .
@xcite the existing uncertainties in the usually adopted formulation for the si case have been discussed as well as those for the sd one .
it has also shown that the sd case is even more uncertain since the nuclear and particle physics degrees of freedom can not be decoupled and a dependence on the assumed nuclear potential exists . moreover ,
further significant uncertainties in the evaluation of the sd interaction rate also arise from the adopted spin factor for the single target - nucleus .
in fact , the available calculated values are well different in different models and , in addition , at fixed model they depend on @xmath73 . thus , not only the target nuclei should have spin different from zero ( for example , this is not the case of ar isotopes ) to be sensitive to dark matter particles with a sd component in the coupling , but also well different sensitivities can be expected among odd - nuclei having an unpaired proton ( as e.g. @xmath38na and @xmath39i ) and odd - nuclei having an unpaired neutron ( as e.g. the odd xe and te isotopes and @xmath82ge ) .
another scenario also considered in the corollary dama / nai model dependent analyses is that of wimps with _ preferred inelastic _ scattering which has been suggested by @xcite . in this case
the dark matter particles can only inelastically scatter off nuclei going to excited levels with a @xmath83 mass splitting .
a specific model featuring a real component of the sneutrino ( for which the mass splitting naturally arises ) has been given in ref .
it has been shown that for this inelastic scattering a kinematical constraint exists which favours target - detectors media with heavy nuclei ( such as @xmath39i ) with the respect to those with lighter ones ( such as e.g. @xmath84ge ) .
in fact , this process can only occur if the particle velocity is larger than a threshold value ; this kinematical constraint becomes increasingly severe as the nucleus mass decreases @xcite .
moreover , this model scenario implies some interesting peculiar features when exploiting the annual modulation signature ; in fact with the respect to the case of wimp elastically scattering it would give rise to an enhanced modulated component of the signal with respect to the unmodulated one and to largely different behaviours with energy for both the components ( showing both higher mean values ) @xcite .
the differential energy distribution of the recoil nuclei in the case of inelastic processes is function of @xmath85 , @xmath86 and @xmath83 , analogoulsy as above other ingredients , as e.g. the form factor , play also a role @xcite .
as mentioned , the expected energy distribution for the scatterings of dark matter particles depends on @xmath87 and on the velocity distribution of the dark matter particles at earth s position .
the experimental observations regarding the dark halo of our galaxy do not allow to get information on them without introducing a model for the galaxy matter density .
the dark halo model widely used in the calculations carried out in the field of direct detection approaches is the naive isothermal sphere that corresponds to a spherical infinite system with a flat rotational curve . despite its simplicity
has favoured its wide use in the calculation of expected rate of dark matter particle - nucleus interaction , it does nt match with astrophysical observations and presents an unphysical behaviour .
in fact , the density profile has a singularity in the origin and implies a total infinite mass of the halo unless introducing some cut - off at large radii .
thus , the use of more realistic halo models is mandatory since the model dependent results significantly vary . an extensive discussion about some of the more credited realistic halo models has been reported in ref .
@xcite and have been considered in the results given in the following ( see table [ tb : halo ] ) . in particular , the considered halo model classes correspond to : i ) spherically symmetric matter density with isotropic velocity dispersion ( a ) ; ii ) spherically symmetric matter density with non - isotropic velocity dispersion ( b ) ; iii ) axisymmetric models ( c ) ; iv ) triaxial models ( d ) ; v ) moreover , in the case of axisymmetric models it is possible to include either an halo co - rotation or an halo counter - rotation .
a0 & isothermal sphere & + a1 & evans logarithmic & @xmath88 kpc + a2 & evans power - law & @xmath89 kpc , @xmath90 + a3 & evans power - law & @xmath91 kpc , @xmath92 + a4 & jaffe & @xmath93 , @xmath94 , @xmath95 , @xmath96 kpc + a5 & nfw & @xmath93 , @xmath97 , @xmath98 , @xmath99 kpc + a6 & moore et al . & @xmath100 , @xmath97 , @xmath101 , @xmath102 kpc + a7 & kravtsov et al . & @xmath103 , @xmath97 , @xmath104 , @xmath105 kpc + + + b1 & evans logarithmic & @xmath88 kpc + b2 & evans power - law & @xmath89 kpc , @xmath90 + b3 & evans power - law & @xmath91 kpc , @xmath92 + b4 & jaffe & @xmath93 , @xmath94 , @xmath95 , @xmath96 kpc + b5 & nfw & @xmath93 , @xmath97 , @xmath98 , @xmath99 kpc + b6 & moore et al . & @xmath100 , @xmath97 , @xmath101 , @xmath102 kpc + b7 & kravtsov et al . & @xmath103 , @xmath97 , @xmath104 , @xmath105 kpc + + c1 & evans logarithmic & @xmath106 , @xmath107 + c2 & evans logarithmic & @xmath88 kpc , @xmath107 + c3 & evans power - law & @xmath89 kpc , @xmath108 , @xmath109 + c4 & evans power - law & @xmath91 kpc , @xmath107 , @xmath92 + + d1 & earth on maj .
axis , rad .
anis . & @xmath110 + d2 & earth on maj .
axis , tang .
& @xmath111 + d3 & earth on interm .
axis , rad .
& @xmath110 + d4 & earth on interm .
axis , tang .
anis . & @xmath111 + [ tb : halo ] the parameters of each halo model have been chosen taking into account the available observational data .
thus , considering the allowed range for the local velocity of dark matter particles @xmath112 km s@xmath113 ( 90% c.l . )
, the allowed range of local density @xmath79 has been evaluated @xcite taking into account the following physical constraints : i ) the amount of flatness of the rotational curve of our galaxy , considering conservatively @xmath114 , where @xmath115 is the value of rotational curve at distance of 100 kpc from the galactic center ; ii ) the maximal non dark halo components in the galaxy , considering conservatively @xmath116 @xcite .
although a large number of self - consistent galactic halo models , in which the variation of the velocity distribution function is originated from the change of the halo density profile or of the potential , have been considered , still many other possibilities exist .
the proper knowledge of other quantities is also necessary such as e.g. the recoil / electron response ratio for the given nucleus in the given detector and energy range , ( named _ quenching factor _ ) .
of course , significant differences are often present in literature for the measured value of this recoil /electron response ratio even for the same nucleus in the same kind of detector as shown in ref .
@xcite . in conclusion , just as a corollary of the model independent result over the seven annual cycles , in the following some of the many possible model dependent quests for a dark matter candidate are summarized .
they have been obtained by considering the halo models previously mentioned for three of the possible values of the local velocity @xmath13 : 170 km / s , 220 km / s and 270 km / s and for the halo density values as in the prescriptions of ref .
the escape velocity has been maintained at the fixed value : 650 km / s .
of course , it is worth to note that the present existing uncertainties affecting the knowledge of the escape velocity will significantly extend allowed regions e.g. in the cases of _ preferred inelastic _ wimps and of light mass wimp candidates ; its effect would be instead marginal at large wimp masses .
all these scenarios have been investigated in some discrete cases either considering the mean values of the parameters of the used nuclear form factors and of the measured quenching factors ( case @xmath117 ) or adopting the same procedure as in refs . @xcite .
the latter one has been obtained by varying either : i ) the mean values of the measured @xmath38na and @xmath39i quenching factors @xcite up to + 2 times the errors ; ii ) the nuclear radius , @xmath118 , and the nuclear surface thickness parameter , @xmath119 , in the si form factor @xcite from their central values down to -20% ; iii ) the @xmath120 parameter in the considered sd form factor from the given value @xcite down to -20% ( case @xmath121 ) .
moreover , we have also considered one of the possible more extreme cases where the iodine nucleus parameters are fixed at the values of case @xmath121 , while for the sodium nucleus one considers : i ) @xmath38na quenching factor at the lowest value measured in literature ; ii ) the nuclear radius , @xmath118 , and the nuclear surface thickness parameter , @xmath119 , in the si form factor @xcite from their central values up to + 20% ; iii ) the @xmath120 parameter in the considered sd form factor from the given value @xcite up to + 20% ( case @xmath122 ) . finally , no restriction on the mass of the dark matter particle has been adopted in these analyses ; hence , we have just marked on those figures the lowest bound on the neutralino mass as derived from the lep data in the adopted supersymmetric schemes based on gut assumptions @xcite . in fact , other model assumptions are possible and would imply significant variations of some accelerators bounds , allowing neutralino mass down to 6 gev ( see e.g. the recent refs .
@xcite ) ; in addition , other low mass candidates can be considered as well .
it is worth to note that the lep model dependent mass limit when considered selects the wimp - iodine elastic scatterings as dominant because of the adopted scaling laws and of kinematical arguments , while dama / nai is intrinsically sensitive both to low and high wimp mass having both a light ( the @xmath38na ) and a heavy ( the @xmath39i ) target - nucleus .
the results presented by dama / nai on the corollary quests for the candidate particle over the seven annual cycles are calculated taking into account the time and energy behaviours of the _ single - hit _ experimental data .
in particular , the likelihood function requires the agreement : i ) of the expectations for the modulated part of the signal with the measured modulated behaviour for each detector and for each energy bin ; ii ) of the expectations for the unmodulated component of the signal with the respect to the measured differential energy distribution and - since ref . @xcite - also with the bound on recoils obtained by pulse shape discrimination from the devoted dama / nai-0 data@xcite .
the latter one acts in the likelihood procedure as an experimental upper bound on the unmodulated component of the signal and as a matter of fact as an experimental lower bound on the estimate of the background levels by the maximum likelihood procedure .
thus , the c.l.s , we quote for allowed regions , already account for compatibility with the measured differential energy spectrum and with the measured upper bound on recoils .
in particular , in the following for simplicity , the results of these corollary quests for the candidate particle are presented in terms of allowed regions obtained as superposition of the configurations corresponding to likelihood function values _
distant _ more than @xmath123 from the null hypothesis ( absence of modulation ) in each of the several ( but still a limited number ) of the possible model frameworks considered here .
obviously , these results are not exhaustive of the many scenarios possible at present level of knowledge ( e.g. for some other recent ideas see @xcite ) and larger sensitivities than those reported in the following would be reached when including the effect of other existing uncertainties on assumptions and related parameters @xcite . in the most general scenario
to which the dama / nai target nuclei are fully sensitive both the si and the sd components of the cross section vs @xmath124 for some of the possible @xmath86 and @xmath73 values .
four sd couplings are reported as examples : i ) @xmath73 = 0 ( @xmath76 = 0 and @xmath125 0 or @xmath126 ) corresponding to a particle with null sd coupling to neutron ; ii ) @xmath127 ( @xmath128 ) corresponding to a particle with the same sd coupling to neutron and proton ; iii ) @xmath73 = @xmath129 ( @xmath130 0 and @xmath75 = 0 or @xmath131 ) corresponding to a particle with null sd couplings to proton ; iv ) @xmath73 = 2.435 rad ( @xmath132 = -0.85 ) corresponding to a particle with sd coupling through @xmath133 exchange .
the case @xmath134 is nearly similar to the case iv ) .
inclusion of other existing uncertainties on parameters and models would further extend the regions ; for example , the use of more favourable form factors and/or of more favourable spin factors than the considered ones would move them towards lower cross sections.,height=340 ] are present . in this general scenario
the data give an allowed volume in the 4-dimensional space ( @xmath86 , @xmath135 , @xmath124 , @xmath73 ) .
[ fg : pan_sisd ] just shows slices of this 4-dimensional allowed volume in the plane @xmath135 vs @xmath124 for some of the possible @xmath73 and @xmath86 values for the considered model frameworks .
we just note that experiments using either even - spin target nuclei ( as ar and most of ge , xe , te isotopes ) or odd - spin ge , xe or te isotopes can not explore most of the allowed volume . from the given figures it is clear that at present either a purely si or a purely sd or mixed si&sd configurations are compatible with the experimental data of the seven annual cycles .
often the purely si interaction with ordinary matter is assumed to be dominant since e.g. most of the used target - nuclei are practically not sensitive to sd interactions ( instead , @xmath38na and @xmath39i nuclei are fully sensitive ) and the theoretical calculations and comparisons are even much more complex and uncertain . therefore , following the analogous procedure as the general case , we have exploited for the same model frameworks the purely si scenario , obtaining the allowed region in the plane @xmath86 and @xmath135 shown in fig .
[ fg : fig_si ]
_ left_. of course , best fit values of cross section and wimp mass span over a large range in the considered model frameworks . , @xmath136 ) .
the vertical dotted line represents a bound in case of a neutralino candidate when supersymmetric schemes based on gut assumptions are adopted to analyse the lep data ; the low mass region is allowed for neutralino when other schemes are considered and for every other wimp candidate ; see text .
while the area at wimp masses above 200 gev is allowed only for few configurations , the lower one is allowed by most configurations ( the colored region gathers only those above the vertical line ) @xcite .
the inclusion of other existing uncertainties on parameters and models would further extend the region ; for example , the use of more favourable si form factor for iodine alone would move it towards lower cross sections .
_ on the right : example of the effect induced by the inclusion of a sd component different from zero on allowed regions given in the plane @xmath137 vs @xmath86 .
_ in this example the evans logarithmic axisymmetric @xmath138 halo model with @xmath139 km / s , @xmath79 equal to the maximum value for this model and a given set of the parameters values ( see @xcite ) have been considered .
the different regions refer to different sd contributions for the particular case of @xmath140 : @xmath141 0 pb ( a ) , 0.02 pb ( b ) , 0.04 pb ( c ) , 0.05 pb ( d ) , 0.06 pb ( e ) , 0.08 pb ( f ) .
analogous situation is found for the other model frameworks.,title="fig:",height=170 ] , @xmath136 ) .
the vertical dotted line represents a bound in case of a neutralino candidate when supersymmetric schemes based on gut assumptions are adopted to analyse the lep data ; the low mass region is allowed for neutralino when other schemes are considered and for every other wimp candidate ; see text .
while the area at wimp masses above 200 gev is allowed only for few configurations , the lower one is allowed by most configurations ( the colored region gathers only those above the vertical line ) @xcite .
the inclusion of other existing uncertainties on parameters and models would further extend the region ; for example , the use of more favourable si form factor for iodine alone would move it towards lower cross sections .
_ on the right : example of the effect induced by the inclusion of a sd component different from zero on allowed regions given in the plane @xmath137 vs @xmath86 . _ in this example the evans logarithmic axisymmetric @xmath138 halo model with @xmath139 km / s , @xmath79 equal to the maximum value for this model and a given set of the parameters values ( see @xcite ) have been considered .
the different regions refer to different sd contributions for the particular case of @xmath140 : @xmath141 0 pb ( a ) , 0.02 pb ( b ) , 0.04 pb ( c ) , 0.05 pb ( d ) , 0.06 pb ( e ) , 0.08 pb ( f ) .
analogous situation is found for the other model frameworks.,title="fig:",height=170 ] moreover , configurations with @xmath135 even much lower than those shown in fig .
[ fg : fig_si ]
_ left _ are accessible in case an even small sd contribution is present in the interaction as shown as in an example in fig .
[ fg : fig_si ] _
right_. analogous situation is found for other model frameworks .
a comparison of the dama / nai purely si allowed region ( given in fig .
[ fg : fig_si ] _ left _ ) with the theoretical expectations for a purely si coupled neutralino candidate in a mssm with gaugino mass unification at gut scale released
is shown in fig .
[ figtheo ] as taken from ref @xcite .
= 2.1 in , @xmath124 ) with @xmath140 and @xmath142 ( @xmath133 coupling ) . for the definition of the vertical line and of the coloured area
see previous figure caption .
inclusion of other existing uncertainties on parameters and models would further extend the sd allowed regions .
for example , the use of more favourable sd form factors and/or more favourable spin factors would move them towards lower cross sections .
values of @xmath124 lower than those corresponding to these allowed regions are possible also e.g. in case of an even small si contribution.,title="fig : " ] , @xmath124 ) with @xmath140 and @xmath142 ( @xmath133 coupling ) . for the definition of the vertical line and of the coloured area
see previous figure caption .
inclusion of other existing uncertainties on parameters and models would further extend the sd allowed regions .
for example , the use of more favourable sd form factors and/or more favourable spin factors would move them towards lower cross sections .
values of @xmath124 lower than those corresponding to these allowed regions are possible also e.g. in case of an even small si contribution.,title="fig : " ] analogously , one can consider the pure sd coupling . in this scenario one
obtain an allowed volume in the 3-dimensional space ( @xmath86 , @xmath124 , @xmath73 ) .
just examples of some slices of this allowed volume at given @xmath73 is shown in fig .
[ fg : fig_puresd ] .
considerations similar to the first case hold . finally , also the inelastic dark matter particle scenario has been analised obtaining an allowed volume in the 3-dimensional space ( @xmath143 , @xmath86 , @xmath83 ) . for simplicity ,
[ fg : fig_inel ] shows just few slices of such an allowed volume at some given wimp masses . ,
@xmath83 , @xmath86 ) for some @xmath86 values for wimp with _ preferred inelastic _ interaction .
inclusion of other existing uncertainties on parameters and models would further extend the regions ; for example , the use of a more favourable si form factor for iodine and different escape velocity would move them towards lower cross sections @xcite.,height=264 ] we remind that in these calculations @xmath144 has been assumed at fixed value ( as in the previous cases ) , while its present uncertainties can play a significant role in this scenario of wimp with _ preferred inelastic _ scattering . in conclusion ,
the data of the seven annual cycles have also allowed corollary investigations on the features of the dark matter particle candidate in some of the many possible scenarios . however , although several scenarios have been investigated , these corollary analyses are not exhaustive at all because of the present poor knowledge on many astrophysical , nuclear and particle physics needed assumptions .
thus , further efforts on the topics are in progress and a relevant contribution is expected by the new dama / libra set - up , now running .
as already mentioned , no other experiment , whose result can be directly compared in a model independent way with that of dama / nai , is available so far in the field of dark matter direct detection .
in fact , most of the activities , started in the 90 s , are still at r&d stage and/or have released marginal exposures with the respect to the many years of existence and to the several used detectors .
this is the case of cdms and edelweiss experiments , while the zeplin experiment is more recent @xcite .
since these experiments have claimed to have `` excluded '' dama / nai , we will briefly point to the attention of the reader only few arguments .
in particular , table [ tb : modepcomp1 ] summarizes some items for comparison .
.features of the dama / nai results on the wimp annual modulation signature over the seven annual cycles @xcite with those of refs .
[ cols="^,^,^,^,^ " , ] [ tb : modepcomp1 ] firstly , let us preliminarily assume as fully correct the `` selected '' number of events , the energy threshold , the energy scale , etc . quoted by those experiments ( see table [ tb : modepcomp1 ] ) and
let us consider if at least under this hypothesis their claims might be justified .
the answer is obviously not ; in fact : i ) they give a single model dependent result using @xmath84ge or @xmath84xe target , while dama / nai gives a model independent result using @xmath38na and @xmath39i targets ; ii ) in the single ( of the many possible ) model scenario , they consider , they `` fix '' all the astrophysical , nuclear and particle physics assumptions at a single choice ; the same is even for the experimental and theoretical parameters values needed in the calculations .
in addition , dama / nai is generally quoted there in an uncorrect , partial and unupdated way and existing scenarios to which dama / nai is fully sensitive on the contrary of the others are ignored .
let us now briefly comment also some of the experimental aspects .
in particular , the counting rate of the ge bolometers experiments is very high and few / zero events are claimed after applying several strong and hardly safe rejection procedures ( involving several orders of magnitude ) .
they usually claim to have an `` event by event '' discrimination between _ noise + electromagnetic background _ and _ recoil + recoil - like ( neutrons , end - range alphas , fission fragments , ... ) _ events by comparing the bolometer and the ionizing signals for each event , but their results are , actually , largely based on `` a priori '' huge data selections and on the application of other preliminar rejection procedures ( such as e.g. the one on the so - called surface electrons ) , which are generally poorly described and often not completely quantified .
moreover , most efficiencies and physical quantities entering in the interpretation of the claimed selected events have never been properly accounted ; as an example , we mention the case of the bolometer quenching factor of the recoil target nuclei . in fact , for the bolometer signals the quenching factor ( on which the energy threshold and the energy scale rely and , hence , also the claimed sensitivity for the given model dependent exclusion plots ) is arbitrarily assumed to be exactly equal to one . up to now
, only one measurement has been made available for a given detector @xcite ; it offers the value : @xmath145 , which is within the error
compatible with one , but at the same time also compatible with much smaller values .
thus , any bolometer result , obtained without considering e.g. the uncertainties about the unknown value of the quenching factor and , hence , about the energy threshold and energy scale , has to be considered partial and arbitrary . for completeness
we also mention that the reproducibility of the results over different running periods has not been proved as well as the values of the effective sensitive volumes for the read - outs of the two signals for each event and related quantities ; obviously , further uncertainties are present when , as done in some cases , a neutron background modeling and subtraction is pursued in addition . as regards zeplin - i @xcite , a very low energy threshold is claimed ( 2 kev ) , although the light response is very poor : between @xmath0 1 ph.e./kev @xcite ( for most of the time ) and @xmath0 2.5 ph.e./kev ( claimed for 16 days ) @xcite .
moreover , a strong data filtering is applied to the high level of measured counting rate ( see table [ tb : modepcomp1 ] ) by hardware vetoes , by fiducial volume cuts and , largely , by applying down to few kev a standard pulse shape discrimination procedure , although the lxe scintillation pulse profiles ( pulse decay time @xmath146 30 ns ) are quite similar to the pmt noise events in the lower energy bins and in spite of the poor light response .
quantitative information on experimental quantities related to the used procedures has not yet been given @xcite .
in conclusion , those claims for contradiction have intrinsecally no scientific bases .
the dark matter particles , via their annihilation either in the celestial bodies ( such as earth and sun ) or in the galactic halo could give rise to high energy neutrinos , positrons , antiprotons and gamma rays .
therefore , they could be _ indirectly _ detected by looking either for `` upgoing '' muons produced by @xmath147 in underground , underwater or under - ice detectors or for antimatter and gamma rays in the space .
however , it is worth to remark that no direct model independent comparison can be performed between the results obtained in direct and indirect searches . in the case of `` upgoing '' muons in terrestrial detectors ,
the expected @xmath148 flux is the key quantity .
however , several sources of uncertainties are present in the related estimates ( and , therefore , in the obtained results ) such as e.g. the assumption that a `` steady state '' has been reached in the considered celestial body and the estimate and subtraction of the existing competing processes , offered by the atmospheric neutrinos .
model dependent analyses with a similar approach have been carried out by large experiments deep underground such as e.g. macro and superkamiokande . in particular , the case of the neutralino candidate in mssm has been discussed in @xcite , showing that their model dependent results were not in conflict with dama / nai .
as we mentioned , the annihilation of the dark matter particles in the galactic halo could also produce antimatter particles and gamma rays .
the antimatter searches have to be carried out outside the atmosphere , i.e. on balloons or satellites .
in particular , the dark matter particles annihilation would result in an excess of antiprotons or of positrons over an estimated background arising from other possible sources .
the estimate and subtraction of such a background together with the influence of the earth and of the galactic magnetic field on these particles plays a crucial role on the possibility of a reliable extraction of a signal .
however , at present an excess of positrons with energy @xmath149 gev has been suggested by @xcite and other experiments . interpreted in terms of dark matter particles annihilation
it gives a result not in conflict with the effect observed by dama / nai @xcite . as regards the possibility to detect @xmath15 s from dark matter particles annihilation in the galactic halo ,
experiments in space are planned . however , at present it is difficult to estimate their possibilities considering e.g. the background level , the uncertainties in its reliable estimate and subtraction as well as the smallness of the expected signal ( even more , if a subdominant component would be present ) when properly calculated with rescaling procedure . however , in ref .
@xcite the presence of a @xmath15 excess from the center of the galaxy in the egret data @xcite has already been suggested .
this excess match with a possible dark matter particles annihilation in the galactic halo @xcite and is not in conflict with the dama / nai model independent result previously published . for completeness
, we remind that recently it has been suggested @xcite that the positive hints from the indirect detection namely the excess of positrons and of gamma rays in the space and the effect observed by dama / nai can also be described in a scenario with multi - component dark matter in the galactic halo , made of a subdominant component of heavy neutrinos of the 4@xmath150 family and of a sterile dominant component .
= 2.3 in = 2.3 in in particular ( see fig . [ figtheo2 ] ) , it has been shown that an heavy neutrino with mass around 50 gev can account for all the observations , while the inclusion of possible clumpiness of neutrino density as well as new interactions in the heavy neutrino annihilation , etc .
can lead to wider mass ranges : from about 46 up to about 75 gev ( see ref .
@xcite for details ) .
the large merits of highly radiopure nai(tl ) set - up have been demonstrated in the practice by dama / nai which has been the highest radiopure set - up available in the field , has effectively pursued a model independent approach to investigate dark matter particles in the galactic halo collecting an exposure several orders of magnitude larger than those available in the field and has obtained many other complementary or by - products results . in 1996
dama proposed to realize a ton set - up @xcite and a new r&d project for highly radiopure nai(tl ) detectors was funded at that time and carried out for several years in order to realize as an intermediate step the second generation experiment , successor of dama / nai , with an exposed mass of about 250 kg .
thus , new powders and other materials have been selected , new chemical / physical radiopurification procedures in nai and tli powders have been exploited , new growing / handling protocols have been developed and new prototypes have been built and tested . as a consequence of the results of this second generation r&d , the new experimental set - up dama / libra ( large sodium iodide bulk for rare processes ) , @xmath0250 kg highly radiopure nai(tl ) crystal scintillators ( matrix of twenty - five @xmath0 9.70 kg nai(tl ) crystals ) , was funded and realised . after the completion of the dama / nai data taking in july 2002 , the dismounting of dama / nai occurred and the installation of dama / libra started .
in particular , the experimental site as well as many components of the installation itself have been implemented ( environment , shield of pmts , wiring , hp nitrogen system , cooling water of air conditioner , electronics and daq , etc . ) . in particular , all the cu parts have been chemically etched before their installation following a new devoted protocol and maintained in hp nitrogen atmosphere until the installation .
all the procedures performed during the dismounting of dama / nai and the installation of dama / libra detectors have been carried out in hp nitrogen atmosphere ( see fig.[fg : fig12 ] ) .
dama / libra is taking data since march 2003 and the first data release will , most probably , occur when an exposure larger than that of dama / nai will have been collected and analysed in all the aspects . just as an example of the quality of the data taking , fig .
[ fg : sta ] shows the stability of the calibration factor and of the ratio of the peaks positions of the @xmath35am source during about one year of data taking . )
of the measured energy distribution of the @xmath35am source during about one year of data taking.,scaledwidth=50.0% ] the highly radiopure dama / libra set - up is a powerful tool for further investigation on the dark matter particle component in the galactic halo having all the intrinsic merits already mentioned in section 2 and a larger exposed mass , an higher overall radiopurity and improved performances with the respect to dama / nai .
thus , dama / libra will further investigate the 6.3 @xmath2 c.l .
model independent evidence pointed out by dama / nai with increased sensitivity in order to reach even higher c.l .. moreover , it will also offer an increased sensitivity to improve corollary quests on the nature of the candidate particle , trying to disentangle at least among some of the many different possible astrophysical , nuclear and particle physics models as well as to investigate other new possible scenarios . in the following some of the main topics
not yet well known at present and which can affect whatever model dependent result and comparison will be mentioned .
they will be addressed by the highly radiopure dama / libra in a way often unique and always reliable , cheap and competitive .
they are : * _ the velocity and spatial distribution of the dark matter particles in the galactic halo_. it has been shown that the naive description of the galactic halo as an isothermal halo is an unphysical and non - realistic approximation which significantly affects model dependent evaluations ( exclusion plots , allowed regions , etc . ) and comparisons .
other modelings ( not exhaustive at all ) , many of them based on n - bodies simulations , have been considered in literature and some of them have been discussed at some extent in @xcite and references therein . some of these models could be significantly discriminated by dama / libra . * _ the effects induced on the dark matter particles distribution in the galactic halo by contributions from satellite galaxies tidal streams_. recently it has been pointed out @xcite that contributions to the dark matter particles in the galactic halo should be expected from tidal streams from the sagittarius dwarf elliptical galaxy . considering that this galaxy was undiscovered until 1994 and considering galaxy formation theories , one has to expect that also other satellite galaxies do exist and contribute as well . in particular , the canis major satellite galaxy has been pointed out as reported in 2003 in ref .
@xcite ; it can , in principle , play a very significant role being close to our galactic plane . at present ,
the best way to investigate the presence of a stream contribution is to determine in accurate way the phase of the annual modulation , @xmath49 , as a function of the energy ; in fact , for a given halo model @xmath49 would be expected to be ( slightly ) different from 152.5 d and to vary with energy ( see fig .
[ fg : stream ] ) .
+ , of the annual modulation signal as function of the energy when considering : i ) only galactic halo ( `` no sgr '' ) ; ii ) galactic halo ( @xmath138 halo model with @xmath151 km / s , @xmath79 equal to the maximum value for this model ) and a contribution from sagittarius dwarf galaxy ( `` c2 '' ) ; iii ) galactic halo ( @xmath152 halo model with @xmath151 km / s , @xmath79 equal to the maximum value for this model ) and a contribution from sagittarius dwarf galaxy ( `` a5 '' ) .
the contributions from sagittarius dwarf galaxy have been taken in both cases with a density equal to 4% of @xmath79 .
the light shadow region is the final result of dama / nai on the @xmath49 value for the cumulative energy interval ( 2 6 ) kev , while the dark shadow region is the expectation on @xmath49 assuming an experiment with the same features as dama / nai , an exposure of @xmath153 kg @xmath1 day and the same central value for @xmath49 .
] * _ the effects induced on the dark matter particles distribution in the galactic halo by the existence of caustics_. it has been shown that the continuous infall of dark matter particles in the galactic gravitational field can form caustic surfaces and discrete streams in the dark matter particles halo @xcite .
the phenomenology to point out a similar scenario is analogous to that in the previous item ; thus , dama / libra can as well test this possibility . *
_ the detection of possible `` solar wakes''_. as an additional verification of the possible presence of contributions from streams of dark matter particles in our galactic halo , dama / libra can investigate also the gravitational focusing effect of the sun on the dark matter particle of a stream .
in fact , one should expect two kinds of enhancements in the dark matter particles flow : one named `` spike '' , which gives an enhancement of dark matter particle density along a line collinear with the direction of the incoming stream and of the sun , and another , named `` skirt '' , which gives a larger dark matter particle density on a surface of cone whose opening angle depends on the stream velocity .
thus , dama / libra will investigate such a possibility with high sensitivity through second - order time - energy correlations .
* _ the study of possible structures as clumpiness with small scale size_. possible structures as clumpiness with small scale size could , in principle , be pointed out by exploiting a large exposure which can be collected by dama / libra . * _ the coupling(s ) of the dark matter particle with the @xmath38na and @xmath39i and its nature_. as mentioned , several large uncertainties are linked to the coupling(s ) between the dark matter particle and the target - nuclei .
dama / libra , exploiting a new large exposure , will allow to better constrain the related aspects .
in addition , analyses in model frameworks suitable for e.g. mirror dark matter @xcite and for particles from multi - dimensional kaluza - klein like theories will be performed as well and compared with the other scenarios analised so far . * _ scaling laws and cross sections_. at present just simple scaling laws are used to scale all the nuclear cross sections to a common nucleon cross section ; however , they are a large source of uncertainties in model dependent results and comparisons . for example ,
recently , it has been pointed out @xcite that , even for the neutralino candidate , these assumptions ( which hold in the case of model with one - nucleon current ) are arbitrary when two - nucleon current with pion exchange are introduced .
thus , the presence of two target - nuclei in the nai(tl ) detectors of dama / libra could in principle offer a probe for that in the large exposure this experiment will collect .
a large work will be faced by dama / libra , which is in addition the intrinsecally most sensitive experiment in the field of dark matter because of its radiopurity , exposed mass and high duty cycle .
these qualities will also allow dama / libra to further investigate with higher sensitivity several other rare processes .
finally , at present a third generation r&d effort toward the possible nai(tl ) ton set - up has been funded and related works have already been started .
dama / nai has been a pioneer experiment running at lngs for several years and investigating as first the model independent wimp annual modulation signature with suitable sensitivity and control of the running parameters . during seven independent experiments of one year each one , it has pointed out at 6.3 @xmath2 c.l .
the presence of a modulation satisfying all the many peculiarities of a wimp induced effect .
neither systematic effects nor side reactions able to account for the observed modulation amplitude and to contemporaneously satify all the requirements of the signature have been found .
dama / nai has also pointed out the complexity of corollary investigations on the nature of the candidate particle , because of the present poor knowledge on the many astrophysical , nuclear and particle physics aspects .
a second generation experiment dama / libra has been realized and put in operation since march 2003 .
this new set - up , having a larger exposed mass and an higher overall radiopurity , will offer a significantly increased sensitivity to further investigate the dark matter particle component in the galactic halo pointed out at 6.3 @xmath2 c.l . by dama / nai .
moreover , further efforts towards the creation of ultimate radiopure nai(tl ) set - ups are already in progress ; in particular , a third generation r&d is in progress towards the possible 1 ton set - up we proposed in 1996 .
99 -2pt f. zwicky , _ helv .
_ 6 ( 1933 ) 110 s. smith , _ astrophys . j. _ 83 ( 1936 ) 23 v.c . rubin and w.k .
ford , _ astrophys .
j. _ 159 ( 1970 ) 379 ; m. roberts and a.h .
rots , _ astron
. astrophys .
_ 26 ( 1973 ) 483 d. n , spergel et al . ,
_ astro - ph/0302209 _ a. riess et al .
, _ astronom .
j. _ 116 ( 1998 ) 1009 ; e.d .
perlmutter et al . , _ astrophys .
j. _ 517 ( 1999 ) 565 c. l. bennett et al .
, _ astro - ph/0302207 _ r.a.c .
croft et al . , _ phys .
_ 83 ( 1999 ) 1092 ; o. elgaroy et al .
_ astro - ph/0204152 _ r. bernabei et al . , p. belli , r. bernabei , c. bacci , a. incicchitti , r. marcovaldi , d. prosperi , dama proposal to infn scientific committee ii , april 24@xmath150 1990 .
r. bernabei et al . , r. bernabei et al . , p. belli et al .
, r. bernabei et al . ,
r. bernabei et al .
c18 ( 2000 ) 283 r. bernabei et al . , r. bernabei el al . , _ eur
j. _ c23 ( 2002 ) 61 p. belli et al . ,
r. bernabei el al . , _
la rivista del nuovo cimento _ 26 n.1 ( 2003 ) 1 - 73 [ astro - ph/0307403 ] k.a .
drukier et al . , k. freese et al . , p.j.t .
leonard and s. tremaine , _ astrophys . j. _ 353 ( 1990 ) 486 ; c.s .
kochanek , _ astrophys .
j. _ 457 ( 1996 ) 228 ; k.m .
cudworth , _ astron .
_ 99 ( 1990 ) 590 d. smith and n. weiner , and hep - ph/0402065 .
k.freese et .
astro - ph/0309279 ; _ phys .
_ 92 ( 2004 ) 11301 .
r. bernabei et al . , _ il nuovo cim .
_ a112 ( 1999 ) 545 p. belli et al . , ;
p. belli et al . , ; p. belli et al . , ( err . ) ; p. belli et al .
, ; p. belli et al . , ; r. bernabei et al . , ; r. bernabei et al . , ;
r. bernabei et al . , ; r. bernabei et al .
phys . j. direct _ c11 ( 2001 ) 1 ; r. bernabei et al . , ; r. bernabei et al . , .
r. bernabei et al .
, r. bernabei et al . , ; r. bernabei et al .
, ; p. belli et al . , p. belli et al . , ; r. bernabei et al . , ; p. belli et al ; r. cerulli et al .
m. ambrosio et al . , p. belli et al . , m. cribier et al . , c. arpesella et al . , _ health phys .
_ 72 ( 1997 ) 629 m. wojcik : _ nucl .
& methods _
b61 ( 1991 ) 8 r. bernabei et al . ,
r. bernabei et al .
, p. belli et al . , _ phys .
. _ b460 ( 1999 ) 236 r. bernabei et al . , .
p. belli et al . , _ phys .
_ c60 ( 1999 ) 065501 .
r. bernabei et al . , .
f. cappella et al .
j._-direct c14 ( 2002 ) 1 .
r. bernabei et al .
_ j. a 23 ( 2005 ) 7 .
r. bernabei et al . , rom2f/2004/30 , submitted for publication . w.h
. press and g. b. rybicki , _ astrophys .
j. _ 338 ( 1989 ) 277 ; j.d .
scargle , _ astrophys .
j. _ 263 ( 1982 )
835 g. prezeau et al .
_ 91 ( 2003 ) 231301 .
a. bottino et al . , _ phys .
lett . _ b402 ( 1997 ) 113 w. dehnen and j. binney , _ mon . not .
r. astron .
_ 294 ( 1998 ) 429 e. i. gates , g. gyuk , m. s. turner , _ phys .
_ d53 ( 1996 ) 4138 r.h .
helm , ; a. bottino et al .
ressell et al .
groom et al .
c 15 ( 2000 ) 1 a. bottino et al . , ; a. bottino et al , _ hep - ph/0304080 _ d. hooper and t. plehn , madph-02 - 1308 , cern - th/2002 - 29 , _ hep - ph/0212226 _ g. blanger , f. boudjema , a. pukhov and s. rosier - lees , _ hep - ph/0212227 _ f.s .
ling , p. sikivie , s. wick , astro - ph/0405231 .
a. bottino et al . , .
cdms collaboration , astro - ph/0405033 ; edelweiss collaboration , in the proc . of ndm03 , japan ( 2003 )
; n. smith , talk given at idm02 , york , september 2002 r. luscher , talk given at moriond , march 2003 e. simon et al .
instrum . & meth .
_ a507 ( 2003 ) 643 .
a. bottino et al .
, ; ; ; ; ; ; ; ; _ nucl . phys .
b _ 608 ( 2001 ) 461 g.l .
kane et al . , _ hep - ph/0108138 _ a. morselli , talk given at vulcano 2002 a. morselli et al . , _ astro - ph/0211327 _ a. strong et al . , _ astrophys .
j. _ 537 ( 2000 ) 763 k. belotsky , d. fargion , m. khlopov and r.v .
konoplich , hep - ph/0411093 r. bernabei et al .
, _ astrop .
_ 4 ( 1995 ) 45 ; r. bernabei , in the identification of dark matter , world sc . pub .
( 1997 ) 574 .
ibata et al . , _ mon . not .
_ 348 ( 2004 ) 12 .
r. foot , _ hep - ph/0308254_. | the dama / nai experiment ( @xmath0 100 kg highly radiopure nai(tl ) ) was proposed , designed and realised to effectively investigate in a model independent way the presence of a dark matter particle component in the galactic halo by exploiting the annual modulation signature . with a total exposure of 107731 kg @xmath1 day , collected over seven annual cycles deep underground at the gran sasso national laboratory of the i.n.f.n . , it has pointed out at 6.3 @xmath2 c.l .
an effect which satisfies all the peculiarities of the signature and neither systematic effects nor side reactions able to mimic the signature were found .
moreover , several ( but still few with respect to the possibilities ) corollary model dependent quests for the candidate particle have been carried out . in this paper
the obtained results are summarized and some perspectives are discussed at some extent .
# 1#2#3#4#1 # 2 ( # 3 ) # 4 p = 0.65 cm * dark matter particles in the galactic halo : results and implications from dama / nai + * r.bernabei , p.belli , f.cappella , r.cerulli , f.montecchia , f.nozzoli _ dip .
di fisica , universita di roma `` tor vergata '' and infn , sez .
roma2 , i-00133 rome , italy _
a.incicchitti , d.prosperi _ dip .
di fisica , universita di roma `` la sapienza '' and infn , sez .
roma , i-00185 rome , italy _
c.j.dai , h.h.kuang , j.m.ma , z.p.ye _ ihep , chinese academy , p.o .
box 918/3 , beijing 100039 , china _ _ keywords : _ dark matter ; wimps ; underground physics _ pacs numbers : _ 95.35.+d |
the large production cross section of bottom quark pairs ( @xmath7 ) at polarized hadron colliders @xcite coupled with the relatively long lifetime of the @xmath8-quark , implies that detailed experimental study of the properties of the @xmath8-quark at those facilities could also be very useful . since the ckm mixing angle
@xmath9 is close to one , the @xmath8 quark couples rather readily to the virtual top quark in loop diagrams .
now the top quark is so heavy ( @xmath10 gev ) @xcite that the @xmath8-quark becomes very sensitive to electroweak radiative corrections as these corrections often tend to grow with the virtual quark mass @xcite .
precision studies of the @xmath8 s are thus very useful in testing the standard model ( sm ) and in searching for new physics . in the hadronic environment , while the production cross sections are high , a quantitative understanding of the effects of qcd can be a very difficult challenge .
for this reason , in testing the sm and in searching for clues of new physics , it is perhaps better to focus on observables that tend to be robust to qcd corrections .
for that reason , in general , the production cross section is not a good observable .
we propose to focus instead on signatures of parity violation in @xmath8-quark production since qcd corrections can not generate parity violation .
parity violating asymmetries that are ratios of cross - sections should be less sensitive to qcd corrections as well as to the uncertainties in the parton distribution functions .
furthermore , in study of parity violation , one may be able to make use of polarized incident @xmath11 beams @xcite .
we will concentrate on one type of parity violating observable , that is the forward - backward asymmetry . in its differential form
, it is defined as : @xmath12 where the subscripts @xmath13 and @xmath14 stand for the forward and the backward hemispheres respectively . for the reactions of interest to us , i.e. @xmath15 or @xmath16 , we consider two sources of contributions to such a parity violating observable .
these are the chromo - anapole form factor of the @xmath17 vertex ( note @xmath18 ) and the tree level electroweak process @xmath19 . the latter contribution ( @xmath20 exchange )
is significant only when the invariant mass ( @xmath21 ) of the @xmath7 pair is close to the @xmath20 mass ( @xmath22 ) .
it can be removed , if necessary , by imposing an appropriate cut on the @xmath21 .
thus , the chromo - anapole form factor is a very important contributor to the parity violating signal and consequently it is our primary focus .
recently , several suggestions have been made to study parity violation asymmetries in polarized hadron collisions for ( 1 ) the production of one jet , two jets , and two jet plus photon in the sm @xcite ; ( 2 ) the production of @xmath23 and @xmath20 in the sm ; and ( 3 ) the inclusive production of one jet with a new handed interaction between subconstituents of quarks .
these works primarily deal with interference of electroweak and strong interactions on light quarks . in this letter , we present the first study on parity violation generated from one loop weak corrections to bottom quark pair production at polarized hadron colliders .
the key difference with the works in refs .
is , as alluded to in the opening paragraph , that the @xmath7 pair in the final state in our study is a very sensitive tool of the effects of the top quark which in turn is sensitive to non - standard effects .
specifically , we will evaluate the forward - backward asymmetry in the bottom quark pair production at polarized @xmath0 and @xmath1 colliders in the standard model and in a two higgs doublet model to examine the effects of parity violation from the interactions of @xmath7 with spin-1 and spin-0 fields .
let us write the @xmath17 vertex as @xmath24 where @xmath25the strong coupling , @xmath26the su(3 ) matrices , @xmath27 and @xmath28 are the dirac spinors of @xmath8 and @xmath29 with outgoing momenta @xmath30 and @xmath31 and @xmath32 is the momentum of the gluon .
at the tree level @xmath33 . the 1-loop vertex function can be expressed as @xmath34 \nonumber \\ & & + ( p_1-p_2)^\mu [ c(k^2 ) - d(k^2 ) \gamma_5 ] \nonumber \\ & & + ( p_1+p_2)^\mu [ e(k^2 ) - f(k^2 ) \gamma_5 ] \label{eq : ff1}\end{aligned}\ ] ] current conservation demands that @xmath35 and @xmath36 , where @xmath32 and @xmath37 in the @xmath7 center of mass ( cm ) frame . applying the gordon identities we can re - write @xmath38 where @xmath39 , @xmath40the chromo - charge ; @xmath41 , @xmath42the anomalous chromo - magnetic moment ; @xmath43 , @xmath44chromo - anapole moment ; @xmath45 , and @xmath46the chromo - electric dipole moment . in the sm ,
the dominant one loop weak corrections that contribute to the chromo - anapole moment arise from diagrams with the @xmath47 and its goldstone counterpart , the @xmath48 . in many extensions of the sm , @xmath49 in models which contain more than one doublets of higgs ,
there are charged higgs bosons ( @xmath50 ) contributing to the chromo - anapole form factor of fermions at the one - loop order . as is well known the simplest of such extensions
consists of two higgs doublets @xcite @xmath51 and @xmath52 with vacuum expectation values ( vevs ) @xmath53 and @xmath54 .
after symmetry breaking , there remain five physical higgs bosons @xcite : a pair of singly charged higgs bosons @xmath55 , two neutral cp - even scalars @xmath56 ( heavier ) and @xmath57 ( lighter ) , and a neutral cp - odd pseudoscalar @xmath58 .
the ratio of the two vevs is usually expressed as @xmath59 . in our analysis , we have considered such a two higgs doublet model ( thdm ) with the yukawa interactions of model ii @xcite , which is required in the minimal supersymmetric model ( mssm ) .
@xcite in this model , one doublet ( @xmath51 ) couples to down - type quarks and charged leptons while the other ( @xmath52 ) couples to up - type quarks and neutrinos .
the one loop weak corrections from diagrams with the @xmath47 , the @xmath48 and the @xmath60 , yield dominant contribution to the b quark anapole moment from the virtual top quark in the loop . in our calculations of these corrections
we will set @xmath61 and also we will ignore cp violating effects . we have employed the t hooft feynman gauge for loop calculations with @xmath62 .
the forward - backward asymmetry in @xmath7 production is primarily driven by @xmath63 which is related to the chromo - anapole form factor @xmath64 via @xmath65 .
therefore , in fig . 1 we present @xmath63 coming separately from diagrams with the @xmath66 , the @xmath67 and the @xmath20 as well as the total contribution in the sm .
as an illustration of a beyond the sm effect , we also show @xmath63 for thdm for @xmath68 gev and various values of @xmath69 .
the @xmath20 boson exchange contribution to the @xmath8-chromo - anapole moment is very small since the coupling product @xmath70 is very small .
the form factor @xmath63 is enhanced near @xmath71 , where @xmath72 , due to the @xmath73 threshold which appears in diagrams with charged boson ( @xmath74 @xmath75 @xmath23 or @xmath50 ) exchanges . since @xmath76
, its value is expected to grow with @xmath21 . from fig . 1 we see that numerically this increase sets in for @xmath77 gev . in the thdm , the form factor @xmath78 gets an additional contribution , @xmath79 , from the charged higgs boson .
clearly @xmath80 is a function of @xmath81 and @xmath82 , being proportional to @xmath83 ^ 2\}$ ] .
therefore , it has the same sign as @xmath84 for @xmath85 , but has an opposite sign for @xmath86 .
thus the effects of parity violation are enhanced in the thdm if @xmath82 is less than @xmath87 . for @xmath88 ,
the @xmath80 vanishes and the total @xmath78 becomes that of the sm .
let us define the cross section for the sub - process , @xmath89 , in each helicity state of quarks in the initial state as @xmath90 where @xmath91 represents a right - handed ( @xmath92 ) or a left - handed ( @xmath93 ) helicity of the quark and the antiquark .
the cross sections in the forward @xmath94 and the backward @xmath95 directions are defined as @xmath96 where @xmath97 , with @xmath98 being the scattering angle of the @xmath8 in the @xmath7 center of mass frame , @xmath99 is the total cross section of the @xmath100 subprocess and @xmath101 is the difference of the cross sections in the forward and the backward directions of the subprocess .
( note that variables in the cm frame of the @xmath100 subprocess are denoted with a @xmath102 on top of them ) .
the tree level expressions for these cross sections are @xmath103 at the tree level , the difference of the cross sections in the forward and backward directions is zero in each helicity state of @xmath100 because parity is conserved in qcd .
the one loop weak corrections to the @xmath100 cross section is @xmath104\end{aligned}\ ] ] and the difference of @xmath100 cross sections in the forward and backward directions is @xmath105\end{aligned}\ ] ] where @xmath106 and @xmath58 , @xmath14 and @xmath107 are form factors defined in eq .
[ eq : ff1 ] .
a nonzero @xmath108 is of course a signature of parity violation . in polarized @xmath0 and @xmath1 collisions ,
the differential cross section for the production of @xmath7 takes the form @xmath109 \nonumber \\ & + & \hat{\sigma}_{lr } [ q_l(x_1,\hat{s})\bar{q}_r(x_2,\hat{s } ) + q_l(x_2,\hat{s})\bar{q}_r(x_1,\hat{s } ] \nonumber \\ & + & \hat{\sigma}_{rr } [ q_r(x_1,\hat{s})\bar{q}_r(x_2,\hat{s } ) + q_r(x_2,\hat{s})\bar{q}_r(x_1,\hat{s } ] \nonumber \\ & + & \hat{\sigma}_{ll } [ q_l(x_1,\hat{s})\bar{q}_l(x_2,\hat{s } ) + q_l(x_2,\hat{s})\bar{q}_l(x_1,\hat{s } ] \end{aligned}\ ] ] where @xmath110 and @xmath111 are momentum fractions of the partons in the initial beam .
parity violation can be studied at hadron colliders by using polarized initial beams , e.g. @xmath112 or @xmath113 etc .
let us consider two hadron colliders with polarized beams ( a ) @xmath112 and ( b ) @xmath113 . with polarized @xmath0 or @xmath1 beams
, we can define @xmath114 [ @xmath115 to be the probability density @xcite for a quark of flavor @xmath116 and momentum fraction @xmath117 with a helicity of the same [ opposite ] sign as the helicity of the proton .
the differential cross section , @xmath118 in polarized @xmath112 collisions become @xmath119 \nonumber \\ \sigma_{rl } & = & \int^{1}_{x_1^{min } } dx_1 \int^{1}_{x_2^{min } } dx_2 \frac{d\sigma_{rl}}{dx_1 dx_2 } \label{eq : sigma}\end{aligned}\ ] ] for @xmath120 , we need to sum over all helicity states : @xmath121 , @xmath122 @xmath123 and @xmath124 .
the difference of @xmath125 in the forward ( @xmath126 ) and the backward ( @xmath127 ) directions at the @xmath7 cm frame in polarized @xmath128 collisions are @xmath129 \nonumber \\ \delta\sigma_{rl } & = & \int^{1}_{x_1^{min } } dx_1 \int^{1}_{x_2^{min } } dx_2 \frac{\delta\sigma_{rl}}{dx_1 dx_2 } \label{eq : dsigma}\end{aligned}\ ] ] at high energy , parity violation from chromo - anapole form factor is expected to increase with @xmath21 .
it is therefore useful to examine the differential forward - backward asymmetry given in eq .
[ eq : deltaa ] .
the numerator has contribution only from @xmath89 , while the denominator has contributions from both @xmath89 and @xmath120 .
gluon fusion produces a large number of @xmath7 pairs , which tend to contribute significantly to the denominator of eq .
[ eq : deltaa ] thus reducing the signal for the parity violating asymmetry from @xmath100 in @xmath0 collisions . in @xmath1 collisions , the antiquark density is greatly enhanced , therefore , parity violation signals from @xmath89 is much larger in @xmath1 than in @xmath0 collisions at the same energy .
this differential asymmetry is presented in figures 2 and 3 for polarized @xmath112 and @xmath113 collisions at several values of energy .
the parity violation signal peaks at the @xmath130 in the differential of forward - backward asymmetry versus @xmath21 .
also shown is the same asymmetry in @xmath19 ; it dominates if @xmath21 is close to @xmath22 , but becomes negligible for @xmath131 gev .
the integrated forward - backward asymmetry is defined as @xmath132 where @xmath133 and @xmath134 are the number of @xmath7 pairs in the forward and the backward directions ; and @xmath135 is the integrated luminosity .
the statistical uncertainty ( @xmath136 ) and the statistical significance ( @xmath137 ) of the integrated asymmetry are @xmath138 the difference ( @xmath139 ) and the total ( @xmath125 ) of the cross sections @xmath140 and @xmath141 with one loop weak corrections , as well as the integrated asymmetry ( @xmath142 ) and its statistical significance ( @xmath137 ) are presented in table 1 for an integrated luminosity of 10 fb@xmath143 and 100 fb@xmath143 . for @xmath144 tev , this asymmetry might be visible with @xmath145 100 fb@xmath143 and a cut on @xmath21 .
we find that requiring a higher @xmath21 can efficiently enhance the asymmetry from weak corrections in @xmath89 .
a high @xmath21 cut not only reduces the cross section from gluon fusion it also reduces the asymmetry from @xmath19 as well .
we close with two brief remarks .
\1 ) in our calculations we have retained only the parity violating effects originating from @xmath146 . in principle , gluon fusion , i.e. @xmath147 , can also contribute to parity violation due to electroweak corrections @xcite .
however , dimensional reasoning indicates that such contributions are probably small at lower energy , at least for @xmath148 . in any case
, it is very unlikely that the contribution to the forward backward asymmetry from such a source will cancel away the @xmath100 contribution for all values of @xmath21 .
\2 ) it would clearly be extremely interesting to consider these parity violating effects in other extensions of the sm ; in particular in the mssm .
as we noted before , in mssm , there will be additional loop contributions not contained in a thdm that could enhance the asymmetry @xcite .
to summarize , at a hadron collider with polarized @xmath0 and @xmath149 gev ( e.g. rhic ) , parity violation in @xmath7 production will be dominated by @xmath19 . in polarized @xmath113 collisions with @xmath150 gev and enough integrated luminosity , it might be possible to observe parity violation signals from sm weak corrections in @xmath7 production . in the thdm with model ii yukawa interactions , the chromo - anapole form factor of the bottom quark generated from leading weak corrections is enhanced for @xmath69 close to one .
it is enhanced by the factor @xmath151 for @xmath69 less than one .
therefore , the signal of parity violation in @xmath7 production is greatly enhanced for @xmath152 and @xmath81 less than about 300 gev .
production @xcite , is greatly enhanced for @xmath153 . in ref .
@xcite , parity violation in @xmath73 production was studied in unpolarized @xmath1 collisions . ]
the study of parity violation in @xmath7 production might provide a good opportunity to study new interactions between the third generation quarks and ( charged ) spin-0 as well as spin-1 bosons .
we are grateful to stephane keller and jeff owens for providing polarized parton distribution functions and for beneficial discussions . c.k . and d.a .
would like to thank bill bardeen and the theoretical physics department of fermilab for hospitality where part of this research was completed .
would like to thank mike tannenbaum for interesting discussions .
this research was supported in part by the the u.s .
department of energy grant nos .
de - fg05 - 87er40319 ( rochester ) , de - fg02 - 95er40896 ( wisconsin ) , de - ac05 - 84er40150 ( cebaf ) , and de - ac-76ch00016 ( bnl ) , and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation .
20 a. p. contougouris , s. papadopoulas and b. kamal , phys .
b * 246 * ( 1990 ) 523 .
m. karliner and r. robinett , phys .
b * 324 * ( 1994 ) 209 .
f. abe _ et al_. , [ cdf ] , phys .
rev . lett .
* 74 * ( 1995 ) 2626
. s. abachi _
et al_. , [ @xmath154 , phys . rev . lett .
* 74 * ( 1995 ) 2632 .
see , e.g. , w.s .
hou , r.s .
willey and a. soni , phys .
* 58 * ( 1987 ) 1608 .
g. bunce _
et al_. , particle world * 3 * , 1 ( 1992 ) ; _ see also , proceedings of the polarized collider workshop _ , penn state university 1990 , edited by j. collins , s. heppelmann , and r. robinett , aip conf .
proceedings no . 223 ( aip , new york , 1991 ) .
cheng , m .-
l . huang and c.f .
wai , phys .
d * 49 * ( 1994 ) 1272 . c. bourrely and j. soffer , nucl .
b * 423 * ( 1994 ) 329 .
p. taxil and j.m .
virey , phys .
b * 364 * ( 1995 ) 181 .
georgi , hadronic j. * 1 * ( 1978 ) 155 .
gunion , h.e .
haber , g.l .
kane and s. dawson , _ the higgs hunter s guide _ , addison - wesley , redwood city , ca ( 1990 ) and references therein .
donoghue and l .- f .
li , phys . rev .
* d19 * ( 1979 ) 945 ; l. hall and m. wise , nucl .
b187 * ( 1981 ) 397 . s. keller and j.f .
owens phys . rev .
* d49 * ( 1994 ) 1199 .
d. atwood , c. kao and a. soni , work in progress . c. kao , phys .
b * 348 * ( 1995 ) 155 .
table 1 . the difference ( @xmath155 ) and the total ( @xmath125 ) of cross sections @xmath140 and @xmath141 , and the asymmetry of @xmath156 as defined in eq .
[ eq : afb ] generated by weak corrections in ( a ) the sm and ( b ) the thdm with @xmath157 gev and @xmath158 , for @xmath149 gev with @xmath159 100 gev , @xmath160 tev with @xmath159 200 gev and @xmath161 tev with @xmath159 300 gev . also shown is the statistical significance ( @xmath137 ) for @xmath162 10 fb@xmath143 and 100 fb@xmath143 . [ cols="^,^,^,^,^,^,^ " , ]
1 . the real part of the chromo - anapole form factor @xmath163 as a function of @xmath21 from ( a ) the diagrams with @xmath48 ( dash ) , the @xmath47 ( dot - dash ) , the @xmath20 ( dot ) and the sm total ; and the thdm total for @xmath164 0.5 , 1 , 3 , 10 and 35 , with ( b ) @xmath165 gev and ( c ) @xmath166 gev . note that @xmath167 .
the differential forward - backward asymmetry ( @xmath168 ) defined in eq .
[ eq : deltaa ] versus @xmath21 , in polarized @xmath112 collisions for ( a ) @xmath169 gev ( b ) @xmath170 gev and ( c ) @xmath171 gev .
this asymmetry has been evaluated using the ( total ) chromo - anapole form factor from leading weak corrections in the sm as well as in the thdm for @xmath165 gev and @xmath164 0.5 , 1 , and 35 .
the differential forward - backward asymmetry ( @xmath168 ) defined in eq .
[ eq : deltaa ] versus @xmath21 , in polarized @xmath113 collisions for ( a ) @xmath169 gev ( b ) @xmath170 gev and ( c ) @xmath171 gev .
this asymmetry has been evaluated using the ( total ) chromo - anapole form factor from leading weak corrections in the sm as well as in the thdm for @xmath165 gev and @xmath164 0.5 , 1 , and 35 . | parity violation induced by the chromo - anapole form factor of the bottom quark , generated from weak corrections , is studied in polarized hadron collisions . the forward - backward asymmetry in the bottom quark pair production at polarized @xmath0 and @xmath1 colliders
is evaluated in the standard model and in a two higgs doublet model to examine the effects of parity violation . in the models
studied , promising results are found for polarized @xmath1 colliders . * ur-1384 * + * madph-96 - 968 * * parity violation in bottom quark pair production + at polarized hadron colliders * chung kao@xmath2 , david atwood@xmath3 and amarjit soni@xmath4 @xmath5_department of physics and astronomy , university of rochester , + rochester , ny 14627 , usa _
+ @xmath6_department of physics , university of wisconsin , madison , wi 53706 , usa _
+ @xmath3_theory group , cebaf , newport news , va 23606 , usa _ + @xmath4_department of physics , brookhaven national laboratory , upton , ny 11973 , usa _ |
the eruption of prominences and magnetic flux ropes from the sun has important implications in view of space weather manifestations .
these eruptions are frequently associated with solar flares and coronal mass ejections .
it is widely accepted that magnetic reconnection is the most fundamental process responsible for the changes in the topology of magnetic fields , as well as the rapid conversion of stored magnetic energy into thermal and kinetic energy of plasma and particles during solar eruptive events ( for a review , see * ? ? ?
solar flares occur in active regions consisting of sunspot groups in the photosphere enveloped by large overlying coronal loops that essentially represent a closed magnetic field configuration associated with sunspots .
the spectrum of energy released from a flare shows that there is a precursor phase , which is generally marked by a gradual increase in thermal radiation mainly in the form of small enhancement of soft x - ray ( sxr ) flux @xcite .
it is followed by the impulsive phase lasting for a few seconds to tens of minutes , comprising of energy release in the form of @xmath3-ray , hard x - ray ( hxr ) , extreme ultraviolet ( euv ) , and microwave ( mw ) emissions .
after the impulsive phase thermal radiation dominates . in this phase , h@xmath4 and sxr emission
attain a maximum level followed by a gradual decline in the intensity . to understand
the diversity of phenomena during eruptive flares , the standard flare model ( also known as cshkp model ) has been proposed which incorporates the pioneering works of @xcite , @xcite , @xcite , and @xcite . in the standard flare model ,
the initial driver of the flare process is a rising prominence ( also called filament when viewed on the solar disk ) formed in the chromosphere along the magnetic polarity inversion line of solar active regions . in a simplistic picture
, one can imagine the structure of a bipolar magnetic configuration in terms of an inner region , called the core field , and the outer region , called the envelope field .
the core fields are rooted close to the neutral line while the envelope fields are rooted away from it .
many observations have confirmed that core fields can usually be traced by a filament .
the standard flare model recognizes that in the early stages of a large eruptive flare , the core fields containing the prominence erupt which would result in the stretching of the envelope fields . with the evolution of the eruption process
, the stretched field lines reclose via magnetic reconnection beneath the erupting filament .
the extensive solar observations over the last decades , mainly from space borne telescopes , have revealed several important features of the multi - wavelength flare that have further strengthened the standard flare model and provided concrete observational inputs to understand the role of magnetic reconnection toward plasma heating and particle acceleration .
these features include : occurrence of hard x - ray ( hxr ) footpoint ( fp ) and looptop ( lt ) sources , formation of h@xmath4 flare ribbons at the opposite polarity regions , rising arcade of the intense soft x - ray ( sxr ) loops , plasmoid ejections , upward motion of hxr lt source , and increasing separation of h@xmath4 flare ribbons along with hxr fp sources ( see reviews by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . according to the standard flare model , the filament activation is a crucial part of the flare process .
observations reveal several categories of the filament activation , viz . full , partial and failed @xcite .
the full or partial eruptions are followed by flares and cmes while the failed eruptions characterize those situations when a filament undergoes a successful activation in the beginning but finally fails to erupt from the corona .
full and partial eruption of filaments have been reported in numerous observations . on the other hand ,
failed eruptions are observed less frequently and therefore form a less studied topic in solar physics .
their study provide an opportunity to investigate the interaction between the erupted plasma and magnetic fields with the overlying coronal loops ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . the rate of decrease of overlying magnetic fields with their heights is considered to be a crucial factor that determines whether an erupting prominence will lead to a successful eruption or confine within the overlying coronal region @xcite . @xcite and @xcite studied a failed eruption of a filament during which multiple hxr sources were observed from the coronal as well as footpoint locations .
the examination of the locations of coronal hxr sources with respect to the dynamical evolution of the filament is of high interest toward a better understanding of the energy release scenario in solar eruptions . sometimes prior to the filament activation and associated flares , small localized brightenings
are observed to occur in the active region close to the site of the main flare @xcite .
such localized pre - flare brightenings are frequently observed as enhanced x - ray flux at low energies ( e.g. , goes light curves ) before the impulsive phase .
it has been recognized that filaments undergo slow yet important changes during the pre - flare phase @xcite
. some studies also show crucial evidence of non - thermal x - ray emission during the early activation phase of solar eruptions @xcite .
these investigations have provided clues to understand the processes occurring in the active region that eventually lead to filament destabilization and thus triggering the impulsive energy release .
it is important to note that , while investigating the pre - flare phase , one should also carefully probe the state of overlying coronal loops besides the core region containing the filament . in this study , we present observations of an m6.2 flare and its associated pre - flare activities observed in active region noaa 10646 on 2004 july 14 between 04:30 ut and 06:00 ut that are associated with the confined eruption of a filament ( sol2004 - 07 - 14 ) . during the pre - flare phase ,
we observe contraction of overlying coronal loops as well as multiple brightenings at its core region where the main flare occurred .
the impulsive phase of the m6.2 flare is characterized by multiple non - thermal peaks during which a rising prominence showed an initial acceleration for @xmath14 minutes , then slowed down and finally stopped by overlying loops .
several aspects of this event related to the confinement of the prominence eruption and associated flare emissions have been analyzed in previous studies @xcite .
@xcite reported that this flare can be well described by a quadrupole model .
this study showed that the failed eruption of the flux rope caused stretching in the overlying loop system .
as a consequence of this interaction , the overlying loops were observed to oscillate radially with an initial amplitude of 9520 km , a period of 377 s and an exponential damping time of 500 s. this flare is one of three events analyzed in @xcite in which they found a close relation among the euv brightenings observed at the footpoints of overlying loop systems , a decrease in velocity of the eruption , and episodes of hardening of the hxr spectra . from these observations
, they concluded that the interaction of the erupting structure with overlying loops was strong enough to heat the high loops which then cooled down in @xmath130 - 60 minute to become visible in the euv range .
here we revisit sol2004 - 07 - 14 and provide a comprehensive analysis of the intriguing flaring and eruptive activities from its prolonged pre - flare phase to the decay of the event .
we obtained significant additional information about the state of overlying coronal loops , the flaring core region and energy release processes during the flare s impulsive phase . on the whole ,
this event was remarkable in several aspects .
( 1 ) we note significant pre - flare activity which is manifested in the form of large - scale contraction of overlying coronal loops .
( 2 ) accompanying with the large - scale loop contraction , sequential brightening in the core region was observed which can clearly be recognized as three pre - flare events in the goes x - ray profile .
( 3 ) a drastic increase in the speed of contraction is observed just @xmath55 minute before the onset of the impulsive phase ( i.e. , during third pre - flare event ) .
( 4 ) during the impulsive phase , we observe multiple non - thermal peaks .
the first stage of the impulsive phase marks the transition from signs of the magnetic implosion scenario @xcite to the generally observed expansion of flaring coronal loops and the activation of filament .
the high spatial resolution , continuous sequence of euv images at 171 from the pre - flare phase to the end of the flare provided us with an excellent opportunity to thoroughly examine activities in the core region of the eruption along with the overlying system of coronal loops .
the signatures of magnetic reconnection and particle acceleration are inferred from hxr and microwave ( mw ) observations .
we further provide a detailed hxr spectroscopy of the flare to investigate the crucial plasma parameters and flare energetics .
the paper is organized as follows . in section
[ data ] , we provide details of observing instruments and data analysis procedures . for the convenience of readers , an overview of the multi - wavelength observations
are presented in section [ overview ] .
we provide a detailed multi - wavelength diagnostic of each phase of the eruption in section [ results ] . in section [ discuss ] , we discuss and interpret our observations .
the summary of the paper is given in the final section .
in this paper , we present a comprehensive multi - wavelength analysis of the sequence of intriguing activities that occurred in solar active region noaa 10646 on 2004 july 14 between 04:30 ut and 06:00 ut .
transition region and coronal explorer ( trace ; @xcite ) observed the active region noaa 10646 during this whole period at high time cadence ( better than 1 min ) with its euv channel at 171 .
the trace telescope has a field of view of [email protected]@[email protected] and a spatial resolution of 1@xmath8 ( [email protected] pixel@xmath9 ) .
the trace 171 filter is mainly sensitive to plasmas at a temperature around 1 mk ( fe ix / x ) .
reuven ramaty high energy solar spectroscopic imager ( rhessi ; @xcite ) detected significant activities in noaa 10646 during its continuous observations from 04:30 ut to 05:35 ut .
rhessi observes the full sun with an unprecedented combination of spatial resolution ( as fine as @[email protected] ) and energy resolution ( 15 kev ) in the energy range of 3 kev to 17 mev . to reconstruct x - ray images ,
we have used the computationally expensive pixon algorithm @xcite which is thought to provide the most accurate image photometry @xcite .
the images are reconstructed by selecting front detector segments 38 ( excluding 7 ) with 20 s integration time .
our analysis involves the identification of x - ray and mw source locations during the dynamical evolution of the filament and coronal loops observed by trace .
however , the pointing information of trace is often not good enough to obtain a correct co - alignment between trace images and imaging data at other wavelengths . for a full disk imager ( such as soho / mdi ; @xcite )
, the solar limb provides a reference that helps in accurate determination of the absolute pointing of the telescope .
moreover , thermal bending of the trace guide telescope also leads to an unknown variation in the pointing of at least a few arcseconds @xcite .
the pointing information of both rhessi and soho is believed to be sufficiently accurate .
therefore , we corrected trace pointing by a cross - correlation alignment between a trace white light ( wl ) and a soho / mdi wl image taken at 04:45:20 ut and 04:47:32 ut respectively using the method described in @xcite .
this method has been found very effective for near - limb events ( see , e.g. , * ? ? ?
* ; * ? ? ?
we found that the trace pointing was offset by [email protected] in the x direction and [email protected] in the y direction ( see figure [ pointing ] ) . in this analysis ,
we include microwave ( mw ) observations by the nobeyama radioheliograph ( norh ; @xcite ) at 17 ghz and 34 ghz . at these two frequencies norh
has spatial resolutions of 10@xmath8 and 5@xmath8 respectively @xcite .
norh has a sensitivity of at least 1 solar flux unit ( sfu ) at 17 ghz and @xmath13 sfu at 34 ghz
. the normal time resolution of norh is 1 s , but a resolution as good as 50 ms can be used for special projects ( see also * ? ? ?
for the present study we have analyzed norh light curves and images with a temporal cadence of 1 s.
on 2004 july 14 , active region noaa 10646 was situated close to the western limb at heliographic coordinates of n13w64 . in figure [ mdi_hsi ]
, we show a trace white light ( wl ) image overlaid by a soho / mdi line - of - sight ( los ) magnetic field map of noaa 10646 .
we find that the active region consists of a set of sunspot groups with leading and trailing groups having positive and negative magnetic polarities respectively .
the active region displayed a rather dispersed structure of sunspots with well separated magnetic polarities .
the active region noaa 10646 exhibited a set of very intriguing flaring and eruptive activities between 04:30 ut and 06:00 ut which can be nicely followed in goes sxr time profiles shown in figure [ goes_profile ] . a major flare of class m6.2 occurred in the active region between 05:17 and 06:00 ut .
it is remarkable to observe significant activities in the active region that started several minute before the onset of the m - class flare that continued until the flare s impulsive phase .
the goes light curves ( figure [ goes_profile ] ) clearly recognizes these pre - flare activities in the form of multiple small peaks at @xmath104:41 ut , @xmath105:06 ut and @xmath105:16 ut before the onset of the impulsive phase of flare at @xmath105:17 ut .
thereafter , the flare attains its peak at @xmath105:23 ut and gradually ended by 06:00 ut .
we mark pre - flare goes peaks as i , ii , and iii in figure [ goes_profile ] .
these activities correspond to the localized and discrete brightenings at low - lying loops at the site of the main flare and simultaneous contraction of large overlying coronal loops .
the detailed imaging analysis of x - ray and euv data presented in the next section clearly show that the pre - flare activity corresponds to phases of episodic energy release at the earliest stages of restructuring of magnetic fields .
it is further noteworthy that multiple peaks at pre - flare phases are more pronounced in the high energy goes channel ( i.e. , in 0.54.0 band ) .
based on goes and trace observations , we consider following phases of flare evolution : ( 1 ) pre - flare activity which includes large - scale loop contraction and brightenings in the core region ( 2 ) m6.2 flare which is accompanied with a failed eruption of a filament . here
we note that although the loop contraction proceeds during the whole pre - flare activity , we describe loop contraction and core region brightenings separately in the next section to provide a better clarity on these two phenomena .
we should note that the contraction was observed in higher overlying coronal loops that are spatially separated from the regions of pre - flare brightenings .
it is , however , quite possible that the two phenomena ( localized brightenings and contraction of overlying loop system ) are physically interconnected . in figure [ mdi_hsi ]
, we further show the locations of a pair of hxr and mw emission centroids observed during the early impulsive phase of m6.2 flare by star ( @xmath11 ) and filled circle ( @xmath12 ) symbols respectively .
it is apparent that the flare occurred close to the neutral line of photospheric magnetic field .
we mark locations of hxr and mw emission centroids as s1 and s2 .
following the main impulsive phase , we note another prominent mw source appeared at a distant location , marked as s3 in figure [ mdi_hsi ] , which does not have any hxr counterpart .
active region noaa 10646 was enveloped by a set of large coronal loops with different orientations which present a peculiar structure of active region corona ( see figure [ loops ] ) .
we have estimated the height of three prominent overlying loop systems ( which are indicated in figure [ loops ] by a , b , and c ) from the source region ( shown by ` @xmath7 ' ) .
the projected height of these three systems of loops were estimated as @xmath192 mm , 88 mm , and 93 mm .
we found significant activities occurring in the coronal as well as source region by examining 171 images available from 04:30 ut onward . from the sequence of euv images ,
it is evident that the higher coronal loops began to collapse from @xmath104:47 ut ( see figures [ pre-2 ] & [ imp_ht ] ; and online movie ) .
we find that only one loop system ` b ' underwent contraction .
this large - scale contraction continued until the impulsive rise of the m6.2 flare at 05:18 ut .
we note some important observational characteristics of the loop contraction .
( 1 ) before and at the beginning of loop contraction ( i.e. , 04:4204:51 ut ) the top of the loop system exhibited intense emission which gives evidence for localized heating in the higher corona ( indicated by the arrow in figure [ pre-2](f ) ) . ( 2 ) from 04:51 ut onward , this loop top brightening started getting dispersed and the brightening mostly dissolves by @xmath105:02 ut .
( 3 ) from @xmath14:54 ut , we observe intermittent brightenings in a set of low - lying loops at the source region besides contraction of higher loops ( figure [ pre-2](h ) ) . in figure [ imp_ht ] ( a )
, we present a height - time plot of the collapsing coronal loops . due to mingling of several loops which altogether evolve dynamically ,
the estimation of loop height was a tedious exercise .
we took utmost care in distinguishing one of the loop systems undergoing the contraction by carefully identifying its legs .
we find that the loop contraction proceeds with varying speed . in order to compare the phases of loop contraction with the thermal emission from the flaring region , we have also plotted the co - temporal @xmath13 flux profiles in both channels in figure [ imp_ht](b ) .
we find that in terms of contraction speed , the collapsing loop system undergoes three phases of evolution . in the beginning , the loops contract with relatively higher speed of @xmath511 km s@xmath14 that sustained for @xmath06 minutes .
this is followed by a prolonged phase of @xmath119 minutes during which the contraction slowed down to @xmath15 km s@xmath9 .
as shown in figure [ goes_profile ] , the pre - flare activity is characterized by three episodes of flux enhancement ( marked as i , ii , and iii ) .
we note that the loop contraction continued during all three stages ( figure [ imp_ht ] ) but there is a drastic increase in the speed of contraction from @xmath15 km s@xmath9 to @xmath025 km s@xmath14 at @xmath05:12 ut which mark the third episode of flux enhancement .
the loop contraction halts at 05:18 ut with the beginning of the flare impulsive phase after which the expansion predominates over the implosion .
the sequence of images during the pre - flare phase clearly indicate that the large - scale contraction of euv coronal loops is accompanied by localized brightenings at the core of active region loops . in figure
[ pre-2 ] , we show euv images showing pre - flare activity and mark the three events of goes flux enhancement ( see figure [ goes_profile ] ) as i , ii and iii in panels ( c ) , ( l ) , and ( p ) respectively . here
it is important to note that the core region exhibited pre - flare emission even before the onset of contraction in the form of x - ray ( 612 kev and 1225 kev ) and mw ( 17 ghz and 34 ghz ) emissions which are co - spatial ( figures [ pre-2](b)-(d ) ) .
however , at this time euv images do not show any significant bright structures associated with x - ray and mw emitting regions . the x - ray and mw
sources in the pre - flare phase clearly indicate distinct events of energy release at the core of active region loops that precede the large - scale contraction of overlying loops .
we observe localized , distinct bright euv sources at the core region from 04:54 ut onward , i.e. , after the initiation of the contraction of the loop system .
the location of euv brightening is shown inside a box in figure [ pre-2](h ) .
although brightening at three locations were observed in the beginning ( @xmath104:57 ut ) , we can clearly recognize two of the three bright structures as low - lying loops at some later stages ( @xmath105:02 ut ) .
the intensity of bright loops rapidly decreased ( @xmath105:07 ut ) and we observe formation of a relatively large system of low - lying loops thereafter ( i.e. , 05:08 ut onward ) .
in fact three new bright loops formed sequentially which appeared in coincidence for a brief interval ( around 05:12 ut ) .
these three loops are marked as 1 , 2 , and 3 in order of their occurrences .
we also note that the loop1 became very intense in the later stages ( marked in figure [ pre-2](p ) at 05:15:17 ut ) and also lasted till the impulsive phase of m6.2 flare .
the comparison of pre - flare euv brightening with the co - temporal rhessi images clearly reveal x - ray sources at the top of low - lying loops . it is interesting to see that although in euv images , loop1 appeared the most intense , the hxr emission in 1225 kev energy band is essentially associated with loops 2 and 3 .
we further note that mw emission at 17 ghz and 34 ghz is observed in the form of single source that shows clear association with the bright low - lying euv loops ( figure [ pre-2](p ) ) . in figure [ flux_profile ] , we present the rhessi and norh light curves along with the goes time profiles for the m6.2 flare .
we find that the hxr light curves @xmath1525 kev show a rapid increase in count rates from 05:17 ut onward indicating the onset of the impulsive phase of the flare .
the non - thermal hxr emission continued till @xmath05:25 ut which can be recognized in 2540 kev light curve and further confirmed by hxr spectroscopy ( see section [ sec : rhessi_spec ] ) .
therefore , we have considered the period from 05:17 to 05:25 ut as the impulsive phase of the flare which is indicated in figure [ flux_profile ] by grey shaded area .
the mw light curves clearly show that the impulsive phase is characterized by three distinct events of flux enhancement ( marked as events ` a ' , ` b ' and ` c ' in figure [ flux_profile ] ) with the first and third peaks at 05:18 ut and 05:21 ut being common in hxr profiles also . during mw peak ` b ' at 05:19:54 ut , the hxr profiles above 25 kev also exhibit flux enhancement but a distinct peak structure as in mw light curves can not be recognized .
further these mw peaks are much more distinguishable at 17 ghz than 34 ghz observations .
after the peak ` c ' , mw flux profiles ( mainly 17 ghz ) exhibit significant fluctuations during the impulsive phase . in figure [ mainevent ] , we present combined imaging observations of the m6.2 flare in euv , hxr , and mw to show the crucial phases of the flare evolution . following the impulsive onset of hxr and mw emissions ( event ` a ' )
, we observe the rise of the prominence at the source region and intense euv brightening beneath it ( figures [ mainevent](a)-(c ) ) . in subsequent euv images ,
we observe the evolution of filament in the corona with distinct appearance of its outermost structure ( shown by arrow in figures [ mainevent](f ) , ( g ) , and ( i ) ) which we term as the ` eruption front ' .
the eruption front likely represents a part of the magnetic structure associated with the erupting filament . in figure
[ exp_ht ] , we provide height - time plot of the eruption front .
the eruption front underwent fastest expansion between 05:21 and 05:23 ut with the speed of @xmath0242 km s@xmath14 .
the eruption speed rapidly decreases after 05:23 ut .
we also note that there is a further decrease in the speed of the eruption front as it moves to higher altitudes ; speed of the eruption front was @xmath165 km s@xmath9 between @xmath005:2305:24 ut which decreased to @xmath126 km s@xmath16 between @xmath005:2405:26 ut .
the eruption front was intact and symmetrical till @xmath005:24 ut while attaining a height of 4.8@xmath710@xmath17 km ( from the source region ) at @xmath005:25 ut .
after 05:25 ut , the eruption front got disrupted ( figure [ mainevent](h ) ) while its major portion ( right segment ; marked by arrow in figure [ mainevent](i ) ) descended which we could track / identify till @xmath505:29 ut . in this interval ,
the downward speed of eruption front was estimated as @xmath095 km s@xmath18 .
the remaining portion of the eruption front narrowed down while moving further upward and eventually disrupted . in figure
[ tr_norh ] , we present a series of mw sources at 17 and 34 ghz over co - temporal euv 171 images to study the spatial evolution of mw emission in relation to the phases of prominence activity .
lasco on board soho did not detect any cme associated with this flare ] .
the absence of cme suggests that all the erupted material either fell down on the source region or arrested within the overlying coronal magnetic loops .
euv images from @xmath005:2105:23 ut showed a very important phase of eruption which requires more attention .
we present enlarged view of euv images in figure [ kink ] to emphasize some interesting morphological structures during the early evolution of the erupting system .
we observe a system of twisted loops at the western side ( leading side of eruption ) that undergoes fast expansion ( marked by yellow arrows ) .
further some bright spiky structures can be identified at the eastern side ( marked by red arrows ) which conceivably could be the projection of the helically twisted loops . to understand associated high energy emission , we present co - temporal hxr images at 1225 ( orange ) and 2540 kev ( red ) in figure [ kink](c ) .
we have studied the evolution of rhessi x - ray spectra during the flare over consecutive 20 s intervals from 05:17 ut to 05:25 ut .
further , we have analyzed spectra for a few selected intervals during the pre - flare peaks .
for this analysis , we first generated a rhessi spectrogram with an energy binning of 1/3 kev from 6 to 15 kev , 1 kev from 15 to 100 kev , and 5 kev from 100 to 200 kev energies .
we only used the front segments of the detectors , and excluded detectors 2 and 7 ( which have lower energy resolution and high threshold energies , respectively ) .
the spectra were deconvolved with the full detector response matrix ( i.e. , off - diagonal elements were included ; @xcite ) .
the time intervals for spectral analysis are carefully chosen to avoid the bad intervals due to changes in the attenuator states . in our case , there were two bad intervals during 05:18:0405:18:08 ( change in attenuator state from a@xmath19 to a@xmath20 , i.e. , thin shutter comes in ) and 05:21:2405:21:28 ( change in attenuator state from a@xmath20 to a@xmath21 , i.e. , thin+thick ( both ) shutters are in ) . in figure [ hsi_spectra ] ,
we illustrate spatially integrated , background subtracted rhessi spectra along with their respective residuals for six selected intervals .
spectral fits were obtained using a forward - fitting method implemented in the ospex code . the hxr emission during pre - flare peaks was observed only up to 20 kev and can be well fitted with an isothermal model ( see figure [ hsi_spectra](a ) ) . during impulsive phase , we note hxr emission up to 100 kev . for this phase , spectral fittings were obtained using the bremsstrahlung and line spectrum of an isothermal plasma and a non - thermal thick - target bremsstrahlung model ( i.e. , combination of vth and thick2 models ; see figures [ hsi_spectra](b)-(f ) ) .
also , we have chosen different time intervals for fitting during pre - flare peaks ( 1 minute ) and impulsive phase ( 20 s ) for better counts statistics . in figure
[ plasma_param ] , we plot various spectroscopic parameters derived from spectral fittings : temperature ( @xmath22 ) and emission measure ( @xmath23 ) from thermal fits ; low - energy cutoff ( @xmath24 ) and electron spectral index ( @xmath25 ) from non - thermal thick - target fits .
rhessi observations reveal a very high plasma temperature ( @xmath2632 mk ) at the time of first hxr peak at 05:18 ut ( see figure [ plasma_param](a ) ) while @xmath23 remains at lower values , which suggest intense heating within localized regions .
later on , both , the @xmath22 and @xmath23 exhibit smooth variations ( see figures [ plasma_param](a ) and ( b ) ) . from the thermal parameters , we have further estimated density ( @xmath27 ) and pressure ( @xmath28 ) of the flaring region ( see figure [ plasma_param](c ) ) .
the density of the thermal plasma is inferred by the formula , @xmath29=@xmath30 , where @xmath23 is the emission measure , @xmath31 is the source volume , and @xmath32 is the volume filling factor .
the volume has been estimated by the relation , @xmath33=@xmath34 , where a is the area of x - ray sources at 50% contour level of the corresponding 615 kev rhessi images reconstructed with the clean algorithm . the filling factor ( @xmath35 ) , ratio of the volume filled by hot x - ray emitting plasma to the total source volume ,
is often assumed to be unity ( see , e.g. , * ? ? ?
the pressure of the plasma in thermal emission has been estimated by the formula , @xmath28=@xmath36__k@xmath37__@xmath22 , where _ k@xmath37 _ is boltzmann s constant .
it is noticeable that the density and pressure show a slow yet steady rise until the second hxr peak at @xmath105:21 ut , but exhibit rather oscillatory behaviour in the later stages ( see figure [ plasma_param](c ) ) .
the evolution of low - energy cutoff ( e@xmath38 ) is shown in figure [ plasma_param](d ) .
the e@xmath38 values derived are the upper estimates due to the dominance of the hot thermal contribution , and thus yield lower limits to the number of non - thermal electrons and their energies derived @xcite .
the temporal evolution of electron spectral index ( @xmath25 ) clearly indicates spectral hardening at the two hxr peaks ( at 05:18 and 05:21 ut ) with electron spectral index ( @xmath25 @xmath15 ; see figure [ plasma_param](e ) and spectra in figures [ hsi_spectra](b ) and [ hsi_spectra](d ) ) which suggest stronger acceleration of high energy particles at these epochs .
[ spectroscopy ] in the preceding subsection ( [ spectroscopy ] ) , we have estimated various spectroscopic parameters ( @xmath39 , @xmath40 , @xmath25 , @xmath41 , and @xmath42 ) which are used to evaluate thermal and non - thermal energies associated with the flare . using instantaneous values for @xmath43 and @xmath39 , the total thermal energy content can be estimated by following expressions @xmath44,\end{aligned}\ ] ] where _
k@xmath37 , t , em , v , _ and _ f _ are the boltzmann constant , the plasma temperature ( k ) , the emission measure ( @xmath45 ) , the source volume ( @xmath46 ) and volume filling factor , respectively . in accordance with previous rhessi studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , we assume unity for the filling factor ( @xmath32 ) .
recently @xcite determined from rhessi imaging spectroscopy for a number of extended loop flares that the filling factor lies within the range 0.1 and 1 which they interpret as somewhat less than , but consistent with unity .
the evolution of @xmath47 is shown in figure [ plasma_param2](a ) .
we find that @xmath47 gradually builds up till the second hxr peak ( 05:21 ut ) and remain roughly constant till the end of impulsive phase .
@xmath48 maximizes @xmath02 minutes after peak hxr flux and resembles the goes sxr profiles ( see figure [ plasma_param2](a ) ) . in collisional thick - target model , the hxrs
are produced by collisional bremsstrahlung during the passage of non - thermal electrons through denser plasma regions , in which the electrons are stopped completely by coulomb collisions @xcite . using the electrons distribution parameters derived from rhessi spectroscopy , the power delivered by non - thermal electrons above low - energy cutoff ( @xmath42 ) can be calculated by the following expression @xmath49 , \label{eq2}\end{aligned}\ ] ] where @xmath24 is the low - energy cutoff , @xmath50 is the total number of electrons per second above @xmath24 in units of 10@xmath51 electrons s@xmath9 , and @xmath52 is the electron spectral index @xcite
. an accurate determination of the low - energy cutoff to non - thermal electron distributions is crucial for the calculation of power and consequently non - thermal energy in solar flares @xcite . in general flares
are thought to have low - energy cutoffs close to or in the region where the emission is dominated by thermal bremsstrahlung @xcite .
we further emphasize that in flares with multiple hxr sub - peaks during the impulsive phase ( like the present one ) , the determination of e@xmath53 is rather illusive during the peak emission as it is difficult to distinguish the signals of flare accelerated electrons against dominant thermal bremsstrahlung .
this often results in the overestimation of e@xmath53 during hxr peak phases . for this event
, we find that e@xmath54 varies in the range of 20 - 32 kev with relatively higher values ( @xmath032 kev ) during the second peak ( see figure [ plasma_param](d ) ) .
therefore , in order to get an estimation of power ( and consequently energy ) of the non - thermal electrons , we have taken e@xmath54 as 25 kev which is the average of e@xmath54 over the flare time interval .
the estimated non - thermal electron flux ( @xmath55 ) above 25 kev is plotted in figure [ plasma_param2](b ) . in figure [ plasma_param2](c ) , we have plotted the power ( @xmath56 ) and energy ( e@xmath57 ) contained in non - thermal electron beams .
the plots of @xmath58 and @xmath56 clearly indicates noticeable enhancement in the particle rate and consequently power of flare - accelerated electrons during the two hxr peaks .
in this paper , we study coronal events that occurred in active region noaa 10646 on 2004 july 14 .
we observed large - scale contraction of higher active region loops for a time span of @xmath130 minutes which was followed by an m6.2 flare and associated failed eruption of a flux rope .
table [ tab1 ] presents a summary of the different phases of flare evolution and associated phenomena .
p1.6inp1.4inp3 in + phases & time & observations + + pre - flare activity & 4:30
05:17 ut & three small goes flare events , euv loop contraction for @xmath130 min , drastic change in speed of contraction during third peak of pre - flare phase , pre - flare events showing localized brightenings in euv 171 , mw and x - ray sources at the core region underneath the contracting coronal loops . + contraction phase & 04:47 05:17 ut & large overlying coronal loops observed at euv 171 undergo contraction by @xmath020 mm ( 40% of original height ) , loop contraction continues through all three episodes of pre - flare activity .
+ flare impulsive phase and rapid activation of flux rope & 05:17 05:25 ut & peak hxr emission up to 100 kev with electron spectral index @xmath59 @xmath05 , multiple peaks in hxr and mw ( 17 and 34 ghz ) emission , rapid rise of flux rope by @xmath040 mm . + decay phase and confinement of eruption & 05:25 05:35 ut & braking and successive disruption of eruption front , multiple mw sources from footpoint and coronal regions , compact hxr footpoint source . +
[ tab1 ] although the active region noaa 10646 had a relatively simple bipolar magnetic structure in the photosphere , it exhibited very complex structures of overlying loops in the corona ( see figure [ loops ] ) .
the euv images at 171 clearly show large - scale contraction of a system of higher coronal loops that started @xmath130 minute prior to the m6.2 flare .
it is important to note that the loop contraction was observed throughout the pre - flare phase and ceased with the onset of the flare impulsive phase .
we observed three small events during the pre - flare phase ( events i , ii , and iii ; figures [ goes_profile ] and [ pre-2 ] ) which are characterized by localized brightenings in the core region of the overlying loops .
the speed of loop contraction increased drastically ( from @xmath15 km s@xmath9 to @xmath125 km s@xmath9 ; figure [ imp_ht ] ) just @xmath15 minute prior to the impulsive phase of the m - class flare , i.e. , during the pre - flare event iii .
rhessi measurements during the pre - flare events reveal hxr emission up to 20 kev . from the x - ray spectra , we find that the x - ray emission was predominantly thermal during this phase with plasma temperatures @xmath6020 mk during pre - flare peaks ( see e.g. , figure [ hsi_spectra](a ) ) .
these observations indicate that the flaring site was already enveloped by hot plasma before the onset of filament activation and associated m6.2 flare .
it is worth to emphasize that the m6.2 flare occurred at the location of pre - flare brightenings which was enveloped by the large contracting coronal loop system .
although the phenomenon of loop contraction has been observed in several recent studies , the contraction reported here is remarkable in several aspects .
first , the contraction was observed in overlying coronal loops at large - scales both at spatial and temporal domains during which the loop height decreased by @xmath120 mm ( @xmath140% of original height ) .
the total duration of loop contraction is @xmath130 minutes which is the longest period of loop contraction reported so far .
further , the contraction phase ends with the onset of the impulsive phase of the m6.2 flare , i.e. , the onset of the impulsive phase can be treated as the transition from inward to outward motion of coronal loops .
the investigations of loop contraction have emerged as a very important aspect of solar eruptions in recent times .
these studies are essentially inspired by the rhessi discovery of altitude decrease in the hxr loop top ( lt ) source during the early impulsive phase in sol2002 - 04 - 15 of class m1.2 @xcite .
this phenomenon of downward motion of hxr lt source was established by many subsequent rhessi observations in flares of different intensity classes ( i.e. , from class c to x ) during their earliest stages to the impulsive phase @xcite .
motivated by the rhessi observations , the dynamics of complex coronal loop system over the flaring core was extensively investigated in euv images .
these studies reveal a significant contraction of coronal loops that envelop the flaring region before and during the flare ( e.g. , * ? ? ?
? * ; * ? ? ?
* ) which was attributed to ` magnetic implosion ' @xcite by many authors @xcite .
@xcite provided the first evidence for the contraction of euv coronal loops that lasted for @xmath15 minutes from trace 195 images which was temporarily and spatially correlated with the shrinkage in rhessi 1225 kev lt source during the early impulsive phase of sol2002 - 04 - 16 flare of class m2.5 .
the study of a long duration m7.6 event ( sol2003 - 10 - 24 ) , characterized by a prolonged rise phase of @xmath120 minutes , provided one of the best examples for the contraction of coronal loops @xcite . in this event , trace 195 loops and hxr lt sources underwent a shrinkage of @xmath135% of initial height during @xmath111 minutes .
more importantly , the downward motion was observed simultaneously with speed of @xmath115 km s@xmath9 in euv loops as well as hxr lt sources in different energy bands , viz .
, 612 , 1225 , and 2550 kev . due to observational limitations ,
it is usually not possible to probe the motions of flare associated lt and footpoint ( fp ) sources simultaneously .
nevertheless some uniquely observed events have given us the opportunity to examine the relationship between the dynamics of lt and fp sources in the early flare phases @xcite .
these studies indicate that shrinkage in lt sources are associated with converging fp motions , making us to predict that simultaneity of the two phenomena could be part of the coronal implosion .
@xcite presented observations of contraction of large coronal loops during the early impulsive phase of sol2005 - 07 - 30 of class c8.9 .
this contraction was found in three clusters of euv loops observed at 171 that sustained for a longer interval with relatively slower speed ( @xmath110 minutes with an average speed of @xmath15 km s@xmath9 ) .
recently , @xcite reported loop contraction in euv 171 images during the impulsive phase of m6.4 flare sol2012 - 03 - 09 that was associated with loop oscillations .
it is widely believed that coronal transients derive their energy from the energy stored locally in the coronal magnetic fields .
@xcite conjectured that the energy conversion process during transient events would involve a magnetic implosion when the following assumptions hold : ( 1 ) the energy required for the event must come from the corona directly ; ( 2 ) gravitational potential energy plays no significant role ; ( 3 ) low plasma @xmath61 in the corona . with these assumptions ,
the conservation of energy implies that magnetic energy decreases between the static states before and after the energy release .
the reduction of magnetic energy , @xmath62 dv , and consequently the reduction of the magnetic pressure , b@xmath63 , would inevitably result in the contraction of overlying field lines so as to achieve a new force balance .
the observations of the contraction of overlying coronal loops during and before the impulsive phase of flares by high temporal cadence images from trace and sdo support the predictions of the conjecture by @xcite . however , the observational evidences for the contraction in coronal loops are uncommon for the majority of flares that exhibit explosive rather than implosive behaviour .
we note that in our observations prolonged coronal implosion displays three distinct stages with varying speed of loop contraction ( figure [ imp_ht](a ) ) . here
it is important to focus on the brief and localized x - ray and mw emissions that were observed from the core region of the overlying loop system a few minute before the onset of contraction ( figures [ pre-2](b)(d ) ) .
later on this region brightened up in euv images also with the localized brightening of multiple low - lying loops .
more importantly , the contraction speed rapidly increased @xmath15 minute before the flare onset and this phase is co - temporal with the sequential brightenings of low - lying coronal loops at the core region during which x - ray and mw emissions intensified ( figure [ pre-2 ] ) .
the observations of sequential and localized brightenings during pre - flare phase imply episodic release of small amount of energy which can be considered as the signature of magnetic reconnection at relatively small - scales ( in contrast to large - scale reconnection during the flare impulsive phase ) @xcite .
rhessi observations of this phase further indicate that this interval is mostly dominated by the thermal emission .
the slow and gradual energy release over the prolonged pre - flare phase will cause the magnetic pressure of the coronal loop system to decrease .
it is likely that during this slow preheating phase , the increase of the thermal pressure is mainly due to localized heating and gentle evaporation which is not enough to compensate for the decrease of the magnetic pressure ( see * ? ? ?
this will result in the implosion of the surrounding flaring region consisting of euv coronal loops .
we therefore conjecture that multiple small - scale , localized events of energy release at the core region over the prolonged preheating phase favourably contributed to sustain a large - scale contraction of coronal loops . with the onset of the ` standard flare reconnection ' during the impulsive energy release , the implosion becomes insignificant as the large - scale magnetic reconnection proceeds successively in higher coronal loops stretched by the eruption of the prominence or magnetic flux rope .
we examine the temporal and morphological evolution of mw and hxr emission during the impulsive phase ( i.e. , shaded area in figure [ flux_profile ] ) and their associations with the prominence activation .
the comparisons of spatial and temporal structures of hxr and mw emission offer us crucial information on different aspects of conditions in the solar atmosphere where the flare occurs as both types of emissions are produced by highly energetic electrons but with different emission mechanisms .
the hxr emission most likely arises from bremsstrahlung produced when energetic electrons are decelerated by coulomb forces in collisions with the ambient ions , either in the chromosphere or in the corona .
bremsstrahlung hxr emission is proportional to the product of the non - thermal electron density and the ambient ion density . on the other hand ,
non - thermal mw emission is produced by the gyrosynchrotron mechanism which depends on the magnetic field intensity and its direction .
the gyrosynchrotron production mechanism is very efficient and allows us to detect electrons at energies of hundreds of kev , even when their numbers are very low .
microwave emission can also be produced by thermal electrons through bremsstrahlung in sufficiently dense thermal plasma ( for a review see * ? ? ?
the impulsive phase is characterized by three events of flux enhancements in mw emission denoted by ` a ' , ` b ' , and ` c ' . it is worth to note that the events ` a ' and ` c ' occurred simultaneously in mw and hxr profiles while ` b ' is largely missing in hxr measurements .
the spatial evolution of hxr emission with respect to euv images ( see figure [ mainevent ] ) provides some important insights to understand the relationship between flare emission and phases of prominence activity .
we find that at the very beginning of the impulsive phase ( @xmath05:17 ut ; figure [ mainevent](a ) ) , two distinct 1225 kev hxr sources are observed and at this time the flare related euv brightening at hxr source locations is insignificant .
soon afterwards , non - thermal high energy hxr emission ( at 40@xmath64100 kev energy band ) was observed from the same location with two distinct , well separated emission centroids ( see figure [ mainevent](b ) ) .
this phase corresponds to the first peak of the impulsive phase ( i.e. , event ` a ' marked in figure [ flux_profile ] ) . at this time ,
plasma temperature ( estimated from rhessi spectra ; figure [ hsi_spectra](b ) ) raised to @xmath032 mk which also corresponds to the maximum temperature during the flare ( figure [ plasma_param](a ) ) .
the high temperature and low emission measure ( figure [ plasma_param](a ) and ( b ) ) indicate impulsive heating of plasma within a small volume which is also consistent with corresponding euv images that show localized brightening .
we also note that this very epoch is associated with significant non - thermal characteristics with an electron spectral index @xmath59 @xmath05 ( figures [ hsi_spectra ] and [ plasma_param ] ) .
it is further noteworthy that during event ` a ' , high energy hxr source at 40@xmath64100 kev resembles the mw source at 34 ghz ; both show emission from an extended region with the appearance of two distinct well separated emission centroids ( see also figures [ mainevent](b ) and [ tr_norh](a ) ) . in figure [ mdi_hsi ] , these conjugate emission centroids are indicated on combined wl / magnetogram image as s1 and s2 .
the associations of s1 and s2 with opposite polarity magnetic regions suggest that they probably represent emissions from the footpoints of a flaring loop .
on the other hand the 17 ghz source resembles the hxr emitting structure at relatively lower energies ( @xmath6540 kev ) .
we also note that event ` a ' is followed by the rise and subsequent eruption of the prominence .
the peak of the impulsive phase ( i.e. , event ` c ' ) occurred simultaneously in hxr and mw emission . at this time , hxr sources at energies @xmath1525 kev exhibit a single structure ( figure [ mainevent](d ) ) .
the non - thermal hxr emission is generally believed to originate from footpoint locations of coronal loops by thick - target bremsstrahlung , thus observations of pairs of hxr sources are expected .
the single hxr source implies that the two footpoints are very close indicating energy release in small low - lying loops ( e.g. , * ? ? ?
it should be noted that the present event occurred close to the limb , so due to projection effects the footpoint sources may appear much closer than their actual separation .
although this hxr peak ( event ` c ' ) is much broader and intense than the first peak ( event ` a ' ) , the non - thermal electron spectral indices during the two peaks are comparable ( @xmath05 ; see figures [ hsi_spectra ] and [ plasma_param ] ) .
p3.2inp2.0 in + flare characteristics & parameters + + total duration of hxr peaks & 430 s + no . of hxr peaks & 2
+ & 94 s and 336 s + total non - thermal energy ( ( e@xmath66)@xmath67 ) & 3.03 @xmath6810@xmath69erg + thermal energy ( e@xmath70 ) & + (e@xmath71)@xmath72 & 3.89 @xmath6810@xmath73 erg + (e@xmath71)@xmath74 & 0.33 @xmath710@xmath73 erg + @xmath75 & @xmath17.5 + [ tab2 ] in table [ tab2 ] , we summarize various aspects of energy release during the impulsive phase of the flare ( see also figure [ plasma_param2 ] ) .
the total duration of hxr impulsive phase ( as revealed by hxr flux profile @xmath6025 kev ) is 430 s which is composed of two peaks of 94 s and 336 s durations ( see figure [ plasma_param2](d ) ) .
we find that the power delivered by the non - thermal electrons ( p@xmath76 ) was maximum during the peak of hxr impulsive phase ( event ` c ' ) while the thermal energy e@xmath70 maximized @xmath12 min later which nearly coincides with the peak of thermal sxr emission ( figures [ plasma_param2](a ) and ( c ) ) .
the total non - thermal energy ( ( e@xmath76)@xmath67 ) emitted during this whole impulsive phase was @xmath13.0@xmath710@xmath69erg while the maximum thermal energy ( ( e@xmath77)@xmath72 ) was noted at the end of impulsive phase as @xmath13.9@xmath710@xmath73ergs , yielding a ratio of ( e@xmath78)@xmath67/(e@xmath77)@xmath72 @xmath17.5 .
@xcite examined the ratio of non - thermal to thermal energies for a set of c- and m - class flares and found that this ratio varies in the range of @xmath11.56 .
it was also noted the flares with longer hxr peaks ( @xmath15200 s ) display larger ratios of non - thermal to thermal energy .
considering that the present event displays a longer hxr impulsive phase of 430 s with two sequential hxr peaks , the ratio of @xmath17.5 is in good agreement with the results of @xcite . in order to understand the relationship between the flare - accelerated electron beam and the flare - associated emission from thermal plasma
, we compare the hxr profile ( rhessi 25100 kev ) with the derivative of sxr flux ( goes 18 ) in figure [ plasma_param2](d ) .
we obtained a good temporal consistency between the two curves which suggests that this flare , associated with confined eruption , follows the neupert effect @xcite .
the consistency with the neupert effect is further complemented by the fact that the curve showing the evolution of cumulative non - thermal energy ( figure [ plasma_param2](c ) ) exhibits a good temporal correspondence with the evolution of thermal energy ( figure [ plasma_param2](a ) ) indicating that the non - thermal energy is eventually converted into the thermal energy @xcite .
the validity of the neupert effect implies that the energetic electrons responsible for the hxr emission by thick - target collisional bremsstrahlung ( imaged as high energy hxr sources ) are the main source of heating and mass supply of the sxr emitting hot coronal plasma @xcite .
we also note that with the onset of prominence activation and increase of flare emission after event ` a ' , there is an enhancement in the density and pressure of the thermal plasma ( figure [ plasma_param](c ) ) . with the onset of event `
c ' , we observe a rapid increase in the speed of the erupting prominence ( figure [ exp_ht ] ) .
it is important to note that hxr images at high energies ( @xmath1525 kev ) do not show significant changes in the location and structure of the hxr source , i.e. , a single compact source is observed that remained at the same location .
we note that the eruption proceeded symmetrically till @xmath105:24 ut with the distinct appearance of an intact eruption front ( shown by arrow in figure [ mainevent](f ) and ( g ) ) .
thereafter , the eruption front evolved into an asymmetrical structure ( first seen in euv image at 05:24:30 ut ) with the disruption of the south - west part of the eruption front ( marked by arrow in figure [ mainevent](h ) ) . in the successive images ,
we clearly observe the downfall of the erupted material which eventually leads to the complete failure of the filament eruption without any cme . @xcite and
@xcite have carried out detailed investigations to understand the reason for the confinement of the prominence for this event .
they found strong evidence that the interaction between overlying coronal magnetic fields with the erupting prominence was capable to suppress the eruption completely .
we have recognized some interesting morphological structures associated with the erupting prominence that overlap the flare s impulsive phase ( @xmath005:21 - 05:23 ut ) and highlighted in euv 171 images in figure [ kink ] .
here we emphasize that this phase corresponds to the interval when the erupting prominence attained the maximum speed of @xmath0242 km s@xmath14 ( figure [ exp_ht ] ) .
these structures consists of rapidly expanding bright twisted loops at the front and spiky patterns at the following part which very likely represent the helically twisted loops associated with the erupting prominence .
these structures can be attributed to the portions of the heated prominence within the flux rope .
the hxr source at 2540 kev along with corresponding spectrum ( figure [ hsi_spectra](f ) ) presents evidence for non - thermal emission from accelerated electrons in that region in addition to hot plasma .
further , from the euv images , it is apparent that the magnetic field lines underwent curving toward the bright central region after being stretched by the erupting prominence - flux rope system ( shown by arrows in figures [ kink](b ) , ( c ) , and ( e ) ) .
flux ropes are considered to be an important structural component in the models of solar eruptions .
many recent studies validate the existence of flux ropes in the lower corona in euv observations , mostly in hot euv channels @xcite . here
we emphasize that evidence for a flux rope was found just after the peak ` c ' ( the largest peak ) of the flare impulsive phase ( figure [ flux_profile ] ) .
it is probable that the flux rope was formed following this most violent episode of energy release .
@xcite presented a very clear example of formation and rise of an euv flux rope structure during the impulsive phase of an x - class flare .
these observations suggest that a flux rope has a multi - temperature structure that possesses several structural components , such as , sigmoid , hot plasma blob or plasmoid , leading edge , etc . in our observations ,
the flux rope structure was seen in euv 171 ( which represents plasma at a temperature of @xmath01 mk ) presumably due to the heating of the prominence that lies within the flux rope .
@xcite illustrated three kinds of magnetic topology that can lead to the diversity of eruptive phenomena : full , partial and nil @xcite . according to this study
, magnetic reconnection can occur completely above the system of a prominence and its supporting flux rope or within it leading to nil or partial eruption of the filament .
we believe that , the eruption of bright prominence material along with the helically twisted loops represents the case of magnetic reconnection occurring within the system of prominence and supporting magnetic flux rope . however , we believe that this is just a part of the flux rope which contain the prominence body .
it is very likely that the remaining part of the flux rope , devoided of the prominence , might exist at a very high temperature which is not visible in 171 euv images .
mw sources at 17 and 34 ghz show an interesting evolution from lower as well as higher coronal regions ( figure [ tr_norh ] ) .
in general , we note significant differences in the morphology of mw and hxr sources although one of the mw sources ( which appeared at the early impulsive phase ; see figure [ mainevent](h ) ) exhibits spatial correlation with the hxr source throughout the flare . during event ` a ' ( i.e. , between 05:1805:19 ut ) , we observe a 34 ghz mw source with two distinct centroids having a separation of @xmath118@xmath8 ( see figure [ tr_norh](a ) and ( b ) ) . on the other hand ,
17 ghz images ( note that 17 ghz images have lower spatial resolution than 34 ghz images ) show a relatively extended single source . as discussed earlier , the conjugate centroids ( s1 and s2 ; figures [ mdi_hsi ] and [ tr_norh](a ) ) , observed at high energy hxr ( 40@xmath64100 kev ) and 34 ghz mw emissions , exhibit structural similarities and co - spatiality during event ` a ' . at the second peak ( event ` b ' ; see figure [ flux_profile ] ) , images at both mw frequencies present similar source structures with a single centroid ( see figure [ tr_norh](c ) ) . during the third mw burst at 05:21
ut ( event ` c ' ; figure [ flux_profile ] ) , a new mw source originated at a distance of @xmath533@xmath79 toward the south - west of the pre - existing source ( see figure [ tr_norh](d)-(i ) ) .
we mark the newly developed source as s3 ( figure [ tr_norh](d ) ) and show its location on wl / magnetogram in figure [ mdi_hsi ] .
this new source s3 prevailed till the end of the flare ( @xmath105:30 ut ) with more prominent appearance at 17 ghz over 34 ghz .
we also note the source s3 never appeared in hxrs .
the temporal and spatial associations of hxr and mw sources ( also see figure [ mdi_hsi ] ) indicate that the main energy release site lies in coronal loops close to the magnetic neutral line , probably formed by connecting footpoints s1 and s2 .
it is also likely that the distant mw source s3 resulted from the injection of flare accelerated electrons onto an overlying coronal loop ( which probably connects s1 and s3 regions ) as the rising flux rope interacted with overlying loop systems .
we note that norh has much better dynamic range than rhessi and therefore it is capable of detecting secondary sources @xcite .
more importantly , we find multiple mw sources from higher coronal regions that intermittently appear at several locations after @xmath005:22 ut ( figures [ tr_norh](e)(i ) ) .
it is also important to note that the whole flaring region ( i.e. , footpoint and coronal regions associated with the prominence eruption and its subsequent confinement ) brightened up in 17 ghz mw emission following the impulsive phase of the flare . in particular
, distinct mw sources appeared along the narrow , bright region seen in euv images where bright blobs of plasma were observed following the confinement of the eruption ( marked by arrows in figures [ tr_norh](g)-(i ) ) .
these multiple coronal mw sources were observed after the beginning of disruption of the eruption front , i.e. , when the ejected prominence and its supporting flux rope were subjected to confinement by the overlying field lines . in view of this , we conclude that the distinct mw sources in the corona presumably represent emission from hot plasma blobs formed within the collapsing magnetic flux rope .
we have presented a comprehensive multi - wavelength study of an m6.2 flare which is associated with a confined eruption of a prominence - flux rope system . the availability of a wide range of multi - wavelength data ( euv , mw , x - ray ) from the beginning of the pre - flare to the end of the impulsive phase of the event provided us with a unique opportunity to probe several flare associated phenomena in detail : large - scale loop contraction , small - scale loop brightenings , eruption of prominence - flux rope system and its subsequent confinement . in the following ,
we summarize the important results of this study : 1 .
we have reported implosion of overlying coronal loops that continued over 30 minute during which overlying loops underwent contraction by 20 mm ( @xmath140 % of their original height ) during the pre - flare phase .
such a large - scale contraction has been reported for the first time .
we observe episodic and localized events of energy release in low - lying loops at the core of the large overlying loops . by synthesizing the multi - wavelength data , we propose that prolonged loop contraction is a manifestation of localized events of energy release that occurred intermittently during the pre - flare phase of the m6.2 flare .
the pre - flare phase was dominated by thermal emission with temperatures beyond 20 mk .
this indicates that the plasma was already substantially preheated at the flare core before the onset of impulsive phase .
we believe that the strong preheating at the flare core will contribute favorably toward efficient particle acceleration during the subsequent impulsive phase of the event .
the impulsive phase of the flare is characterized by multiple non - thermal peaks .
after the first impulsive peak , associated with strong hxr emission up to 40 - 100 kev , we observe the activation of the prominence and its supporting flux rope .
our observations imply that the onset of impulsive flare emission ( which probably corresponds to large - scale magnetic reconnection ) has triggered the eruption of the flux rope .
4 . rhessi spectroscopy reveals high plasma temperatures ( @xmath130 mk ) and substantial non - thermal characteristics with electron spectral index ( @xmath805 ) during the impulsive phase of the flare . during this phase , characterized by two sequential hxr peaks
, the ratio of total non - thermal energy to maximum thermal energy was found to be @xmath17.6 which is consistent with earlier studies .
more importantly , the time - evolution of thermal energy nicely correlates with the variations of the cumulative non - thermal energy throughout the impulsive phase of the flare .
this can be interpreted in terms of conversion of the energy of accelerated particles to hot flare plasma and is well consistent with the neupert effect .
the prominence along with its supporting flux rope could not have a successful escape through the overlying coronal loops and therefore leads to a confined eruption .
the observations of the confinement process of the flux rope are remarkable ; we detect multiple coronal mw sources along the trajectory of the eruption .
the euv images show hot plasma blobs on the location of these coronal sources . in our opinion ,
the coronal mw sources represent emission from hot plasma blobs which are formed within the collapsing magnetic flux rope as a result of its interaction with the overlying and surrounding magnetic field lines .
this paper highlights the importance of studying the pre - flare activity .
this study also underlines that confined eruptions form a very interesting category of solar eruptive phenomena .
their investigations provide a unique opportunity to probe the interaction among different magnetic field systems in the corona besides exploring the triggering mechanisms and energy releases processes .
we acknowledge rhessi , trace , norh , soho , and goes for their open data policy .
rhessi and trace are nasa s small explorer missions .
soho is a joint project of international cooperation between the esa and nasa .
this work was supported by the bk21 plus program through the national research foundation ( nrf ) funded by the ministry of education of korea .
a.m.v . gratefully acknowledges the austrian science fund ( fwf ) : p27292-n20 .
we sincerely thank the anonymous referee for providing constructive comments and valuable suggestions that have enhanced the quality and presentation of this paper . | we present a detailed multi - wavelength study of the m6.2 flare which was associated with a confined eruption of a prominence using trace , rhessi , and norh observations .
the pre - flare phase of this event is characterized by spectacular large - scale contraction of overlying extreme ultraviolet ( euv ) coronal loops during which the loop system was subjected to an altitude decrease of @xmath020 mm ( 40% of the initial height ) for an extended span of @xmath030 min .
this contraction phase is accompanied by sequential euv brightenings associated with hard x - ray ( hxr ) ( up to 25 kev ) and microwave ( mw ) sources from low - lying loops in the core of the flaring region which together with x - ray spectra indicate strong localized heating in the source region before the filament activation and associated m - class flare . with the onset of the impulsive phase of the m6.2 flare , we detect hxr and mw sources that exhibit intricate temporal and spatial evolution in relation with the fast rise of the prominence . following the flare maximum , the filament eruption slowed down and subsequently confined within the large overlying active region loops ; the event did not lead to a coronal mass ejection ( cme ) . during the confinement process of the erupting prominence
, we detect mw emission from the extended coronal region with multiple emission centroids which likely represent emission from hot blobs of plasma formed after the collapse of the expanding flux rope and entailing prominence material .
rhessi observations reveal high plasma temperature ( @xmath130 mk ) and substantial non - thermal characteristics with electron spectral index ( @xmath25 ) during the impulsive phase of the flare . the time - evolution of thermal energy exhibits a good correspondence with the variations in cumulative non - thermal energy which suggest that the energy of accelerated particles efficiently converted to hot flare plasma implying an effective validation of the neupert effect . |
this paper is about three - dimensional chern - simons gauge theories .
these theories have a variety of applications in topology , conformal field theory , the ads / cft correspondence , and in the fractional quantum hall effect ( fqhe ) . here
we will investigate quantum equivalences between chern - simons theories with gauge group @xmath2 . in @xcite witten
pointed out that nontrivial quantum equivalences between such theories exist .
this raises the question of the classification of such theories .
the present paper answers that question .
[ [ classical - spin - chern - simons - action . ] ] classical ( spin ) chern - simons action .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the classification of abelian chern - simons theories was mentioned in @xcite but , especially in view of the potential applications to the fqhe , the problem of real interest is the classification of _ spin _ chern simons theories , i.e. chern - simons theories at half - integer level .
we now briefly recall these theories , which were originally introduced in @xcite . for any compact gauge group @xmath3
the chern - simons actions are classified by an element @xmath4 @xcite . for @xmath5 , @xmath6 , and
the element is simply the integer @xmath7 appearing in the familiar expression for the @xmath8 chern - simons action on a 3-manifold @xmath9 : @xmath10\ ] ] ( in this paper the connection is normalized so that @xmath11 has integral periods . ) as usual the expression is defined by choosing an extension to a bounding 4-manifold @xmath12 and defining @xmath13.\ ] ] where @xmath14 is the fieldstrength of an extending connection .
this expression does not depend on the extension provided that @xmath7 is integral .
this follows since @xmath15 on a closed 4-manifold is an integer .
since it can be any integer the minimal value of @xmath7 is @xmath16 .
now , if we endow @xmath9 with a spin structure and require that the bounding 4-manifold @xmath12 have a compatible spin structure then @xmath17 is always an _ even _ integer and hence @xmath7 can be half - integral .
this defines _ a spin _ ( or half - integral ) @xmath8 chern - simons theory .
more generally , if the gauge group is a torus @xmath18 then it is convenient to think of @xmath19 as the space of integral _ even _ symmetric bilinear forms . indeed , choosing a basis @xmath20 for @xmath21 an element in @xmath19 can be written as @xmath22 corresponding to the chern - simons action @xmath23 = \exp\bigl [ i \pi \int_{z_4 } m_{\alpha\alpha}\tilde{f}^\alpha \wedge \tilde{f}^\alpha + \sum_{\alpha<\beta}2m_{\alpha\beta}\tilde{f}^\alpha \wedge \tilde{f}^\beta\bigr ] .
\label{csaction}\ ] ] this expression does not depend on the extension provided that @xmath24 is an even integer and @xmath25 for @xmath26 is an arbitrary integer .
thus the matrix @xmath25 indeed defines an integral even symmetric bilinear form .
every integral symmetric bilinear form defines a lattice .
thus the chern - simons actions with the gauge group @xmath2 are classified by _ even _ integral lattices of rank @xmath27 . now
, if we endow @xmath9 with a spin structure and require that the bounding 4-manifold @xmath12 have a compatible spin structure then @xmath28 is always an _ even _ integer and hence @xmath24 in can be any integer , not necessary even .
thus _ spin _ chern - simons theories are classified by integral symmetric bilinear forms , or more invariantly by integral ( generically odd ) lattices .
having an integral symmetric bilinear form @xmath29 we define an action for _ spin _ chern - simons theory by @xmath30 .
\label{csaction}\ ] ] we denote by @xmath31 the lattice @xmath32 with the bilinear form @xmath33 . in general the spin chern - simons theories are classified by a group in a certain cohomology theory denoted by @xmath34 @xcite . in the case at hand @xmath35
is isomorphic to the space of arbitrary integral symmetric bilinear forms .
further details about spin chern - simons theories can be found in section 2 .
we will assume the bilinear form is nondegenerate and also that the induced form on @xmath36 is nondegenerate .
[ [ the - quantum - equivalence . ] ] the quantum equivalence .
+ + + + + + + + + + + + + + + + + + + + + + + + while the classification of abelian spin chern - simons actions is the classification of integral lattices , there are , as we have mentioned , quantum equivalences between theories with different classical actions .
we will define two theories to be equivalent if they have the same 3-manifold invariants .
equivalently , we declare them to be equivalent if the `` conformal blocks '' ( which can be nonholomorphic ) transform in equivalent projective representations of the modular group . the main result of this paper is a theorem on the classification of the quantum theories under this equivalence relation .
our result says that up to isomorphism the quantum theory is determined by the following @xmath37 invariants of @xmath31 subject to a constraint : * data : * 1 .
@xmath38 the signature of the bilinear form on @xmath31 modulo @xmath39 .
2 . discriminant group @xmath40 with the bilinear form @xmath41 inherited from the bilinear form on @xmath42 .
3 . the equivalence class of a quadratic refinement
@xmath43 $ ] of the bilinear form @xmath44 on @xmath45 .
the quadratic refinement is a map @xmath46 such that for any elements @xmath47 of @xmath45 we have @xmath48 we say that two quadratic refinements @xmath49 and @xmath50 are equivalent if there exists @xmath51 such that for all @xmath52 @xmath53 * constraint : * the constraint on the data follows from the gauss - milgram sum formula @xcite .
note that if @xmath54 is a quadratic refinement then so is @xmath55 for any constant @xmath56 .
we fix @xmath57 $ ] by the requirement that : @xmath58 notice that the constraint does not depend on a particular choice of representative @xmath54 of the equivalence class @xmath57 $ ] . for an arbitrary lattice @xmath31
the constraint is very restrictive and defines a unique equivalence class of quadratic refinements .
the construction goes as follows .
denote by @xmath59 a representative of the characteristic class @xmath60\in\lambda^*/2\lambda^*$ ] : for all @xmath61 it satisfies @xmath62 .
then the quadratic refinement satisfying may be represented by @xmath63 where @xmath64 denotes the bilinear form on @xmath42 .
the equivalence class of the quadratic refinement is filled out by choosing different representatives @xmath59 of the characteristic class @xmath60 $ ] .
evidently this set of quadratic refinements satisfies the equivalence relation .
this result raises the natural question of whether there is a converse result .
using classical theorems about lattices ( see @xcite ) one can prove that every quartet @xmath65)$ ] satisfying equation is realized by some lattice .
moreover if there exists a representative @xmath54 of @xmath57 $ ] such that @xmath66 then the quartet is realized by an _ even _ lattice .
thus we have the following * classification of quantum chern - simons theories : * 1 .
the set of quantum _ spin _ chern - simons theories is in one - to - one correspondence with the quartets @xmath65)$ ] subject to the gauss - milgram constraint .
the set of quantum _ integral _ chern - simons theories is in one - to - one correspondence with the quartets subject to the gauss - milgram constraint such that there exists a representative @xmath54 of the equivalence class @xmath57 $ ] with @xmath66 .
this is the main result of this paper .
we prove the above result by careful construction of the groundstate wavefunction ( or conformal blocks ) of the theory on a compact riemann surface of genus @xmath67 .
this gives a very explicit projective representation of the modular group .
the level @xmath7 @xmath8 chern - simons theory was carefully constructed in @xcite . in the language of this paper , ref .
@xcite investigates the theory corresponding to the quadratic form @xmath68 .
the space of quantum chern - simons theories has a structure of monoid ( a semigroup with an identity element ) .
the sum operation is defined by @xmath69 ) \oplus ( \sigma_2\mod 24,\,\mathscr{d}_2,\,b_2,\,[q_2 ] ) = ( \sigma_1+\sigma_2\mod 24,\,\mathscr{d}_1\oplus\mathscr{d}_2,\,b_{12},\,[q_{12}])\ ] ] where the bilinear form @xmath70 and the quadratic refinement @xmath71 are defined by @xmath72 the quadratic refinement defined in this way satisfies the gauss - milgram constraint with the signature @xmath73 . the identity element is the trivial quartet .
one can ask whether the space of quantum chern - simons theories has a group structure .
the answer appears to be _
no_. see the end of section [ sec : sum ] for more details .
[ [ fractional - quantum - hall - effect . ] ] fractional quantum hall effect .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we hope that our result will be useful in the theory of the fractional quantum hall effect ( fqhe ) .
it has been known for some time @xcite that the classification of quantum hall fluids is related to the classification of quantum chern - simons theories .
most of these discussions have been phrased in terms of the classification of lattices . in section [ sec : qfhe ] we recall some aspects of the relation to the fqhe .
one consequence of our result is that an effective theory with invariants 1,2,3 specified above can give the following fractional hall conductivities @xmath74 in the ground state : @xmath75 the electric charges ( modulo @xmath16 ) of the vortices @xmath52 are given by @xmath76 .
an example of this construction is given in section 7 .
the paper is organized as follows . in section [ sec :
intro ] we review the construction of spin chern - simons actions . in section [ sec : ham ] we consider hamiltonian quantization of the spin chern - simons theory . in section [ sec : gauss ] we formulate the quantum gauss law . in section [ sec : wave ] we derive the physical wave function and discuss its properties .
section [ sec : sum ] summarizes the main theorem and section [ sec : qfhe ] discusses the fractional quantum hall effect .
first let us recall the construction of chern - simons theories in @xmath37 dimensions .
let @xmath3 be a compact lie group and let @xmath77 be a principal @xmath3-bundle with connection @xmath78 .
choose a real representation @xmath79 and let @xmath80 be the associated bundle .
we will denote a representation of the lie algebra @xmath81 by the same symbol @xmath79 .
if @xmath3 is a connected simply connected compact lie group then the bundle @xmath82 is topologically trivial .
thus @xmath78 is a globally well defined lie algebra valued @xmath16-form on @xmath9 . in this case
the chern - simons functional can be defined by the formula @xmath83 @xmath84 is a @xmath3-invariant quadratic form on the lie algebra @xmath81 .
we temporarily refer to standard normalization of gauge field .
if @xmath82 is a nontrivial @xmath3 bundle then can not be used to define cs theory .
however if there exists a @xmath85-manifold @xmath12 which bounds @xmath9 and an extension of the bundle @xmath82 over @xmath9 to the bundle @xmath86 over @xmath12 then we can define the chern - simons functional by @xmath87 here @xmath88 is the curvature of the extended connection @xmath89 .
notice that if the bundle extends then the connection can always be extended via a partition of unity .
the term @xmath90 appears in the path integral and thus it must be single - valued .
this requires the integral of @xmath91 over an arbitrary closed @xmath85-manifold @xmath12 to be integral . in virtue of its relations to the first pontragin class
it is indeed true .
recall that there is a universal bundle @xmath92 . to construct a @xmath3 bundle over @xmath9
one has to find a classifying map form @xmath9 to @xmath93 , and then take a pullback of the universal bundle .
thus the question about existence of @xmath12 which bounds @xmath9 such that the bundle @xmath94 extends is the question of whether the oriented cobordism group @xmath95 is zero or not .
recall that any compact oriented @xmath37-manifold admits a spin structure @xcite .
if we choose a spin structure on @xmath9 we can modify definition by requiring that the spin structure on @xmath9 must be extendable to @xmath12 .
hence the @xmath85-manifold @xmath12 must also be a spin manifold . in this case
the pontragin class @xmath96 is actually divisible by @xmath97 ( this follows from the index theorem and the quaternionic structure of the spin bundle ) . thus we can define a square root from the chern - simons functional @xmath98 the theory defined by this action is called a _ spin _ chern - simons theory .
the existence of a bounding spin @xmath85-manifold @xmath12 is related to the spin cobordism group @xmath99 .
if @xmath100 is a globally well defined lie algebra valued one form then we have the following variational formula for the spin chern - simons functional : @xmath101 for the abelian group @xmath0 it happens that both cobordism groups @xmath102 and @xmath103 vanish .
the spin chern - simons theory is constructed as follows .
consider an oriented compact 3-dimensional manifold @xmath9 without boundary .
let @xmath104 be oriented principal @xmath105 bundles corresponding to the @xmath106-th @xmath8 group in @xmath107 .
the lie algebra of @xmath105 contains one element @xmath108 .
we denote by @xmath109 a connection on @xmath110 . in the abelian case we find it convenient to rescale the connection @xmath111 by @xmath112 so that its curvature @xmath113 has integer periods .
we henceforth use this convention . to construct an arbitrary real @xmath107 bundle @xmath82
one has to choose a real representation @xmath79 of the group @xmath107 .
any real irreducible representation of @xmath107 is two dimensional .
we denote @xmath114 the following matrix @xmath115 then an arbitrary real irreducible representation is of the form @xmath116 where @xmath117 and @xmath118 parameterizes the angle in the @xmath106-th @xmath8 group .
any virtual real representation @xmath79 of @xmath107 can be written as @xmath119 where @xmath120 is the multiplicity of the irreducible representation @xmath116 and we assume @xmath121 for all but finitely many @xmath122 .
usually multiplicities @xmath123 are positive integers , however here we consider virtual representations and thus multiplicities can also be negative .
it is easy to see that the curvature is @xmath124 using one finds that the spin chern - simons functional is @xmath125 this expression coincides with which was obtained in introduction using simpler arguments .
it is interesting to notice that an arbitrary representation of @xmath2 defines a spin chern - simons action but not an integral chern - simons action .
an integral chern - simons action is defined only by very special representations @xmath126 : they are generated by @xmath127 .
for these types of representations we have @xmath128 = \sum_v m_v v_{\alpha}(v_{\alpha}+1)\,[f^{\alpha}\wedge f^{\alpha } ] + 2\sum_{\alpha<\beta}\sum_v m_v v_{\alpha}v_{\beta}\,[f^{\alpha}\wedge f^{\beta } ] \in h^4(bt,{\mathbb z}).\ ] ] the corresponding matrix @xmath25 has even integers on the diagonal , and thus it defines an integral chern - simons action ( see introduction ) .
the action depends on the choice of spin structure .
the conceptual explanation for this is given in the next subsection .
in the thesis of @xcite it is shown that for general compact group @xmath3 spin chern - simons actions are associated with elements of @xmath34 , which fits in @xmath129 for the group @xmath0 we choose a basis @xmath130 in @xmath21 and write an element @xmath131 where @xmath132 is a matrix defined in .
the projection @xmath133 onto @xmath21 is just the mod @xmath97 reduction of the element of @xmath21 : @xmath134 where @xmath135 are the stiefel - whitney classes of @xmath110 .
points in the @xmath136 component of the picard group .
there is no natural origin of @xmath137 . ]
the space @xmath138 of the spin structures of @xmath139 is a principal homogenous space for the group of translations @xmath140 ( see figure [ fig : picard ] ) .
the group @xmath140 is in one - to - one correspondence with flat real line bundles on @xmath139 .
if @xmath141 is the spinor bundle corresponding to the spin structure @xmath142 and @xmath143 then the bundle @xmath144 can be identified with @xmath145 where @xmath146 is a line bundle with the flat connection @xmath147 . to get the spin structure dependence of we use the close relation between chern - simons and the @xmath148-invariant of a dirac operator
( see next subsection ) .
the dirac operator associated to @xmath144 is the dirac operator @xmath149 twisted by @xmath147 .
thus to change the spin structure we have to shift @xmath150 .
the variational formula yields @xmath151 where @xmath152 is a solution of the equation @xmath153 equation does not depend on a particular choice of solution . from it
is easy to see that the equation above can be rewritten in the form @xmath154 clearly the matrix @xmath29 depends on a choice of basis in @xmath21 .
however two different matrices define the same classical action if they are related by a conjugation by an element of @xmath155 .
it is therefore convenient to describe the matrix @xmath29 in a more invariant language .
recall that any integral matrix defines an integral lattice which is a free @xmath156 module @xmath157 with the integral valued bilinear form given by the matrix @xmath33 .
we denote the dual basis in @xmath42 by @xmath158 . in this notation
the action takes the form @xmath159 \label{csphase}\ ] ] where @xmath88 the curvature of the extended connection @xmath89 . in this notation
the curvature @xmath160 takes values in @xmath161 , the space of two forms with periods in @xmath31 .
the vector @xmath162 appearing in is a representative of the characteristic class @xmath60\in\lambda^*/2\lambda^*$ ] defined by the condition that @xmath163 for all @xmath61 .
this condition is nothing else but the basis independent form of equation .
it is easy to see that the characteristic class @xmath60 $ ] is uniquely determined by the bilinear form on the lattice .
if @xmath164 is a globally well defined @xmath16-form then we have the following variational formula @xmath165 . \label{variat}\ ] ] if @xmath166 is the spin chern - simons functional corresponding to the spin structure @xmath142 and @xmath167 then the chern - simons functional corresponding to the spin structure @xmath168 is when we wish to emphasize the boundary manifold @xmath139 and @xmath169 when @xmath139 is understood and we want to emphasize the dependence on spin structure @xmath142 . ] @xmath170 [ vf ] here @xmath59 is _ any _ representative of the characteristic class @xmath60 $ ] . in @xcite jenquin
applied the results of dai and freed @xcite on the @xmath171-invariant of the dirac operator to study spin chern - simons theory .
the spin chern - simons action and the @xmath171 invariant are sections of the same line bundle over the space of connections , and hence they can be identified up to multiplication by a metric dependent function @xmath172 : @xmath173 the dirac operator @xmath174 on a compact @xmath37-dimensional manifold @xmath9 is self - adjoint and elliptic .
its spectrum is real and discrete so one can define @xmath175 where the sum ranges over the nonzero spectrum of @xmath174 .
this series converges for @xmath176 and defines meromorphic function which is regular at @xmath177 .
the value @xmath178 is called the @xmath148-invariant of the dirac operator @xmath174 .
the @xmath171 invariant is @xmath179 on a general odd dimensional manifold @xmath180 is a smooth function of geometric parameters such as the connection and metric .
however in @xmath37 dimensions ( and more generally in @xmath181 dimensions ) it happens that @xmath182 is also a smooth function of the geometric parameters .
the variation of the @xmath171-invariant under the change of smooth parameters follows from the aps index theorem for families .
consider a path @xmath183 and @xmath184 , @xmath185 $ ] , in the space of connections and metrics respectively then @xmath186 } \left[\hat{a}\bigl(\frac{\mathcal{r}(t)}{2\pi}\bigr)\wedge \mathrm{ch}(f_{a(t)})\right]_{(4 ) } \\
= -\frac{1}{8\pi^2}\int_{x_3\times [ 0,1 ] } { \mathop{\mathrm{tr}}\nolimits}_{\rho}(f_{a(t)}\wedge f_{a(t)})+\mathrm{rank}(e)\int_{x_3\times [ 0,1]}\hat{a } \bigl(\frac{\mathcal{r}(t)}{2\pi}\bigr)\end{gathered}\ ] ] where @xmath187 is the roof - genus , @xmath188 is the riemann tensor corresponding to the metric @xmath184 . comparing this equation with one
sees that the @xmath171 invariant and the spin chern - simons action are sections of the same line bundle over the space of connections .
consider maxwell - chern - simons theory with gauge group @xmath2 in three dimensions .
let @xmath110 be principal @xmath8 bundles corresponding to the @xmath106-th gauge group in @xmath2 .
we denote by @xmath190 ( @xmath191 ) the gauge potential ( @xmath192-valued ) on @xmath110 .
the action function @xmath193 of our theory is @xmath194e^{2\pi i { \mathop{\textsl{cs}\,}\nolimits}_{x_3}^{\text{spin}}(a^{\alpha } ) } \mathrm{hol}_{a^{\alpha}}(\{c_r , n^{r}_{\alpha}\ } ) .
\label{action}\ ] ] @xmath195 , @xmath196 is a symmetric positive definite matrix with @xmath197 .
it defines the relative couplings of @xmath8 gauge potentials .
@xmath198 where @xmath199 is the number of vortices , @xmath200 is a set of contours @xmath201 ( vortices ) together with representation @xmath202 of @xmath2 .
@xmath203 is the holonomy of the gauge field @xmath78 around the contours @xmath201 in the representation @xmath204 .
when @xmath190 is globally well defined we can write the last term in as @xmath205 .
@xmath206 is the spin chern - simons term which is defined by .
it is characterized by an integral ( generically odd ) lattice @xmath31 .
we choose a basis @xmath207 in @xmath31 and denote the bilinear form @xmath208 in this basis by @xmath29 .
the dual basis in @xmath42 we denote by @xmath158 .
we can think of the integers
@xmath202 appeared in as defining elements @xmath209 in @xmath42 .
it is convenient to introduce an auxiliary vector space @xmath210 in which the lattice @xmath31 is naturally embedded .
if @xmath164 is a globally well defined @xmath16-form with values in @xmath211 then we have variational formulas and @xmath212 the characteristic property of the holonomy is that if the contour @xmath201 is a boundary of the disk @xmath213 then @xmath214 using the variational formulas one obtains the following equations of motion : @xmath215 where @xmath216 is a @xmath97-form supported on the curve @xmath201 . from this equation
it follows that @xmath217 where @xmath218 .
this equation determines the first chern classes @xmath219 in the presence of vortices ( see details in the beginning of section [ sec : ham : ham ] ) .
this simple arithmetic equation has solutions iff @xmath220 defines an element of @xmath31 . in the fqhe
@xmath221 has the interpretation of the total number of electrons in the @xmath106-th landau level .
let @xmath222 where @xmath223 is a compact oriented riemann surface of genus @xmath67 .
the metric on @xmath9 is a product metric : @xmath224 in this background we can write @xmath225 .
further we will assume that @xmath201 is a straight line in the time direction passing through the point @xmath226 . using @xmath227 one can rewrite the classical equations of motion as @xmath228 [ cleoms ] the space of the gauge potentials @xmath229 is a disjoint union of affine spaces @xmath230 labeled by the first chern classes @xmath130 .
the translation group of the affine space @xmath230 is @xmath231 . in the presence of vortices the gauge potentials @xmath190 are connections on nontrivial line bundles whose first chern classes @xmath232 satisfy equation .
in other words @xmath233 . to parameterize the affine space @xmath234 we choose a reference potential @xmath235 so an arbitrary gauge potential @xmath236 where @xmath164 is a globally well defined @xmath16-form on @xmath9 with values in @xmath211 .
it is convenient to choose the reference potential @xmath235 such that it satisfies equations with _ only one vortex _ of net charge @xmath237 located on the contour @xmath238 .
the gauge potential @xmath233 can be written as @xmath239 where @xmath240 and @xmath241 are respectively a globally well defined @xmath16-form and function on @xmath242 with values in @xmath211 .
one obtains the following expression for the lagrangian : @xmath243 + 2\pi\sum_{r=1}^{n_v } n^{r}_{\alpha}\bigl[a_0^{\alpha}(x_r ) -a_0^{\alpha}(x_{\bullet})\bigr].\end{gathered}\ ] ] here @xmath244 is the lagrangian corresponding to the reference connection @xmath245 .
the corresponding action has exactly the form with @xmath78 changed to @xmath246 .
the momentum is defined as @xmath247 , thus @xmath248 the last equation shows that @xmath249 is a globally well defined gauge invariant @xmath16-form .
@xmath249 has interpretation of the electric field .
the symplectic form is @xmath250=i\phi_j^{\beta}(\sigma ) .
\label{comrel}\ ] ] the hamiltonian is defined by the legendre transform : @xmath251\ ] ] where @xmath252 and @xmath253 it is important for rest of the paper to notice that @xmath254 has a natural khler structure .
we will denote by @xmath255 the subspace of one forms in @xmath256 whose periods lie in @xmath31 .
the symplectic structure is defined by the bilinear form @xmath29 @xmath257 notice that this symplectic form takes integral values on @xmath258 .
the matrix of couplings @xmath196 defines a positive definite metric on @xmath256 : @xmath259 we will show that the data of @xmath260 and @xmath261 determine a complex structure @xmath262 on @xmath256 .
this complex structure is related to the hodge complex structure by a matrix @xmath263 : @xmath264 @xmath263 is a @xmath132-symmetric matrix ( @xmath265 ) and @xmath266 . in order to extract the matrix @xmath263 from the metric @xmath261 we first construct the operator @xmath267 satisfying @xmath268 this yields @xmath269 . since @xmath270 is @xmath132-symmetric and @xmath132 is a symmetric matrix it follows that the matrix @xmath271 is positive definite .
thus there exists a positive definite square root @xmath272 .
now the matrix @xmath263 is defined by appearing here is exactly the same matrix @xmath263 which was constructed in @xcite . ]
@xmath273 notice that if @xmath132 is a positive definite matrix then @xmath274 .
in general @xmath132 has indefinite signature and we have the following statement : @xmath275 and @xmath276 . from the construction of @xmath263 it follows that @xmath277 .
finally the configuration space @xmath256 is actually a khler vector space with the symplectic structure and compatible complex structure .
the metric on @xmath256 which is canonically associated to the complex structure @xmath262 is @xmath278 where @xmath279 .
in the following two subsections we derive the quantum gauss law . in subsection
[ subsec : naive ] we derive the gauss law for the small gauge transformations using the hamiltonian formalism . in subsection
[ subsec : qgl ] we derive the quantum gauss law using some geometrical facts about ( spin ) chern - simons theory .
the classical gauss law is @xmath280 + 2\pi\sum_{r=1}^{n_v } n^{r}_{\alpha}\bigl[\delta^{(2)}(x - x_r ) -\delta^{(2)}(x - x_{\bullet})\bigr ] = 0 .
\label{smallgl}\ ] ] the _ small _ gauge transformations @xmath281 , where @xmath282 are globally well defined , are generated by this gauss law @xmath283 } e^{i\pi \int_{\sigma_g}(df,\,\bar{a } ) } \psi(a+df ) .
\label{ttaction}\ ] ] one can try to generalize the above equation for the _ large _ gauge transformations by changing @xmath284 and @xmath285 where @xmath286 is a closed @xmath16-form with periods in the lattice @xmath31 .
however this naive procedure fails to give a consistent gauss law .
this happens because it ignores the key geometrical fact that the wave function is a section of a certain line bundle with connection which has a nonzero curvature .
we explain this in detail in the next subsection .
[ [ the - magnetic - translation - group . ] ] the magnetic translation group .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + now let us look at the gauss law from a different point of view .
first one sees that the sum of the delta functions in is an exact form .
indeed , we can choose a set of non - intersecting contours @xmath287 which connect point @xmath288 with @xmath289 and write @xmath290 where @xmath291 is a @xmath16-form supported on the contour @xmath292 .
thus the quantity @xmath293 is a closed form .
moreover from the equation of motion it follows that @xmath294 where @xmath295 is a certain function .
thus for any closed time - independent @xmath16-form @xmath296 with values in @xmath211 we can construct a conserved charge @xmath297 although it is not indicated in the notation the charge @xmath298 depends on the choice of contours @xmath287 . in section [ subsec : qgl ]
we will present specific choice of the contours . the charges for different @xmath16-forms @xmath299 and @xmath300 do not commute : @xmath301 = \frac{i}{2\pi}\,\omega_k(\phi_1,\phi_2)\ ] ] where @xmath260 is the symplectic form defined in . moreover
the charge @xmath298 is not gauge invariant . to overcome these difficulties we have to choose a polarization on @xmath256 .
so , choose a `` canonical '' homology basis @xmath302 on @xmath242 , @xmath303 .
it defines a dual basis @xmath304 in the cohomology . then we can write @xmath305 where @xmath306 ( @xmath307 ) is a part of @xmath296 which contains only @xmath308 ( @xmath309 ) .
if @xmath296 has periods in @xmath42 then the following operators are gauge invariant : @xmath310 these operators satisfy the following commutation relations : @xmath311 notice that if @xmath312 then @xmath313 is a gauge transformation
. thus we can identify @xmath296 with @xmath314 where @xmath171 is any vector in @xmath315 . with this identification @xmath316 and
@xmath317 generate the finite heisenberg group @xmath318 : @xmath319 this group is also known as the magnetic translation group . the hilbert space of the theory should be a representation of this group ; we confirm this below .
the classical gauss law fixes the first chern class of the connection on @xmath320 ( see eq .
[ firstchern ] ) .
the space of allowed connections is then an affine space modelled on @xmath254 .
so , in the presence of vortices the configuration space is @xmath321 here the second factor describes the position of the vortices . when evaluated on a manifold with boundary
the spin chern - simons action is a section of a line bundle @xmath322 over @xmath256 .
it was explained at the end of section [ sec : acs ] that this line bundle depends on a choice of spin structure @xmath142 .
thus the quantum wave function is a section of @xmath323 where @xmath324 is the line bundle associated to the principal bundle @xmath320 in which the group @xmath2 acts in the representation @xmath325 .
the line bundle @xmath326 has a natural connection defined by the spin chern - simons phase together with the holonomies .
consider a path @xmath327 in the space of connections @xmath328 and @xmath242 , @xmath185 $ ] is the coordinate on the path .
then the parallel transport in @xmath326 is defined by : @xmath329}(a^{\alpha}(t ) ) } \,\mathrm{hol}_{a^{\alpha}(t)}(\{x_r(t),n^{r}\})\in \mathrm{hom } \bigl(\mathcal{l}_{\sigma}\bigl|_{\{a^{\alpha}(0),x_r(0)\ } } , \mathcal{l}_{\sigma}\bigl|_{\{a^{\alpha}(1),x_r(1)\}}\bigr ) .
\label{connection}\ ] ] has non - zero curvature , and thus the parallel transport does not define the lift of the gauge group . in figure
b ) we present the gauge invariant configuration space @xmath256 .
the parallel transport along straight line path on figure a ) corresponds to the holonomy around the closed loop in @xmath330 . ]
the tangent vector to the path @xmath331 is a pair @xmath332 where @xmath333 and @xmath334 is a tangent vector at the point @xmath335 .
the curvature of the connection is @xmath336 now , for any 1-form @xmath337 introduce the straightline path in the space of connections @xmath328 @xmath338 .
\label{path}\ ] ] here we do not move the position of the vortices . using the formula for the curvature we find @xmath339 it follows from that parallel transport _ does not _ define a lift of the gauge group to the total space of @xmath326 ( see figure [ fig : tor ] ) . to define the lift of the group action we choose a standard path , say , and define the action on a section @xmath340 of @xmath326 by @xmath341 where @xmath342 is a phase , and @xmath286 is an element of @xmath258 ( a closed one form with periods in @xmath31 ) .
the `` lifting phase '' @xmath342 must satisfy @xmath343 this subsection is a little technical .
first , we construct the canonical cocycle @xmath344 , see equation .
however it happens that this is not the cocycle we need to construct the lift of the gauge group .
other cocycles differ from the canonical one by the local terms .
second , we fix this local term . the final result is summarized by equations and . [
[ construction . ] ] construction .
+ + + + + + + + + + + + + the cocycle can be constructed using the phase @xcite . to construct the cocycle we proceed as in @xcite .
consider the following three dimensional manifold @xmath345 where we choose the antiperiodic spin structure on the circle .
( only the antiperiodic spin structure on the circle can be extended to a spin structure on the disk ) .
construct the twisted bundle @xmath346 over @xmath345 ( see figure [ fig : cocycle]a ) : @xmath347/\{(p,0)\sim ( g\cdot p,1)\}\ ] ] where @xmath67 corresponds to a gauge transformation @xmath286 , and @xmath94 was defined in the beginning of section [ sec : spincs ] .
the connection satisfies @xmath348 .
the cocycle is the spin chern - simons phase on this @xmath37-fold times holonomies @xmath349 one sees that the connection @xmath350 does not have indices along the @xmath351 direction which means that the term with the holonomy is a locally constant function of @xmath78 .
@xmath352 defined in this way satisfies the cocycle relation . over @xmath345
the twist is defined by identification of @xmath353 with @xmath354 by the gauge transformation @xmath67 corresponding to @xmath286 . in figure
b ) we present the surface @xmath355 ( a pair of pants ) which is used to prove the cocycle relation . ]
the proof is a standard cobordism argument .
we consider a connection on the @xmath37-manifold @xmath356 restricting to @xmath357 on the three components ( @xmath358 means change of orientation of manifold @xmath139 ) .
then we choose the extending spin @xmath85-manifold to be @xmath359 where @xmath355 is a pair of pants bounding the three circles with spin structure restricting to @xmath360 on the three components ( see figure [ fig : cocycle]b ) . to be explicit we can choose @xmath355 to be the simplex @xmath361 with identifications @xmath362 .
now we extend the connection to be @xmath363 which clearly restricts to the required connection on the boundary .
the field strength is @xmath364 . we can therefore use to show that the product of phases around the boundary is @xmath365 this proves the desired cocycle relation .
the dependence of the cocycle on @xmath78 and the position of the vortices @xmath289 can be extracted from the variational formulas , and : @xmath366 [ var - varphi ] where @xmath100 is an arbitrary element of @xmath231 and @xmath367 is an arbitrary set of nonintersecting contours starting at points @xmath289 .
we expect that there are @xmath368 different cocycles corresponding to @xmath368 different spin structures ( see figure [ fig : picard ] ) .
the group @xmath369 acts transitively on the space of spin structures .
let @xmath370 then from equation it follows that the cocycles corresponding to spin structures @xmath142 and @xmath168 are related by @xmath371 where @xmath59 is _ any _ representative of the characteristic class @xmath60\in \lambda^*/2\lambda^*$ ] which was defined in section 2.3 .
recall that @xmath372 and thus we have @xmath373 [ [ dependence - on - the - base - point . ] ] dependence on the base point .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + the cocycle is not unique . the are other cocycles which differ by local terms . in section [ sec : ham : ham ] we chose a very particular coordinate system in the space of connections and vortices .
using the variational formulas we can the express the cocycle @xmath374 through its value at the base point @xmath375 : @xmath376\ , e^{2\pi i\omega_k(\omega , a)}\ , \varphi_{\sigma}(a_{\bullet},\omega ; x_{\bullet } ) \label{varphi2}\ ] ] where @xmath287 is a set of nonintersecting contours connecting @xmath288 with @xmath289 .
a particular choice of these contours will be explained in the beginning of the next section .
notice however that the cocycle does not depend on the homotopy class of the contours .
the cocycle at the base point satisfies @xmath377 from this equation it follows that @xmath378 is _ linear _ on the space of closed @xmath16-forms with periods in @xmath379 : @xmath380 this equation determines @xmath381 modulo @xmath382 . on the other hand using we can explicitly evaluate the cocycle for a small gauge transformation : @xmath383 recall that @xmath384 satisfies the equation of motion and thus the combination appearing in the exponential is an exact form : @xmath385 the consistency of with requires that @xmath386 . thus @xmath387 is just another trivialization of the left hand side of .
recall that @xmath387 is determined by equation only modulo @xmath382 . to define the cocycle
we must fix this ambiguity modulo @xmath388 .
we claim that there is a distinguished choice of @xmath389 . to prove this we choose a solution @xmath387 of @xmath390
. then an arbitrary solution is of the form @xmath391 where @xmath392 .
now we introduce the cocycle @xmath393 first , notice that the function @xmath394 descends to @xmath395 and satisfies the cocycle law .
second , note since @xmath396 it follows that @xmath397 .
now a choice of a canonical basis @xmath398 in @xmath399 determines a preferred spin structure @xmath400 ( see above equation ) . in the coordinates ,
@xmath401 , the cocycle @xmath402 is necessarily of the form @xmath403 \bigr\}\ ] ] where @xmath404 .
now the distinguished choice of @xmath148 is determined by setting @xmath405 , so @xmath406 .
having done this the variational formula determines @xmath407 \bigr\ } \label{fbullet : coord}\ ] ] using the affine transformation of the characteristics @xmath408 one can show that this choice is in fact @xmath409 invariant . hence @xmath389 determined in this way is independent of choice of basis in @xmath399 .
[ [ summary . ] ] summary .
+ + + + + + + + the cocycle @xmath410 which we use to define the lift of the gauge group is @xmath411 where @xmath394 is defined by .
it is a @xmath412-valued function on @xmath395 : @xmath413 , @xmath414 and @xmath397 .
the dependence on the spin structure is summarized by the following variational formula @xmath415 to define the action of the gauge group on the wave functions , we must first trivialize the line bundle @xmath326 .
we have already trivialized the space of connections @xmath234 by choosing a reference connection @xmath246 in section [ sec : ham : ham ] .
we must also trivialize the space of the vortex configurations on @xmath242 . to this end
we endow @xmath242 with marking @xmath416 , i.e. we choose a point @xmath295 and a `` canonical '' basis of one cycles @xmath417 , @xmath303 ( see figure [ fig : cuts]a ) .
next we cut the riemann surface along these cycles and obtain a @xmath418-gon @xmath419 presented in figure [ fig : cuts]b .
then the vortices @xmath226 can be canonically connected to @xmath288 by a set of straight line contours @xmath292 ( see section [ sec : ham : ham ] for definition of @xmath288 ) . and
@xmath44 cycles to cut the riemann surface @xmath242 to obtain a connected @xmath418-gon .
the straight white lines represent the canonical paths which were used to trivialize the line bundle @xmath420 the space @xmath421 is contractible and thus the line bundle @xmath326 over @xmath422 is trivial .
using first the parallel transport of the vortices along the straight line pathes @xmath287 , and then the parallel transport of the gauge field along the path keeping the vortices fixed we define a canonical nowhere vanishing section @xmath423 of @xmath326 .
more precisely , @xmath424 the _ ratio _ @xmath425 is a function , rather than a section ( to simplify the notations we will write @xmath426 instead of @xmath427 in the argument of the wave function ) .
the action of the gauge group on this function is @xmath428 where the cocycle @xmath429 is defined by .
the gauss law @xmath430 for the wave function takes the form @xmath431 since for the small gauge transformations @xmath413 this gauss law for @xmath432 agrees with the gauss law .
this fact partially explains the statement that the wave function is a section of @xmath326 .
from the geometrical point of view the gauss law says that the gauge invariant wave function is a section of a line bundle @xmath326 over @xmath433 .
the symplectic form @xmath260 descends to a nontrivial element of @xmath434 . from it
follows that the first chern class of @xmath326 along @xmath256 is @xmath435 $ ] .
using the hodge decomposition we can write @xmath240 as @xmath436 where @xmath437 , @xmath438 is the projection on the image of @xmath439 and @xmath440 is the projection on the image of @xmath441 .
this is an orthogonal decomposition with respect to the hodge metric on @xmath331-forms .
the magnetic hamiltonian @xmath442 depends only on @xmath438 , while the electric hamiltonian @xmath443 is a sum of terms depending on @xmath444 , @xmath438 and @xmath440 .
thus the wave function @xmath340 factorizes as a product @xmath445 .
the massive sector has a unique groundstate , but the harmonic sector has many ground states , leading to a hilbert space of `` conformal blocks ''
. we will be mostly focussing on the harmonic sector .
recall that the configuration space @xmath256 has a natural complex structure @xmath446 defined by eq . .
@xmath446 consists of two operators : the hodge @xmath447 and matrix @xmath263 .
recall that @xmath266 . using @xmath447 we can decompose @xmath448 .
notice that @xmath449 are holomorphic forms with respect to the hodge complex structure which in the case of a riemann surface is opposite to the standard complex structure .
in other words @xmath450 corresponds to abelian differentials @xmath451 while @xmath452 to abelian antidifferentials @xmath453 . using @xmath263
we can construct two projection operators @xmath454 and decompose @xmath455 .
thus with respect to the complex structure @xmath446 we have the following decomposition of @xmath256 : @xmath456 \oplus\bigl[\omega^{1,0}(\sigma_g , v_-)\oplus\omega^{0,1}(\sigma_g , v_+)\bigr ] .
\label{cdecom}\ ] ] now we have to choose a basis in this decomposition . for the riemann surface
there is a natural way to choose a basis of harmonic @xmath16-forms .
we choose a `` canonical '' homology basis @xmath457 ( @xmath303 ) in @xmath458 .
the homology basis defines a dual basis of @xmath16-forms @xmath459 .
it is convenient to normalize a basis @xmath460 of abelian differentials such that @xmath461 has period one through the cycle @xmath462 and zero through the cycles @xmath463 , @xmath464 .
then the integrals over the @xmath44-cycles are fixed and define the period matrix : @xmath465 the subscript @xmath416 shows that the period matrix depends on a choice of the symplectic basis ( in some formulas below we will omit it ) .
decomposing into real and imaginary parts , @xmath466 is a symmetric matrix with positive definite imaginary part @xmath467 .
relations imply that @xmath468 .
since the hodge complex structure is opposite to the usual one we have the relations @xmath469 it is easy to see that an arbitrary harmonic 1-form with values in @xmath211 : @xmath470 decomposes under as @xmath471 where @xmath472 the index @xmath473 denotes the projection by @xmath474 . we can expand the gauge fields @xmath164 and momenta @xmath475 in the basis @xmath476 : @xmath477 the nonzero commutation relations are @xmath478=-i\delta^{p}{}_q\,\delta^{\beta}{}_\alpha \quad\text{and}\quad [ \pi_{\alpha}^{2\,p},\,a^{\beta}_{2\,q}]=-i\delta^{p}{}_q\,\delta^{\beta}{}_\alpha.\ ] ] certainly it is more convenient to use holomorphic coordinates , so we introduce @xmath479 and similarly for @xmath480 and @xmath481 .
this yields the following nonzero commutation relations : @xmath482=-i\delta_{p}{}^{q}\delta_{\alpha}{}^{\beta } \quad\text{and}\quad [ \pi^{\bar{\tau}}_{\alpha , p},\,a^{\beta , q}_{\bar{\tau}}]=-i\delta_{p}{}^{q}\delta_{\alpha}{}^{\beta}.\ ] ] the hamiltonian in the harmonic sector is @xmath483\ ] ] where @xmath484 we have the following commutation relations : @xmath485= \pi\,(\tau_2^{-1})_{pq}k_{\alpha\beta } ; \label{ccrel0 } \\ & [ \mathscr{h}_{\text{harm}},\tilde{\pi}^{\tau}_{p,\alpha}]= -2\pi e^2\,\tilde{\pi}^{\tau}_{p,\beta}(\lambda k)^{\beta}{}_{\alpha } \quad\text{and}\quad [ \mathscr{h}_{\text{harm}},\tilde{\pi}^{\bar{\tau}}_{p,\alpha}]= 2\pi e^2\,\tilde{\pi}^{\bar{\tau}}_{p,\beta}(\lambda k)^{\beta}{}_{\alpha } .
\label{ccrel1}\end{aligned}\ ] ] now recall that to take advantage of the khler structure on the configuration space @xmath256 we have to use the decomposition @xmath486 .
notice that the matrix @xmath132 is positive definite on @xmath487 and negative definite on @xmath488 .
thus from one concludes that @xmath489 and @xmath490 are creation operators while @xmath491 and @xmath492 are annihilation operators .
equation shows that the creation operators increase energy by @xmath493 and annihilation decrease by the same amount .
clearly the ground state function @xmath494 must be annihilated by the annihilation operators @xmath491 and @xmath492 .
this yields the following expression for @xmath494 : @xmath495 \phi(a_{\tau}^-,\,a_{\bar{\tau}}^+ ) .
\label{grstate}\ ] ] here @xmath496 is the metric which is canonically associated to the complex structure @xmath262 , and @xmath497 is an arbitrary holomorphic function of its variables .
the energy of this ground state is @xmath498 . comparing the expression above with the decomposition one
concludes that @xmath499 is an antiholomorphic function on the configuration space @xmath256 . in section [ sec : gauss ] we have shown that the physical wave function must satisfy the gauss law constraint .
this constraint together with the holomorphy condition is very restrictive and allows one to determine @xmath296 completely . in this section
we write a very explicit basis of wavefunctions .
the final answer is given in equation below .
[ [ functional - equation - on - phi . ] ] functional equation on @xmath296 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + substituting into the gauss law constraint one finds that @xmath296 must satisfy @xmath500 where @xmath286 is a harmonic element of the real space @xmath258 .
the function @xmath296 depends only on @xmath501 thus for this expression to be well defined the exponential must also depend only on @xmath501 .
indeed , the combination @xmath502 .
introduce the hermitian form @xmath503 on the complex vector space @xmath504 and a factor of automorphy @xmath505 : @xmath506 where @xmath507 is a map from @xmath258 to @xmath8 : @xmath508 in these terms equation acquires the form @xmath509 this is a very well known equation @xcite .
it determines global antiholomorphic sections of the line bundle @xmath510 over the _ complex _
torus @xmath511 . the first chern class of the bundle @xmath512 is @xmath435 $ ] and the cocycle @xmath513 determines the holonomies . the holomorphic sections of this line bundle can be expressed in terms of the theta functions .
a simple argument shows that there are @xmath514 different sections .
a more pedestrian method to obtain the wavefunctions involves writing an overcomplete set of functions and projecting on the gauge invariant ones . in this way one obtains a theta - series which admits a holomorphic factorization ( see appendix [ app : gs ] ) . [
[ explicit - expression - for - the - cocycle . ] ] explicit expression for the cocycle .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + after choosing a basis in @xmath256 , @xmath515 , we can write the cocycle @xmath513 explicitly ( see ) @xmath516(m_2^p ) -2\pi
i \bigl[\epsilon_2^p\otimes w_2+c_2^p\bigr](m^1_p ) \bigr\ } \label{chi}\ ] ] where @xmath400 is the preferred spin structure associated to the homology basis @xmath416 ( see explanation below ) .
the dependence on @xmath147 is obtained from the variational formula .
the terms @xmath517 and @xmath518 are elements of @xmath519 defined by @xmath520 [ cocycle1 ] notice that the cocycle depends only on the homotopy classes of contours @xmath521 $ ] and classes @xmath522,[\epsilon_2]\in(\frac12{\mathbb z}/{\mathbb z})^g$ ] , and @xmath60\in\lambda^*/2\lambda^*$ ] but not on any particular representative .
it is crucial to notice that the expression for the cocycle depends on the choice of the marking @xmath416 of the riemann surface .
let us recall some useful facts about spin structures on @xmath242 .
the component @xmath137 of the picard group consists of the bundles of degree @xmath136 .
the riemann vanishing theorem can be used to characterize the zeros of theta function .
the @xmath523 divisor is an analytic subset of the jacobian defined by equation @xmath524 .
it implies that for any homology basis @xmath416 there is a preferred spin bundle @xmath525 .
if @xmath295 is a point on @xmath242 then @xmath526 $ ] is point in the jacobian called the vector of riemann constants @xcite . here
@xmath527 is the abel map .
let @xmath528 be the characteristics of the theta function , then @xcite @xmath529 here @xmath530 is the number of holomorphic sections of the spin bundle @xmath531 ( @xmath146 is a line bundle with the flat connection determined by @xmath147 ) .
thus the left hand side is the mod 2 index @xcite .
the spin structure is called even or odd depending on whether the mod 2 index is @xmath532 or @xmath533 . [ [ a - basis - in - mathcalc01.-the - theta - function . ] ] a basis in @xmath504 . the theta function .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we choose a complex basis in @xmath256 : @xmath534 where @xmath535 is the basis in @xmath42 .
the complex period matrix @xmath107 is determined by the equation @xmath536 explicitly : @xmath537 an arbitrary harmonic element @xmath538 of the real vector space @xmath256 can be written in the complex basis as @xmath539 + c^{1,0}\cdot\bigl[(\mathbbmss{1}_g\otimes k)\lambda_2+\overline{t}\cdot \lambda^1\bigr ] .
\label{c.decomp}\ ] ] further we will denote @xmath540 by @xmath541 .
this expression shows that the complex vector space @xmath504 is isomorphic to @xmath542 .
the lattice @xmath543 is embedded into @xmath504 according to this decomposition , and is isomorphic to @xmath544 .
substituting the decomposition into one finds @xmath545 these data is already enough to solve the functional equation using the fourier transform .
the details can be found in @xcite , here we only present the result .
define @xmath546 here @xmath547 , @xmath548 , @xmath549 is a representative of the dual lattice , and @xmath550 are characteristics of the theta function .
this canonical theta function satisfies the functional equation : @xmath551 where @xmath552 [ vartheta_cocycle ] the equations above become almost obvious if one rewrites the canonical theta function in the form @xmath553 [ [ the - physical - wave - function . ] ] the physical wave function .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + for the period matrix the canonical theta function can be written in terms of the siegel - narain theta function : @xmath554 where @xmath555 , @xmath64 denotes the bilinear form on @xmath211 , @xmath556 , @xmath557 is a representative of the discriminant group @xmath36 .
@xmath122 and @xmath558 in are related to @xmath171 and @xmath559 by @xmath560 , @xmath561 and @xmath562 .
comparing with we can identify @xmath563 and @xmath564 and write @xmath565 here @xmath566 , and @xmath567 , @xmath568 are _ particular _ representatives of the classes @xmath522,\,[\epsilon_2]$ ] and @xmath60 $ ] , @xmath558 are defined by .
the subscript @xmath416 in the notation for the wave function shows that it depends on the choice of marking @xmath416 .
@xmath569 is some normalization `` constant '' which we are going to fix later .
the wave functions form a basis in the hilbert space of the ground states @xmath570 .
it is interesting to notice that the wave function in the harmonic sector depends on the coupling constants @xmath196 only through the matrix @xmath263 which determines the complex structure on @xmath256 .
the dependence of the wave function on a choice of representatives @xmath567 of the classes @xmath522,\,[\epsilon_2]$ ] is summarized by the following formula @xmath571 + i\pi ( w_2,\,n_2^pc^1_p - n^1_p c_2^p ) + 2\pi in_2^p(\gamma_p , w_2 ) } \ , \psi_{\gamma+n^1\otimes \overline{w}_2,\,\epsilon^1,\epsilon_2}^{\text{phys}}. \label{e - dep}\ ] ] here @xmath572 is the projection of @xmath59 onto the discriminant group @xmath36 .
the dependence of the wave function on a choice of representative @xmath59 of the characteristic class @xmath60\in \lambda^*/2\lambda^*$ ] is @xmath573 where @xmath355 is an arbitrary element of @xmath42 .
the generators of the magnetic translation group @xmath318 are of the form @xmath574 thus the magnetic translation group @xmath318 acts on the physical wave functions as follows : @xmath575 [ wrep ] where @xmath576 and @xmath577 .
one sees that the space of the physical wave functions @xmath578 is a representation space for the group @xmath318 . from the representation one
concludes that the magnetic translation group @xmath318 represents translations of the theta function by the elements from the discriminant group .
put differently , after fixing the complex structure @xmath262 we are left with finitely many holomorphically inequivalent choices of embedding of the lattice @xmath258 into the complex vector space @xmath504 .
these choices are parameterized by the discriminant group @xmath36 .
the magnetic translation group is a permutation group on the space of these embeddings . since the hamiltonian and hilbert space factorize into flat and massive sectors we can consider the wavefunction restricted to the flat fields .
the inner product on the space of flat fields is defined by @xmath579 where the integral runs over all closed @xmath16-forms on @xmath242 with values in @xmath211 .
the integral descends to one on the space of gauge inequivalent flat fields .
this space is `` the jacobian '' @xmath580 .
it is interesting to notice that if we substitute the expression into one obtains the natural inner product on the sections of the line bundle @xmath581 .
we use the hodge decomposition to write @xmath582 where @xmath583 is a harmonic @xmath16-form and @xmath584 is a function with values in @xmath211 .
the measure as usual can be obtained from the norm : @xmath585 where we used @xmath197 , and @xmath586 is the determinant of the laplacian operator @xmath587 on the space of functions , and @xmath588 means that we excluded the zero modes .
the integral over @xmath589 cancels the volume of the small gauge transformations .
the volume of the large gauge transformations is cancelled by restricting the integral over @xmath590 to integral over @xmath591 .
so after gauge fixing one obtains the following expression for the @xmath592-norm : @xmath593 now we substitute into .
notice that @xmath594 appears linearly in the exponential ( this follows from and ) , and the integral over @xmath594 yields two kronecker symbols @xmath595 and @xmath596 where @xmath597 are summation variables in the definition of the siegel - narain theta function : @xmath598 the sum over @xmath599 combines with the integral over @xmath600 to give an integral over @xmath192 .
this gaussian integral is easy to calculate and one finds : @xmath601 evidently , from equation @xmath569 is some kind of square root of the right hand side of .
we now observe that there is a very natural squareroot provided we view @xmath569 as a section of a line bundle rather than a function .
notice that the factor @xmath602 is the quillen norm of the section @xmath603 of the determinant line bundle @xmath604 over the space of complex structures on @xmath242 : @xmath605 where @xmath606 is the volume of the surface @xmath242 in the metric .
thus we can rewrite as @xmath607 here @xmath608 are the dimensions of @xmath609 .
thus @xmath610 and @xmath611 . to obtain a holomorphic splitting
we will consider @xmath569 as _ a section of the determinant bundle _ @xmath612 over the moduli space of complex structures @xmath613 .
the rationale for this choice is that we expect the dual conformal field theory to have @xmath614 left - moving and @xmath615 right - moving bosons .
the norm of the wave function is now defined as a product of the quillen norm for @xmath569 and the @xmath592-norm for the theta function .
the normalization condition for the wave function @xmath616 implies @xmath617 we can identify this expression with the partition function of noncompact left- and right- moving bosons .
we will interpret @xmath618 as an automorphic form on teichmller space .
interpreted as a function on teichmller space , @xmath569 has the following modular properties : @xmath619 and @xmath620 where @xmath621 is a symmetric @xmath622 matrix and @xmath623 is an integer .
we are being sloppy at this point .
the line bundle @xmath604 has curvature , and there is no canonical trivialization over teichmller space .
we expect ( given the results in @xcite ) that there is a canonical isomorphism of @xmath624 with the line in which @xmath523 is valued , allowing us to give a canonical definition of @xmath625 as a function of @xmath626 . for genus
@xmath16 we can be rather explicit .
any metric on the torus can be written as @xmath627 where @xmath628 and @xmath626 is a complex number with positive imaginary part .
the determinant of the @xmath629 operator is @xmath630 where @xmath631 is the liouville action and @xmath632 is the dedekind @xmath148-function .
the dedekind @xmath148-function has the following modular properties : @xmath633 and @xmath634 .
thus , @xmath635^{-n/4 } \,\frac{|\det k|^{1/4}}{\eta^{\,r_-}(\tau ) \bar{\eta}^{\,r_+}(\bar{\tau})}.\ ] ] the action of the teichmller modular group on the wavefunctions factors through the action of the symplectic modular group , defined by the action of large diffeomorphisms on @xmath399 .
this group is isomorphic to @xmath409 .
the group @xmath409 consist of the matrices of the form @xmath636 it is generated by 1 .
@xmath637 , @xmath638 i.e. @xmath639 ; 2 .
@xmath640 where @xmath621 is any symmetric integral @xmath622 matrix ; 3 .
@xmath641 .
the modular properties of the siegel - narain theta - function are as follows : @xmath642 a - transform : @xmath643 @xmath642 b - transform ( generalization of the t - transform ) : @xmath644 where @xmath59 is a representative of the characteristic class @xmath60\in \lambda^*/2\lambda^*$ ] .
the left hand side of clearly does not depend on a particular choice of @xmath645 $ ] , thus the right hand side also does not depend on this choice .
this freedom in the choice of a representative will be important later when we will consider b - transformation of the wave function .
@xmath642 @xmath193-transform : @xmath646 [ theta - mod ] before applying these modular transformations to the wave function we need to introduce some notation .
the functions are labeled by elements of the discriminant group @xmath40 .
we denote by @xmath41 the bilinear form on @xmath45 inherited from the bilinear form on @xmath42 . some quantities in the expressions below
are naturally written in terms of the functional @xmath647 : @xmath648 where @xmath64 is the bilinear form on @xmath42 . the fact that @xmath59 is a characteristic vector guarantees that @xmath647 descends to a functional on the discriminant group .
moreover the functional @xmath647 does not change under the shifts @xmath649 where @xmath650 is an arbitrary element of the lattice @xmath31 .
the constant term in was chosen in such a way that under the shifts of @xmath59 by @xmath651 , @xmath652 the functional satisfies the equality @xmath653 here @xmath654 denotes the projection of @xmath355 onto the discriminant group .
actually @xmath647 is a quadratic refinement of the bilinear form @xmath44 ( see equation for the definition ) .
it is important to understand that the quadratic refinement contains some extra information about the lattice which is not encoded in the bilinear form on @xmath45 . as we show below
, if @xmath44 is nondegenerate then every quadratic refinement of @xmath44 defines a distinguished element in @xmath45 . for the case at hand
it is @xmath572 the projection of @xmath59 onto the discriminant group .
using the modular properties of the theta function and the normalization section one can derive the modular transformations of the wave function : @xmath642 @xmath78-transform : @xmath655 @xmath642 @xmath621-transform : @xmath656}\ , \\
\times e^{-2\pi i\sum_{p < q } b^{pq}\,b(\gamma_p,\gamma_q)}\ , \psi_{\gamma,\,\epsilon^1,\epsilon_2-b\epsilon^1-\frac12\mathrm{diag}(b)}(a_1,a_2,\tau ;
c^1,c_2-bc^1 ) \label{pret}\end{gathered}\ ] ] where @xmath623 is some integer number which depends on the entries of @xmath657 .
this phase arises from the factor @xmath658 in @xmath569 , see equation , and follows simply from the net gravitational anomaly @xmath659 of @xmath614 left - moving and @xmath615 right - moving noncompact bosons .
@xmath642 @xmath193-transform : @xmath660 [ modular ] to obtain the actual @xmath107 and @xmath193 matrices one has to use for some values of @xmath661 .
evidently expression can be rewritten in terms of the quadratic refinement @xmath647 , : @xmath662 + i\pi ( w_2,\,n_2^pc^1_p - n^1_p c_2^p ) + 2\pi i n_2^p[q_{w_2}(-\gamma_p)-q_{w_2}(\gamma_p ) ] } \ , \psi_{\gamma+n^1\otimes \overline{w}_2,\,\epsilon^1,\epsilon_2}^{\text{phys}}. \label{e - dep - inv}\ ] ] notice that only the term with the vortices depends on @xmath59 not through the quadratic refinement . from now on we will not consider the vortices .
that is , we put @xmath663 .
the main result of this discussion is that the representation of the modular group ( without vortices ) is completely encoded in the discriminant group @xmath45 , the bilinear form @xmath44 on the discriminant group , the quadratic refinement @xmath647 and the signature @xmath664 of the bilinear form on the lattice @xmath31 modulo @xmath39 .
* lemma : * two representations of the modular group in the hilbert spaces @xmath665 and @xmath666 are isomorphic .
the isomorphism is completely determined by the discriminant group @xmath45 , the bilinear form @xmath44 and the quadratic refinement @xmath647 .
* proof : * here we consider only the representation of the modular group without the vortices , so @xmath667 .
the hilbert space @xmath665 admits the following decomposition : @xmath668 where @xmath669 is the hilbert space of wavefunctions with fixed spin structure .
we denote an operator @xmath670 acting from @xmath671 to @xmath672 by @xmath673 $ ] .
the wave functions form a basis in @xmath578 . from equation
we learn that the bases in the hilbert spaces @xmath674 and @xmath675 are related by the operator : @xmath676 notice that all factors except for the last one depend only on the projection @xmath654 of @xmath355 onto the discriminant group .
if we shift @xmath355 by an arbitrary element @xmath650 of the lattice @xmath31 then the corresponding operators @xmath677 are related by @xmath678^{(\lambda,\lambda ) } u(\delta,\epsilon^1,\epsilon_2).\ ] ] the expression in the square brackets is the mod 2 index , see .
it equals @xmath532 or @xmath533 depending on whether the spin structure @xmath679 is even or odd .
thus if the spin structure is even the operator depends only on the data appearing in the lemma .
however if the spin structure is odd we have an ambiguity in the multiplication by @xmath680 .
the operators acting in the hilbert spaces @xmath666 and @xmath665 are related by @xmath681 = u(\delta,\epsilon^{1},\epsilon_2)_{\gamma}{}^{\gamma_1 } { \mathcal{o}}(w_2)_{\gamma_1}{}^{\gamma_2}\ ! \left [ \begin{smallmatrix } 2\epsilon^{\prime\,1 } & 2\epsilon'_2 \\ 2\epsilon^1 & 2\epsilon_2 \end{smallmatrix } \right ] ( u^{-1})(\delta,\epsilon^{\prime\,1},\epsilon'_2)_{\gamma_2}{}^{\gamma'}.\ ] ] substituting one finds @xmath682 = { \mathcal{o}}(w_2)_{\gamma+2\epsilon^1\otimes\bar{\delta}}{}^{\gamma'+2\epsilon^{\prime\,1}\otimes\bar{\delta}}\ ! \left [ \begin{smallmatrix } 2\epsilon^{\prime\,1 } & 2\epsilon'_2 \\ 2\epsilon^1 & 2\epsilon_2 \end{smallmatrix } \right ] \\ \times e^{2\pi i\,b(\overline{\delta},2\epsilon_2\cdot\gamma-2\epsilon_2'\cdot\gamma ' ) } \,e^{2\pi i ( \delta,\delta)\,\frac12[4\epsilon^1\cdot\epsilon_2 -4\epsilon^{\prime\,1}\cdot\epsilon_2']}. \label{oorel}\end{gathered}\ ] ] again all one on the right hand side depend only on the projection @xmath654 of @xmath355 onto the discriminant group _ with one exception _ , namely the last term .
the expression above is true for any operator .
however the operators which form a representation of the modular group have a special property they preserve the mod 2 index .
thus if @xmath683 $ ] is a modular transformation then the integer @xmath684 is an _ even _ integer .
hence we can change @xmath685 in the last term of to @xmath686 .
so we prove that two representations of the modular group in @xmath666 and @xmath665 are isomorphic and the isomorphism can be completely written in terms of @xmath45 , @xmath44 and @xmath647 .
@xmath687 for genus one it is easy to write the modular matrices explicitly ( without vortices ) . as in the previous section
we denote an operator @xmath670 acting from @xmath688 to @xmath669 by @xmath689 $ ] .
the wave functions form a basis in @xmath672 . to extract @xmath107 and @xmath193 matrices one has to use for some values of @xmath690 .
the nonzero @xmath107-matrices are : @xmath691 & = t_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 0 & 0 \\ 0 & 1 \end{smallmatrix } \right]= e^{2\pi i\sigma/24 - 2\pi i\ , [ q_{w_2}(-\gamma)-q_{w_2}(0)]}\,\delta_{\gamma\gamma ' } ; \\ t_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 0 \\ 1 & 0 \end{smallmatrix } \right ] & = t_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 1 \\ 1 & 1 \end{smallmatrix } \right ] = e^{2\pi i\sigma/24 - 2\pi i\ , q_{w_2}(-\gamma)}\,\delta_{\gamma\gamma'};\end{aligned}\ ] ] [ t ] where @xmath664 is the signature of @xmath31 modulo @xmath39 .
the nonzero @xmath193-matrices are : @xmath692 & = s_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 0 & 1 \\ 1 & 0 \end{smallmatrix } \right ] = \\ s_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 0 \\ 0 & 1 \end{smallmatrix } \right ] & = |\lambda^*/\lambda|^{-1/2}e^{2\pi i\ , b(\gamma,\gamma'+\overline{w}_2 ) } ; \\
s_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 1 \\ 1 & 1 \end{smallmatrix } \right ] & = + 4\pi iq_{w_2}(0)}\end{aligned}\ ] ] [ s ] one can verify that the @xmath193 and @xmath107 operators presented above satisfy the equivalence relation . in the process of verification it is enough to use only the properties of the quadratic refinement and but not the explicit expression .
abstractly the modular group in genus one without vortices is generated by two transformations @xmath193 and @xmath107 subject to the relations @xmath693 let us check that and satisfy these properties .
explicit calculation shows that @xmath694 & = \delta_{\gamma+\gamma',0 } ; & \qquad\qquad ( s^2)_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 0 \\ 1 & 0 \end{smallmatrix } \right ] & = \delta_{\gamma+\gamma',0}\,e^{2\pi i\,b(\gamma,\overline{w}_2 ) } ; \\ ( s^2)_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 0 & 1 \\ 0 & 1 \end{smallmatrix } \right ] & = \delta_{\gamma+\gamma'+\overline{w}_2,0 } ; & \qquad\qquad ( s^2)_{\gamma}{}^{\gamma'}\ ! \left [ \begin{smallmatrix } 1 & 1 \\ 1 & 1 \end{smallmatrix } \right ] & = \delta_{\gamma+\gamma'+\overline{w}_2,0}\ , e^{8\pi i q_{w_2}(0)+2\pi i\ , b(\gamma,\overline{w}_2)}\end{aligned}\ ] ] where @xmath572 denotes the projection of @xmath59 onto the discriminants group . it
easy to verify that @xmath695 for the first three operators above .
however @xmath696 = 1 $ ] requires the following equality @xmath697 which is certainly true for the quadratic refinement .
straightforward calculation shows that the equation @xmath698 ( for all spin structures ) is equivalent to the following equality : @xmath699}=e^{2\pi i q_{w_2}(-\gamma_1-\gamma_3)}.\ ] ] after summation over @xmath700 we arrive at the equality : @xmath701 in virtue of the relation this sum does not depend on a particular choice of representative @xmath59 of the class @xmath60\in\lambda^*/2\lambda^*$ ] .
moreover it is easy to see that the absolute value of the sum in the left hand side is @xmath16 .
actually equation is one of the versions of the celebrated gauss - milgram sum formula @xcite .
the original gauss - milgram sum formula was proven for an even lattice for @xmath702 by milgram ( for a one dimensional lattice the formula was proven by gauss and is called the gauss sum formula ) . however the shortest and most elegant proof was given by j. milnor and d. husemoller @xcite ( appendix 4 ) .
the proof of the gauss - milgram formula for the quadratic refinement can be found in @xcite on page 70 , proposition 5.41 .
given an integral lattice @xmath31 we construct the following set of invariants : 1 .
@xmath703 . the signature of the quadratic form on @xmath31 mod @xmath39 .
2 . the discriminant group @xmath40 with the bilinear form @xmath41 inherited from the bilinear form on @xmath42 .
the equivalence class of the quadratic refinement @xmath43 $ ] of the bilinear form @xmath44 on @xmath45 .
the quadratic refinement is a map @xmath46 such that for any two elements @xmath704 and @xmath705 of @xmath45 @xmath706 we say that two quadratic refinements @xmath49 and @xmath50 are equivalent if there exists @xmath51 such that for all @xmath52 @xmath707 these invariants satisfy the gauss - milgram constraint : @xmath708 note that this constraint fixes the ambiguity of adding a constant to @xmath54 .
notice too that this constraint does not depend on a choice of representative @xmath709 of the equivalence class @xmath43 $ ] .
* theorem : * a representation of the modular group of a spin chern - simons theory defined by a lattice @xmath31 is completely encoded in the invariants 1,2,3 of the lattice @xmath31 subject to the constraint .
* proof : * the proof of this statement was already indicated in section [ sec : modular ] . here
we only have to show that every representative of the equivalence class of quadratic refinements @xmath43 $ ] is of the form @xmath647 , see equation .
once this is established , we can use equations and to extract the projective representation of the modular group .
all these expressions depend only on @xmath45 , @xmath44 and @xmath647 . from the lemma in section
[ sec : modular ] it follows that the representation of the modular group corresponding to different choices of @xmath710 $ ] are isomorphic .
thus the theorem follows .
@xmath687 now let us explain why any quadratic refinement subject to the constraint is necessarily of the form @xmath647 .
there is a theorem which states that the quadratic refinements @xmath711 with @xmath712 are in one - to - one correspondence with the projections @xmath572 of representatives @xmath59 of the characteristic class @xmath60 $ ] onto the discriminant group .
the idea of the proof is as follows .
suppose we are given a quadratic refinement @xmath713 .
then there exists a distinguished element of @xmath45 .
indeed , consider the map @xmath714
@xmath715 from it follows that @xmath716 is a linear map .
since this map is defined on @xmath45 and @xmath44 is nondegenerate it follows that there exists an element @xmath717 such that @xmath718 .
now for the converse : given a representative @xmath59 of the characteristic class @xmath60 $ ] we construct the quadratic refinement @xmath719 this functional is invariant under a shift of @xmath59 by @xmath720 where @xmath61 .
thus @xmath713 does not depend on a particular choice of the lift @xmath721 .
an arbitrary quadratic refinement is of the form @xmath722 where @xmath723 is a linear map and @xmath56 is a constant . using that @xmath44 is nondegenerate on @xmath45 we can represent @xmath724 for @xmath725 .
now it can be shown that the addition of a linear function @xmath726 corresponds to a shift @xmath727 where @xmath355 is an arbitrary lift of @xmath728 to @xmath42 .
this shift is just the change of a representative of the class @xmath60 $ ] .
thus the only real ambiguity is the constant @xmath56 .
it can be fixed from the gauss - milgram sum formula @xcite : @xmath729 we are looking for the quadratic refinement which satisfies the constraint , thus @xmath730 .
so as we expected @xmath731 .
* corollary : * two quantum spin chern - simons theories defined by the lattices @xmath732 and @xmath733 are equivalent ( have the same @xmath37-manifold invariants ) iff the invariants 1,2,3 for the lattices @xmath732 and @xmath733 are equal .
example 1 . as simple examples , the three invariants vanish for the even unimodular lattice @xmath734 .
this was the case studied in @xcite .
another example is the odd unimodular lattice @xmath735 where again the invariants 1,2,3 vanish .
thus the chern - simons theory determined by this lattice is trivial .
indeed , the conformal field theory dual is a single left - moving and right - moving weyl fermion .
the partition function on a riemann surface with fixed spin structure is clearly invariant under modular transformations preserving the spin structure .
example 2 .
the quantum chern - simons theories defined by the lattices @xmath31 and @xmath736 are equivalent . example 3 .
consider the @xmath737 lattice .
the @xmath737 lattice is an even unimodular lattice and thus its discriminant group is trivial , but it has signature @xmath738 which is nonzero mod @xmath39 .
so this theory is nontrivial .
example 4 .
any even unimodular lattice of signature @xmath739 is trivial .
this result raises the natural question of whether there is a converse result .
* theorem ( converse result ) : * every quartet : an integer @xmath142 modulo @xmath39 , a finite abelian group @xmath45 together with a bilinear form @xmath41 , and an equivalence class @xmath57 $ ] of a quadratic refinement of @xmath44 subject to the gauss - milgram constraint is realized by some lattice @xmath31 .
moreover if there exists a representative @xmath740 $ ] such that @xmath66 then the quartet is realized by an _ even _ lattice .
* corollary : * a ) the set of quantum _ spin _ chern - simons theories is in one - to - one correspondence with these quartets subject to the gauss - milgram constraint .
+ b ) the set of quantum _ integral _ chern - simons theories is in one - to - one correspondence with the quartets subject to the gauss - milgram constraint such that there exists a representative @xmath54 of the class @xmath57 $ ] with @xmath66 .
* proof : * suppose we are given the quartet @xmath65)$ ] .
first , from corollary 1.16.6 in @xcite it follows that for sufficiently large @xmath741 there exists an odd lattice of signature @xmath742 which produces the discriminant form @xmath44 .
thus we can always realize the triplet ( @xmath743 ) by some odd lattice .
now we can use the fact that given a lattice the quadratic refinement @xmath54 is necessarily of the form @xmath647 .
thus the theorem follows .
second , suppose that there exists @xmath740 $ ] such that @xmath66 . from corollary 1.10.2 in @xcite
it follows that for sufficiently large @xmath744 subject to the constraint @xmath745 there exists an _ even _ lattice of the signature @xmath742 which produces the quadratic refinement @xmath54 .
one of the ways to define the signature of the quadratic refinement ( for an even lattice ) ( see appendix 4 in @xcite ) is to use the gauss - milgram sum formula .
the @xmath142 on the right hand side is the signature of @xmath54 modulo 8 .
thus in our case the condition of the corollary 1.10.2 is automatically satisfied .
therefore the quartets for which there exists a representative @xmath746 $ ] such that @xmath66 are realized by _ even _ lattices .
example 5 . as
an example of the converse result consider the group @xmath747 .
there are two bilinear forms we can define on @xmath45 : @xmath748 and @xmath749 .
one can easily solve equation for the quadratic refinement .
one finds that for each bilinear form there is only one equivalence class of the quadratic refinements .
the representatives which satisfy the gauss - milgram constraint are @xmath750 where [ cols="^,^,^,^",options="header " , ] an odd lattice exists for any signature @xmath751 .
for example , the quartet @xmath752)$ ] is realized by the lattice @xmath753 with @xmath754 ; the quartet @xmath755)$ ] is realized by the lattice @xmath756 with @xmath757 .
even lattices usually do not exist for all signatures .
the quartet is realized by an even lattice if there is a representative @xmath54 of @xmath57 $ ] such that @xmath758 .
equations @xmath759 and @xmath760 for the quadratic refinement yield @xmath761 and @xmath762 respectively ( compare with table 15.4 in @xcite ) .
thus the quartets @xmath763)$ ] and @xmath764)$ ] are realized by even lattices .
for example , the second quartet is realized by the lattice @xmath756 with @xmath765 .
[ [ monoid - structure . ] ] monoid structure .
+ + + + + + + + + + + + + + + + + the space of quantum chern - simons theories has a structure of monoid .
the sum operation is defined by @xmath69 ) \oplus ( \sigma_2\mod 24,\,\mathscr{d}_2,\,b_2,\,[q_2 ] ) = ( \sigma_1+\sigma_2\mod 24,\,\mathscr{d}_1\oplus\mathscr{d}_2,\,b_{12},\,[q_{12}])\ ] ] where the bilinear form @xmath70 and the quadratic refinement @xmath71 are defined by @xmath72 evidently the quadratic refinement defined in this way satisfies the gauss - milgram constraint with the signature @xmath766 .
the identity element is the trivial quartet .
one can ask whether the space of quantum chern - simons theories has in addition an abelian group structure .
it is well know that the space of quadratic refinements ( with nondegenerate @xmath44 ) over a commutative ring with identity form a witt ring ( see appendix 1 in @xcite ) .
the sum operation is defined in the same way as above , while the subtraction operation is defined using the grothendieck construction .
the trivial elements are so - called split modules . for the case at hand
the ring is @xmath156 .
the quartet @xmath767)$ ] is called split if there exists a @xmath156-submodule @xmath27 of @xmath45 such that @xmath768 and @xmath769 for some @xmath51 . here
@xmath770 is a set of all @xmath771 such that @xmath772 for all @xmath773 .
notice that in this definition the nondegeneracy of @xmath44 is important .
notice also that the definition of the split quartet does not depend on a particular choice of quadratic refinement from @xmath57 $ ] .
two quartets @xmath774)$ ] @xmath775 are called equivalent if there exist two split quartets @xmath776)$ ] such that @xmath777 .
the zero element is the equivalence class of split quartets .
however these split quartets correspond to _ non trivial _ quantum chern - simons theories ( for example @xmath778 ) .
for this reason we can not impose this equivalence relation . hence the quantum chern - simons theories do not seem to have an abelian group structure .
this section is based on refs .
@xcite . a quantum hall fluid in the long - distance limit has an effective description in terms of _ spin _ chern - simons theory with maxwell kinetic term .
this is precisely the theory we consider in this paper .
the field strengths @xmath113 play the role of electric currents associated with the @xmath106-th landau level . in the long distance limit
these currents interact very weakly , namely , only through the chern - simons term .
let @xmath779 be the gauge potential of the external electro - magnetic field .
we assume that it is topologically trivial .
the currents couple to the external field in the standard way : @xmath780 where @xmath781 is a set integers .
the integers @xmath781 must be very special in order to describe couplings to the landau levels ( see below ) .
we are interested in the following partition function @xmath782}.\ ] ] the variation with respect to @xmath779 yields the expectation value of the electro - magnetic current @xmath783 in the system : @xmath784 this expectation value is easy to calculate . since @xmath229 is topologically trivial one can shift the integration variable @xmath785 : @xmath786 where @xmath787 .
thus the expectation value of the electric current is @xmath788 the hall conductivity is @xmath789 .
the total electric charge is given by the integral of this current over the surface .
the quantity @xmath790 is naturally associated to the electric charge of the vortex @xmath791 . notice that both @xmath792 and @xmath791 are elements of the dual lattice @xmath42 , thus generically the electric charge is fractional .
the statistics of the vortex @xmath791 is described by the product @xmath793 @xcite .
hence in general the vortices @xmath791 have fractional statistics .
however if @xmath791 is an element of the lattice then both its electric charge and statistics are integral .
certainly we want the excitations with an odd charge ( such as electrons ) to have fermi statistics , and the excitations with an even charge to have bose statistics . in other words
we must require that @xmath794 thus @xmath792 is actually a representative of the characteristic class @xmath60\in \lambda^*/2\lambda^*$ ] .
one of the invariants of spin chern - simons theory is the quadratic refinement @xmath709 of the bilinear form on the discriminant group @xmath36 . for a particular choice of basis it reads
@xmath795 one immediately observes that @xmath796 .
now recall that two quadratic refinements @xmath54 and @xmath797 are equivalent if there exists @xmath798 such that @xmath799 . for the case at hand
this means @xmath800 .
the set of all possible fractional hall conductivities associated with a given lattice @xmath31 is thus constructed by : @xmath801\ } = \{8q(\gamma)\mod 8\,|\,\gamma\in\lambda^*/\lambda\}.\ ] ] for the example 5 .
the basis vectors of the dual lattice are @xmath802 and @xmath803 .
the lattice @xmath31 is spanned by the vectors @xmath804 .
the red dots together with the origin represent the discriminant group @xmath78 .
the circle on the top represents the values of the quadratic refinement @xmath805 .
] example 5 .
consider a two dimensional lattice @xmath806 with the bilinear form @xmath807 ( see figure [ fig : lattice ] ) .
the dual lattice @xmath42 is generated by vectors @xmath808 which are defined by @xmath809 .
we identify the lattice @xmath810 as @xmath811 where @xmath812 .
the discriminant group is a cyclic group of dimension @xmath813 .
it is represented by the classes @xmath814,\quad
\gamma_1=[\omega^1],\quad \gamma_2=[2\omega^1],\quad \gamma_3=[3\omega^1],\quad \gamma_4=[4\omega^1].\ ] ] the characteristic vector @xmath59 can be chosen to be @xmath802 .
the values of the quadratic refinement and the corresponding fractional hall conductivities are @xmath815 the fractional electric charge @xmath816 : @xmath817
we would like to thank m. douglas , d. freed , l. ioffe , j. jenquin , p. landweber , s. lukic , v. nair , n. read , r. stong , l. takhtadjan , and e. witten for discussions and correspondence .
this work was supported in part by doe grant de - fg02 - 96er40949 , d.b .
was supported in part by rfbr grant 02 - 01 - 00695 .
here we want to solve the following functional equation @xmath818 where the cocycle @xmath819 is ( @xmath515 ) @xmath820 first we choose an over complete basis of antiholomorphic functions @xmath821 second , using we average this function over the large gauge transformation : @xmath822 we now follow a standard procedure and use poisson resummation to split this sum in a form so that we can extract the conformal blocks .
the details are presented below .
using one finds that the sum can be written as @xmath823 thus identifying the cocycle with one finds that the physical wave function is given by equation .
we want to express the gaussian sum @xmath828 in terms of the siegel - narain theta function . here
@xmath819 is a quadratic refinement of the form @xmath260 , i.e. @xmath829 in the basis introduced in section [ sec : basis ] the metric @xmath830 has the form ( @xmath515 ) @xmath831 thus the sum takes the form @xmath832 where @xmath833 and @xmath834 . using @xmath835 we do poisson resummation with respect to @xmath836 in : @xmath837 where @xmath838 .
representing @xmath839 , @xmath840 and @xmath841 we can rewrite this sum as where @xmath843_+ + \bigl[\mathbbmss{1}_g\otimes k^{-1}t\bigr]_- , & \quad p_{r}&=\bigl[m^1+\mathbbmss{1}_g\otimes k^{-1}t\bigr]_- + \bigl[\mathbbmss{1}_g\otimes k^{-1}t\bigr]_+ ; \\
\bar{\ell}_{\tau}&=\bar{\ell}_2-\tau \bar{\ell}^1 , & \qquad \bar{\ell}_{\bar{\tau}}&=\bar{\ell}_2-\bar{\tau } \bar{\ell}^1 .
\label{psipsi}\end{aligned}\ ] ] now @xmath844 can be uniquely written as @xmath845 .
we can associate the element @xmath846 of @xmath847 to @xmath848 .
so one sees that @xmath849 and @xmath850 form two independent lattices . using that @xmath851 and
@xmath852 one sees that these sums exactly correspond to the sum in the siegel - narain theta function . using relations @xmath853 and
@xmath266 one obtains that @xmath854 .
finally the sum splits as follows : @xmath855 where @xmath856 and @xmath857 .
99 e. witten , `` sl(2,z ) action on three - dimensional conformal field theories with abelian symmetry , '' arxiv : hep - th/0307041 .
s. gukov , e. martinec , g. moore and a. strominger , `` chern - simons gauge theory and the ads(3)/cft(2 ) correspondence , '' arxiv : hep - th/0403225 .
r. dijkgraaf and e. witten , `` topological gauge theories and group cohomology , '' commun .
phys . * 129 * , 393 ( 1990 ) .
d. s. freed , `` classical chern - simons theory .
part 1 , '' adv .
math . * 113 * , 237 ( 1995 ) [ arxiv : hep - th/9206021 ] ; `` classical chern - simons theory .
part 2 , '' http://www.ma.utexas.edu/users/dafr/cs2.pdf j. frohlich , u. m. studer and e. thiran , `` quantum theory of large systems of nonrelativistic matter , '' lectures given at nato advanced study institute : les houches summer school , session 62 , aug 1994 , arxiv : cond - mat/9508062 .
stipsicz , `` on the vanishing of the third spin cobordism group @xmath858 '' , zap .
nauchn . sem .
otdel . mat .
inst . steklov .
( pomi ) 267 ( 2000 ) , geom .
5 , 290302 , 332 ; translation in j. math .
( n. y. ) 113 ( 2003 ) , no .
6 , 898905 .
x. z. dai and d. s. freed , `` eta invariants and determinant lines , '' j. math .
phys . *
35 * , 5155 ( 1994 ) [ erratum - ibid . *
42 * , 2343 ( 2001 ) ] [ arxiv : hep - th/9405012 ] .
e. witten , `` five - brane effective action in m - theory , '' j. geom .
* 22 * , 103 ( 1997 ) [ arxiv : hep - th/9610234 ] .
e. diaconescu , d. s. freed , and g. moore `` the m - theory 3-form and e(8 ) gauge theory , '' arxiv : hep - th/0312069 .
d. mumford , _ tata lectures on theta i _ , birkhuser , 1983 . | we derive a simple classification of quantum spin chern - simons theories with gauge group @xmath0 .
while the classical chern - simons theories are classified by an integral lattice the quantum theories are classified differently .
two quantum theories are equivalent if they have the same invariants on 3-manifolds with spin structure , or equivalently if they lead to equivalent projective representations of the modular group .
we prove the quantum theory is completely determined by three invariants which can be constructed from the data in the classical action .
we comment on implications for the classification of fractional quantum hall fluids .
@xmath1 may 2005 |
the diquark concept has been used in a large number of studies , ranging from one - gluon exchange to lattice qcd calculations . for example , see refs .
@xcite .
recently , an interacting quark - diquark model has been introduced by santopinto in ref .
this is a nonrelativistic potential model , where baryons are described as two - body quark - diquark bound states and the relative motion between the quark and diquark constituents in terms of a relative coordinate @xmath0 .
the hamiltonian contains a coulomb - like plus linear confining interaction and an exchange one , depending on the spins and isospins of the quark and the diquark . in this article
, we will discuss the main features of the interacting quark - diquark model , as relativistically reformulated in refs .
@xcite within the point form formalism @xcite .
finally , we will summarize our results for the strange and nonstrange baryon spectra @xcite and baryon magnetic moments in the quark - diquark model @xcite .
we will also briefly discuss the formalism to calculate other baryon observables in the quark - diquark model , including open - flavor strong decays and baryon masses with self - energy corrections .
in quark - diquark models , baryons are assumed to be composed of a constituent quark and a constituent diquark @xcite .
up to an energy of 2 gev , the diquark can be described as two correlated quarks with no internal spatial excitations @xcite .
then , its color - spin - flavor wave function must be antisymmetric .
moreover , as we consider only light baryons , made up of @xmath1 , @xmath2 , @xmath3 quarks , the internal group is restricted to su@xmath4(6 ) .
if we denote spin by its value , flavor and color by the dimension of the representation , the quark has spin @xmath5 , flavor @xmath6 , and color @xmath7 .
the diquark must transform as @xmath8 under su@xmath9(3 ) , hadrons being color singlets .
then , one only has the symmetric su@xmath4(6 ) representation @xmath10(s ) , containing @xmath11 , @xmath12 ( the scalar diquark ) and @xmath13 , @xmath14 ( the axial - vector diquark ) @xcite .
the hamiltonian of the model is @xcite @xmath15 here , @xmath16 is a constant , @xmath17 and @xmath18 are the direct and exchange quark - diquark interactions , respectively , @xmath19 and @xmath20 the diquark and quark masses . the direct term , @xmath21 is the sum of a coulomb - like interaction with a cut off plus a linear confinement term .
the exchange interaction is given by @xcite , @xmath22 \end{array } \mbox { } , \label{eqn : vexch - nonstrange}\ ] ] where @xmath23 and @xmath24 are the spin and isospin operators . to calculate the strange baryon spectrum , eq .
( [ eqn : vexch - nonstrange ] ) has to be generalized in the form of a grsey - radicati interaction @xcite , @xmath25 \end{array } \mbox { } , \label{eqn : vexch - strange}\ ] ] where @xmath26 are su@xmath27(3 ) gell - mann matrices . in the nonstrange sector
, we also have a contact interaction @xcite , @xmath28 introduced to reproduce the @xmath29 mass splitting . in eqs .
( [ eq : vdir]-[eqn : vcont ] ) , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 are free parameters , fitted to the reproduction of the experimental masses @xcite .
our results , extracted from refs .
@xcite , are shown in figs .
[ fig : spectrum - nd]-[fig : spectrum - xi - omega ] .
these results can be compared to those of three quark models @xcite and other quark - diquark model calculations @xcite . as fig .
[ fig : spectrum - nd ] , but for @xmath39 and @xmath40 resonances .
figure taken from ref .
@xcite ; aps copyright . ] as fig .
[ fig : spectrum - nd ] , but for @xmath39 and @xmath40 resonances .
figure taken from ref .
@xcite ; aps copyright . ] + as fig .
[ fig : spectrum - nd ] , but for @xmath41 , @xmath42 and @xmath43 resonances .
figure taken from ref .
@xcite ; aps copyright . ] as fig .
[ fig : spectrum - nd ] , but for @xmath41 , @xmath42 and @xmath43 resonances .
figure taken from ref .
@xcite ; aps copyright . ] in ref .
@xcite , we improved the `` relativized '' model of ref .
@xcite by introducing a spin - isospin transition interaction , which induces a mixing between scalar and axial - vector diquark states .
we computed the nonstrange baryon spectrum within point form dynamics and used the resulting wave functions to calculate the magnetic moments of the proton and neutron ( see sec .
[ baryon magnetic moments ] ) .
the spin - isospin transition interaction , @xmath44 , was chosen as @xmath45 where @xmath46 and @xmath47 are free parameters and the matrix elements of the spin transition operator , @xmath48 , are defined as @xmath49 } \left| s_1 , m_{s_1 } \right\rangle \neq 0 \mbox { for } s_1 ' \neq s_1 \mbox { } , \ ] ] with @xmath50 and @xmath51 .
the matrix elements of the isospin transition operator , @xmath52 , are defined analogously . and @xmath53 nonstrange baryon resonances ( up to 2 gev ) and the experimental masses from pdg @xcite ( boxes ) .
figure taken from ref .
@xcite ; societ italiana di fisica / springer - verlag copyright.,width=302 ] the spin - isospin transition interaction of eq .
( [ eqn : vtr(r ) ] ) mixes quark - scalar diquark and quark - axial - vector diquark states , i.e. states with @xmath54 ( @xmath55 ) and @xmath56 ( @xmath57 ) , whose total spin ( isospin ) is @xmath58 ( @xmath59 ) .
thus , in this version of the model the nucleon state , as well as states such as the @xmath60 , the @xmath61 and the @xmath62 , contain both a @xmath54 and a @xmath56 component . the introduction of the interaction of eq .
( [ eqn : vtr(r ) ] ) determines an improvement in the overall quality of the reproduction of the experimental data ( considering only @xmath63 and @xmath53 resonances ) @xcite , with respect to that obtained with the previous version of this model @xcite .
in particular , the roper resonance , @xmath64 @xmath65 , is far better reproduced than before and the same holds for @xmath66 @xmath67 .
[ fig : spectrum3e4 ] .
below , we discuss the formalism to calculate other baryon observables in the interacting quark - diquark model of ref .
we stress the importance of having a model in which @xmath68 s wave functions contain both a @xmath69scalar diquark and a @xmath69axial - vector diquark component , which play a role analogous to that of @xmath70- and @xmath71-type components in the nucleon wave function .
we think that the presence of both components is necessary to obtain a good reproduction of baryon observables in a quark - diquark model , including e.g. baryon magnetic moments and open - flavor strong decays . as a starting point , for the magnetic dipole operator @xmath72 we consider a non - relativistic expression @xcite @xmath73 in order to use the permutational symmetry of the quarks 1 and 2 of the diquark , we re - write eq .
( [ eqn : mu - nr ] ) as @xcite : @xmath74 where @xmath75 is the charge operator of the diquark in the isospin space , @xmath76 represents its spin , @xmath77 is the difference between the charges of quarks 1 and 2 and @xmath78 is the difference between the spin operators of the two quarks . eq .
( [ eqn : mu - eff ] ) can be re - written as @xcite @xmath79 representing the effective magnetic dipole operator of our model .
note that the second term gives the @xmath80 transitions .
the mean values of the effective dipole operator can be easily calculated , so that the numerical values of free parameters ( @xmath81 and @xmath82 ) can be fitted to reproduce the proton and neutron magnetic moments , that are @xmath83 n.m.u . and
@xmath84 n.m.u . @xcite .
because the number of free parameters is larger than that of the experimental informations , we take @xmath85 ; the resulting values for the other parameters are @xmath86 and @xmath87 @xcite .
a more detailed study of the electromagnetic current will be carried out in a subsequent paper , with the aim of calculating the nucleon electromagnetic form factors and the helicity amplitudes of baryon resonances @xcite . as discussed in refs .
@xcite , the baryon spectra of quark - diquark and three quark models are quite distinct , the main differences being in the amount and energy of theoretically predicted states .
when theoretical results and experimental observations are compared , it is clear that three quark models predict an eccessive number of states .
this is the problem of missing resonances which can be , at least partially , solved in quark - diquark models in terms of a smaller number of effective degrees of freedom @xcite .
nevertheless , experimental informations on a hadron are not only limited to its mass , but also include ( open- and hidden - flavor ) strong decay widths , weak and e.m .
radiative transitions , and so on .
thus , we think it is also worthwhile to compare three quark and quark - diquark model predictions for some of these observables . here
, we discuss the formalism to compute the open - flavor strong decays of nonstrange baryons in the well - known @xmath88 pair - creation model @xcite . in the @xmath88 model
, a hadron decay takes place in its rest frame and proceeds via the creation of an additional @xmath89 pair ( @xmath90 and @xmath91 ) with vacuum quantum numbers , i.e. @xmath92 . the decay amplitude @xmath93 can be expressed as @xcite @xmath94 where @xmath95 is the so - called @xmath88 amplitude , @xmath96 the phase space factor for the decay , @xmath97 and @xmath98 the relative momentum and orbital angular momentum of the @xmath99 pair .
[ fig : qd-3p0 ] .
we write the phase space factor in the usual relativistic form , @xmath100 , where @xmath101 ( i = @xmath102 ) .
the transition operator of the @xmath103 model is given by @xcite : @xmath104^{(0)}_0 \\ & \times & b_1^{\dagger}(\vec{p}_{\rm q_1 } ) \
, d_2^{\dagger}(\vec{p}_{\rm q_2 } ) \mbox { } .
\end{array } \label{3p0}\ ] ] here , @xmath105 is the pair - creation strength , @xmath106 and @xmath107 the creation operators for a quark and an antiquark with momenta @xmath108 and @xmath109 , respectively , @xmath110 the pair - creation vertex or quark form factor .
the @xmath89 pair is characterized by a color singlet wave function @xmath111 , a flavor singlet wave function @xmath112 , a spin triplet wave function @xmath113 with spin @xmath114 and a solid spherical harmonic @xmath115 , since the quark and antiquark are in a relative @xmath116 wave .
pair - creation model of hadron vertices is here used within a quark - diquark picture for baryons . @xmath38 and @xmath69 are , respectively , the spectator diquark and quark ; @xmath117 is the created @xmath89 pair , with @xmath88 quantum numbers .
springer - verlag wien copyright.,width=226 ] [ [ baryon - spectrum - with - self - energy - corrections - in - the - quark - diquark - model ] ] baryon spectrum with self - energy corrections in the quark - diquark model ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the formalism described in the previous section can also be used to compute the baryon @xmath118 baryon + meson vertices relevant to the calculation of baryon self - energies . for a calculation of octet and decuplet baryon self - energies within a three quark model ,
@xcite .
we consider the hamiltonian @xmath119 where @xmath120 is the `` unperturbed '' part , describing the interaction between constituent ( valence ) quarks [ see eq . ( [ eqn : h0 ] ) ] , and @xmath121 the interaction which can couple a baryon state to the baryon - meson continuum .
the physical mass of a baryon , @xmath122 , can be written as @xmath123 where @xmath124 is the bare mass , obtained by solving the eigenvalue problem of @xmath120 , and @xmath125 the self - energy correction . in eq .
( [ eqn : self - a ] ) , one has to sum over a complete set of baryon - meson intermediate states , @xmath126 .
these channels , with relative momentum @xmath69 between @xmath127 and @xmath128 , have quantum numbers @xmath129 and @xmath98 coupled to the total angular momentum of the initial state @xmath130 .
@xmath131 stands for the coupling between the intermediate state @xmath126 and the unperturbed wave function of the baryon @xmath132 .
several choices for @xmath121 are possible , including the @xmath88 pair - creation model operator @xmath133 of eq .
( [ 3p0 ] ) , which is ours . in the following section , we provide an example of a @xmath88 model vertex calculation ( @xmath134 ) in the quark - diquark model .
this particular vertex provides the largest contribution to the @xmath135 self - energy @xcite . as an example
, we briefly discuss the calculation of the @xmath134 vertex .
the matrix elements can be calculated according to refs .
@xcite , with a few differences . in the special case of @xmath136
, the transition amplitude can be written as ( * ? ? ?
6 ) @xmath137 where @xmath138 and @xmath139 are the flavor and spatial matrix elements , respectively , @xmath140 the spin of the @xmath99 pair , @xmath141 and @xmath142 the spin of the spectator diquark and quark , @xmath143 and @xmath144 the spins of the @xmath88 quark and antiquark .
we also use the notation @xmath145 . in the @xmath146 case ,
( [ eqn:3p0-general ] ) reduces to @xmath147 the flavor matrix element can be written as @xcite @xmath148 where @xmath149 is the su(3 ) flavor singlet , @xmath150 , @xmath151 and @xmath152 the flavor wave functions of the corresponding hadrons @xcite . in the @xmath146 case ,
( [ eqn : flavor - me ] ) can be written as @xmath153 \\ & & \left [ \left| \{u , u\}u \right\rangle \otimes \left ( \left| u \bar u \right\rangle + \left| d \bar
d \right\rangle + \left| s \bar s \right\rangle \right ) \right ] = \frac{1}{\sqrt 6 } \mbox { } , \end{array}\ ] ] where @xmath154 @xmath155 , @xmath156 , and the notation @xmath157 stands for an axial - vector diquark @xcite .
the spatial matrix element are the same as those for meson @xmath118 meson + meson transitions @xcite .
we use the coordinate system @xmath158 , @xmath159 , @xmath160 , @xmath161 , where @xmath162 , @xmath163 , @xmath164 , and @xmath165 .
the spatial matrix elements can be written explicitly as @xmath166 where @xmath167 , @xmath168 and @xmath169 are single harmonic oscillator ( ho ) wave functions with ho parameters @xmath170 gev@xmath171 @xcite and @xmath172 gev@xmath171 @xcite ; @xmath173 $ ] is a gaussian quark form factor , with @xmath174 gev@xmath171 @xcite . in the case of ground - state hadrons , namely with no radial and orbital excitations , eq .
( [ eqn : spatial - general ] ) reduces to @xmath175 where @xmath176 @xmath177 @xmath178 \mbox { } , \ ] ] and @xmath179 ( i = a , b , c ) .
e. santopinto and j. ferretti , strange and nonstrange baryon spectra in the relativistic interacting quark - diquark model with a grsey and radicati - inspired exchange interaction .
c 92:025202 ( 2015 ) .
m. de sanctis , j. ferretti , e. santopinto and a. vassallo .
relativistic quark - diquark model of baryons with a spin - isospin transition interaction : non - strange baryon spectrum and nucleon magnetic moments .
j. a 52:121 ( 2016 ) .
m. ferraris _
et al._. a three body force model for the baryon spectrum .
b 364:231 ( 1995 ) ; m. aiello _
et al._. a three body force model for the electromagnetic excitation of the nucleon .
b 387:215 ( 1996 ) ; m. m. giannini and e. santopinto . the hypercentral constituent quark model and its application to baryon properties .
. j. phys .
53:020301 ( 2015 ) .
u. loring , b. c. metsch and h. r. petry .
the light baryon spectrum in a relativistic quark model with instanton induced quark forces : the nonstrange baryon spectrum and ground states .
j. a 10:395 ( 2001 ) .
a. le yaouanc , l. oliver , o. pene and j. -c .
naive quark pair creation model of strong interaction vertices .
d 8:2223 ( 1973 ) ; naive quark pair creation model and baryon decays .
d 9:1415 ( 1974 ) . | we describe the relativistic interacting quark - diquark model formalism and its application to the calculation of strange and nonstrange baryon spectra .
the results are compared to the existing experimental data .
we also discuss the application of the model to the calculation of other baryon observables , like baryon magnetic moments , open - flavor strong decays and baryon masses with self - energy corrections . |
thermoelectric material , which can directly interconvert thermal and electrical energy through the seebeck and peltier effects , has attracted increasing attention as a solution to realize a sustainable and emission - free society without environmental pollution @xcite .
it offers simple and compact solid - state energy - conversion devices that are adaptive in a variety of situations to utilize various kinds of waste heat such as car engine and garbage furnace . in this context ,
oxide thermoelectrics is potentially promising because it is fairly stable at high temperatures in air , and since the discovery of the good thermoelectric properties of the layered cobalt oxide na@xmath11coo@xmath12 @xcite , oxide thermoelectrics including transition - metal elements has been intensively studied in the past decade @xcite . among the oxide thermoelectrics ,
the layered ca@xmath0co@xmath1o@xmath2 is a peculiar example because of its highly unusual transport behaviors intimately related to the complicating magnetic properties .
ca@xmath0co@xmath1o@xmath2 consists of alternate stacking of two types of layered subsystems with different @xmath13-axis parameter ( fig .
1 ) @xcite : the cdi@xmath12-type coo@xmath12 layer responsible for the charge transport and the rocksalt - type ca@xmath12coo@xmath0 layer which behaves as the charge reservoir to supply holes into the coo@xmath12 layer @xcite .
the anomalous charge transport of this compound is clearly seen in the non - monotonic temperature variations of the electrical resistivity and the seebeck coefficient @xcite , but an underlying origin for these behaviors remains unclear .
for instance , below @xmath14 k , the resistivity increases with decreasing temperature like an insulator while the seebeck coefficient shows a metallic behavior , which has been discussed in terms of the variable range hopping @xcite , quantum criticality @xcite , and pseudogap opening @xcite associated with a spin - density - wave ( sdw ) formation @xcite . at higher temperature
, the resistivity shows a fermi - liquid behavior varying as @xmath15 up to the coherence temperature @xmath16 k , above which an incoherent bad - metal state is realized .
the seebeck coefficient reaches a large value of 130 @xmath17v / k at around 150 k , and becomes almost temperature independent up to 400 k , as described within the modified heikes formula in which the spin and orbital degrees of freedom of correlated co @xmath10 electrons are involved @xcite . on the other hand ,
the seebeck coefficient again increases with increasing temperature above 400 k , suggested to be associated with a spin - state crossover @xcite , but the high - temperature spin - state nature of ca@xmath0co@xmath1o@xmath2 remains puzzling @xcite .
note that the similar issue regarding an excited spin state is greatly debated in the perovskite lacoo@xmath0 @xcite .
co@xmath1o@xmath2 viewed from ( a ) @xmath13-axis and ( b ) @xmath18-axis directions .
the solid and dashed lines in ( b ) respectively represent the unit cells for coo@xmath12 and rocksalt layers with different @xmath13 axis parameters . ] to clarify a precise role of correlated co @xmath10 electrons for the intriguing electronic properties in this compound , it is quite valuable to substitute the co ions by isovalent transition - metal elements with different orbital size and correlation effect @xcite .
the physical properties of polycrystalline samples of isovalent rh - substituted ca@xmath0co@xmath1o@xmath2 have been recently examined for this purpose @xcite , however , the polycrystalline samples have an unavoidable extrinsic effect on the charge transport such as the boundary scattering . also note that the anisotropy of the resistivity in the parent compound is significantly large @xcite .
the aim of this paper is to experimentally investigate how an extended rh 4@xmath19 orbital affects the transport properties of ca@xmath0co@xmath1o@xmath2 using single - crystalline samples .
as claimed in the previous polycrystalline study @xcite , in the low - temperature fermi - liquid regime , the @xmath8 value in the resistivity is reduced by rh substitution , indicating that an electron - electron correlation effect is weakened by rh @xmath9 electrons with broad orbitals .
we find that , in contrast to such a weak correlation effect , the low - temperature seebeck coefficient is unexpectedly enhanced in the rh - substituted single crystals , which may attribute to an enlargement of the pseudogap in the rh - substituted samples associated with the short - range order of sdw . in the incoherent transport regime above room temperature , the resistivity is reduced by rh substitution , which also stems from the extended @xmath20 orbitals of rh @xmath9 electrons with light effective mass .
the seebeck coefficient of the rh - substituted single crystals is found to become almost temperature - independent at high temperatures , qualitatively understood from the spin state of rh ions , which favors the low - spin state owing to a large crystal field splitting for the @xmath9 electrons .
the experiments were performed using ca@xmath0co@xmath5rh@xmath6o@xmath2 single crystals ( @xmath21 ) with typical sample dimensions of @xmath22mm@xmath23 grown by a flux method @xcite .
precursor powders of caco@xmath0 ( 99.9% ) , co@xmath0o@xmath24 ( 99.9% ) , and rh@xmath12o@xmath0 ( 99.9% ) were mixed in a stoichiometric ratio and calcined two times in air at 1173 k for 24 h with intermediate grindings . then kcl ( 99.999% ) and k@xmath12co@xmath0 ( 99.999% ) powders mixed with a molar ratio of @xmath25
was added with the calcined powders of ca@xmath0co@xmath5rh@xmath6o@xmath2 as a flux .
the concentration of ca@xmath0co@xmath5rh@xmath6o@xmath2 was set to be 1.5% in molar ratio .
the mixture was put in an alumina crucible and heated up to 1123 k in an electrical furnace with a heating rate of 200 k / h . after keeping the maximum temperature for 1 h ,
the furnace was cooled down slowly with a rate of 1 k / h , and at 1023 k , the electrical power of the furnace was switched off .
as - grown samples were rinsed in distilled water to remove the flux and platelet crystals with a shiny surface were obtained .
the x - ray diffraction measurement was performed with cu @xmath26 radiation in a @xmath27-@xmath28 scan mode .
the scattering vector was normal to the crystal surface .
the in - plane resistivity was measured using a standard four - probe method .
excitation current of @xmath29 @xmath17a was provided by a keithley 6221 current source and the sample voltage was measured with a synchronized keithley 2182a nanovoltmeter .
these two instruments were operated in a built - in delta mode to cancel unwanted thermoelectric voltage .
the in - plane seebeck coefficient was measured using a steady - state technique with a typical temperature gradient of 0.5 k / mm made by a resistive heater . the thermoelectric voltage of the sample was measured with keithley 2182a nanovoltmeter .
the temperature gradient was measured with a differential thermocouple made of copper and constantan in a liquid he cryostat from 4 k to 300 k and with two platinum temperature sensors in a high - temperature equipment from 300 k to 650 k. in both measuring apparatuses , the thermoelectric voltage from the wire leads was carefully subtracted .
co@xmath1o@xmath2 and ( b ) ca@xmath0co@xmath30rh@xmath31o@xmath2 .
inset ( a ) compares these two spectra around the ( 002 ) reflection .
the interplanar distance @xmath32 of ca@xmath0co@xmath5rh@xmath6o@xmath2 determined from @xmath33 peaks is shown in the inset ( b ) as a function of rh content @xmath3 . ]
figures 2(a ) and 2(b ) represent typical x - ray diffraction patterns of the as - grown single crystals of ca@xmath0co@xmath1o@xmath2 ( @xmath34 ) and ca@xmath0co@xmath30rh@xmath31o@xmath2 ( @xmath4 ) , respectively , measured with the scattering vector normal to the surface of the plate - like samples .
all the peaks are indexed as @xmath33 reflections of the parent compound , meaning that the measured crystal surface is the @xmath35 plane . in bi - substituted ca@xmath0co@xmath1o@xmath2
, it is reported that an impurity phase is grown even in single crystals @xcite .
the present crystals do not show such impurity phase even for @xmath4 sample .
the inset of fig .
2(a ) compares the diffraction patterns of @xmath34 and @xmath4 samples around the ( 002 ) reflection .
this ( 002 ) peak is shifted low angle with increasing rh content @xmath3 owing to the larger ionic radius of rh than that of co. since the crystal structure of ca@xmath0co@xmath1o@xmath2 is monoclinic ( @xmath36 ) , it is impossible to determine the @xmath37-axis parameter from only @xmath33 peaks .
instead , we have obtained the interplanar distance @xmath32 from @xmath38 , where @xmath39 is integer and @xmath40 is the wavelength of incident x - ray , by using all the @xmath33 reflections . as shown in the inset of fig .
2(b ) , @xmath32 linearly increases as a function of @xmath3 , indicating successful rh substitutions . at present
, it is unclear whether co site of coo@xmath12 or the rocksalt layer is substituted by rh ions .
while the precise position of rh ions should be studied in future spectroscopic experiments , a considerable and systematic change in the following transport properties implies that rh ions may substantially enter the conductive coo@xmath12 layer .
co@xmath5rh@xmath6o@xmath2 single crystals ( @xmath41 ) .
the fermi - liquid transport coefficient @xmath8 is shown in the inset of ( a ) as a function of rh content @xmath3 .
inset of ( b ) shows the temperature variations of the power factor @xmath42 . ]
figure 3(a ) shows the temperature variations of the electrical resistivity measured along the in - plane direction . around room temperature ,
the in - plane resistivity is about 4 m@xmath43 cm for all the samples , which is the same as the single - crystalline result of the parent compound @xcite .
below room temperature it is largely increased with increasing rh content @xmath3 . for @xmath44 samples ,
the resistivity decreases with decreasing temperature down to around 100 k and then shows an upturn towards lower temperatures . in the metal - like regions around 120 k ,
the resistivity exhibits a fermi - liquid behavior varying as @xmath15 , and the @xmath8 value decreases with rh content @xmath3 as displayed in the inset of fig .
this behavior is also reported in the previous study on the polycrystalline samples @xcite , although the absolute value is quite different because of the single - crystalline nature in the present study .
since @xmath8 can be scaled to @xmath45 , where @xmath46 is the effective mass @xcite , this result indicates that the effective mass becomes small with rh substitutions , as is naturally understood from a weak correlation effect of rh @xmath9 electrons compared with co @xmath10 electrons . for @xmath4 sample ,
such a fermi - liquid behavior is no longer observed .
in contrast to the low - temperature results , the resistivity decreases with increasing rh content above room temperature .
the resistivity is less than 2 m@xmath43 cm for @xmath4 sample at 600 k , which is beneficial for oxide thermoelectrics at high temperatures as discussed later . in the parent compound
, the resistivity shows a broad maximum around 500 k and then decreases with heating . with increasing rh content , the broad peak shifts to the lower temperature , and disappears for @xmath4 sample , in which the resistivity shows an insulating behavior in the whole temperature range we measured .
we infer that such an insulating behavior is ascribed not to the carrier concentration , which is essentially constant above room temperature , but to the mobility , as will also be discussed in the results of the seebeck coefficient .
such a nonmetallic resistivity due to the mobility has also been observed in several oxide materials @xcite .
figure 3(b ) displays the temperature dependence of the in - plane seebeck coefficient .
below room temperature , the seebeck coefficient slightly but systematically changes with rh content ; in particular , below 100 k where the seebeck coefficient shows a metallic behavior , it is considerably enhanced by rh substitution .
the present results are distinct from the previous report using polycrystalline samples which shows that rh substitution does not affect the seebeck coefficient @xcite , probably attributed to an extrinsic effect of averaging large transport anisotropy in the polycrystalline samples . above room temperature ,
the enhancement of the seebeck coefficient seen in the parent compound is significantly suppressed with rh substitution . for @xmath4 sample , the seebeck coefficient becomes almost temperature independent up to 650 k , where the heikes formula with a constant carrier density looks applicable .
let us discuss an origin of the transport behaviors in this system .
we first focus on the low - temperature upturn of the resistivity coexisting with a metallic seebeck coefficient below 100 k. although it has been often discussed in terms of variable range hopping due to carrier localization @xcite , it is claimed that the temperature dependence of the resistivity is much milder than that of the variable - range - hopping transport @xmath47 ( @xmath48 ) @xcite and that the fitting range is narrow .
a scenario of quantum criticality is quite intriguing @xcite , but it is difficult to explain the seebeck coefficient of the rh - substituted samples in following reason ; as mentioned before , the @xmath8 value in the resistivity , a measure of strength of the correlation effect , decreases with increasing rh content , as is understood from a weak correlation of rh @xmath9 electrons .
this indicates that the rh - substituted samples locate far from the quantum critical point . since the @xmath8 value is intimately related to the electronic specific heat and seebeck coefficient through the effective mass @xcite , the rh - substituted samples with lower effective mass should have a smaller seebeck coefficient than that of the parent ca@xmath0co@xmath1o@xmath2 in the case of the quantum critical scenario
this is , however , opposite to the present results in which the seebeck coefficient is increased by rh substitution at low temperatures . as an origin of the low - temperature transport behaviors , we consider a formation of the pseudogap associated with the short - range order of sdw , which is suggested in bi - substituted ca@xmath0co@xmath1o@xmath2 single crystals @xcite .
at low temperatures , the seebeck coefficient in a two - dimensional metal is expressed as @xmath49 where @xmath50 is the charge of carrier , @xmath51 the @xmath37-axis length , @xmath39 the carrier concentration @xcite .
now the conduction carrier is hole , hence @xmath52 . using eq .
( 1 ) , @xmath53 can be evaluated from @xmath54 as @xmath55 figure 4 depicts the temperature dependence of @xmath54 at low temperatures .
hsieh @xmath56 @xmath57 have shown that the decrease of @xmath54 with lowering temperature originates from the decrease of the carrier concentration @xmath39 according to eq .
( 2 ) @xcite , which is also supported from the results of the hall coefficient measurements @xcite , and proposed that this phenomenon is ascribed to an opening of the pseudogap caused by the short - range order of sdw below 100 k @xcite .
we also note that such a reduction of @xmath54 below a pseudogap temperature is observed in other layered cobalt oxide @xcite and the kondo semiconductor cenisn @xcite . of ca@xmath0co@xmath5rh@xmath6o@xmath2 single crystals ( @xmath41 ) at low temperatures .
the inset depicts the rh content @xmath3 dependence of @xmath58 at which the resistivity takes the minimum value in the low - temperature range . ]
we then discuss how the rh substitution affects the pseudogap behavior . in the present study
, @xmath54 systematically decreases with rh substitution below 100 k. within the pseudogap scenario , the present result shows that @xmath39 decreases with increasing rh content , indicating that the pseudogap associated with sdw becomes large by rh substitution .
now we stress that this is consistent with our resistivity results . in this system ,
the resistivity minimum around 70 k is associated with an onset of sdw gap formation @xcite .
therefore , according to above scenario , the onset temperature should increase with increasing rh content .
the inset of fig .
4 displays the rh content dependence of @xmath58 at which the resistivity takes the minimum value in the low - temperature range .
indeed , the onset temperature @xmath58 increases with increasing rh content , supporting the pseudogap scenario . on the other hand
, it is an open question at present why the pseudogap is stabilized by rh substitution .
an underlying mechanism should be examined by future microscopic measurements .
next we remark the high - temperature transport in this system . in the parent compound ,
the seebeck coefficient around room temperature is well explained by the modified heikes formula @xcite , which considers a hole hopping between co@xmath59 and co@xmath60 sites both of which take low - spin states @xcite .
in contrast , an origin of the enhancement of the seebeck coefficient above 400 k is not clear .
it is theoretically suggested that a mixture of low - spin and high - spin states may enhance the seebeck coefficient @xcite .
thus , a thermally - induced spin - state crossover from low - spin into higher - spin state is suggested for the enhancement of seebeck coefficient @xcite , but the nature of spin states at high temperatures is still puzzling @xcite . in the present study , the rh substitution suppresses such an enhancement , and the seebeck coefficient becomes almost temperature independent for @xmath4 sample .
this behavior may be naively understood within a suppression of the spin - state crossover by rh substitution .
in contrast to co ions , the spin state of which often dramatically varies with temperature , rh ions usually show the stable low - spin state owing to the large crystal field splitting between @xmath20 and @xmath61 manifolds @xcite .
thus , at high temperatures , the number of the transition - metal ions with higher - spin states , which can enhance the seebeck coefficient , would be decreased in average by increasing rh content .
this leads to a possible suppression of the high - temperature seebeck coefficient in rh - substituted crystals .
let us discuss an origin of the reduction of the high - temperature resistivity in the rh - substituted samples . in the high - temperature range ,
the seebeck coefficient is described by the heikes formula , indicating that the carrier concentration is temperature - independent .
we also note that it shows almost the same value of 130 @xmath17v / k from 200 to 300 k for all the samples .
this implies that the carrier concentration does not change with rh substitution , which is also supported by the isovalent substitution .
therefore the decrease of the resistivity in rh - substituted samples originates from an increase of hole mobility by rh substitution . here
, the hole hopping occurs in the @xmath20 manifolds of co / rh ions , and the effective mass of carriers becomes small with increasing rh content because of the weak correlation and the broad orbital of the rh @xmath9 electrons .
this may lead to a higher mobility in the rh - substituted crystals .
we also mention a bad - metal state above @xmath16 k realized in the parent compound . in a bad - metal state , a hallmark of strongly correlated electron systems @xcite , the resistivity is increased with increasing temperature beyond the mott limit . in the present study , as seen in the fermi - liquid parameter @xmath8 in the resistivity , the correlation effect is weakened in rh - substituted samples and consequently the bad - metal state disappears in @xmath4 sample . since the resistivity increases in the bad - metal state , the disappearance of bad - metal state , along with the increase of hole mobility discussed above , may lead to a reduction of the resistivity in @xmath4 sample . finally let us comment on the thermoelectric energy - conversion efficiency of the present crystals .
the inset of fig .
3(b ) shows the temperature variations of the power factor @xmath42 of the present crystals . for @xmath4 sample ,
the power factor reaches a large value of @xmath62 w / mk@xmath63 at 600 k thanks to the low resistivity .
it is interesting to enlarge the power factor by decreasing effective mass in contrast to the usual approach to the high thermoelectric performance @xcite .
moreover , previous report using the polycrystalline samples shows that the rh substitutions strongly suppress the thermal conductivity in this system @xcite .
thus the thermal conductivity of the single - crystalline samples should be investigated as a future issue .
an importance of the thermoelectric study using single crystals is also highlighted by the discovery of the high thermoelectric efficiency in snse single crystals @xcite .
we study the electrical and thermoelectric transport properties of the layered ca@xmath0co@xmath5rh@xmath6o@xmath2 single crystals in a broad temperature range from 10 k up to 650 k. in the metal - like region around 120 k , the fermi - liquid transport coefficient @xmath8 decreases with increasing rh content @xmath3 , indicating smaller effective mass in the rh - substituted samples because of the weak correlation effect of rh @xmath9 electrons compared with co @xmath10 electrons . at lower temperatures , in contrast to the weak correlation effect found in the rh - substituted samples , we observe that the seebeck coefficient increases with increasing rh substitution , which is also different from the results of the previous polycrystalline study .
we suggest that this behavior is attributed to an enlargement of a pseudogap in the rh - substituted crystals , which is accompanied with the short - range order of the spin density wave below 100 k , as is supported from the increase of the spin - density - wave onset temperature by rh substitution . at higher temperatures ,
the resistivity is decreased with increasing rh content owing to a small effective mass of rh - substituted samples .
such a low resistivity in rh - substituted crystal indeed enhances the power factor at high temperatures .
the seebeck coefficient of ca@xmath0co@xmath30rh@xmath31o@xmath2 becomes almost temperature independent up to 650 k , possibly originating from a suppression of a spin - state crossover by the substitution of rh ions which favor the low - spin state .
the microscopic measurements to elucidate the spin state in the present system are highly required as a future issue .
this work was supported by a grant - in - aid for challenging exploratory research ( no .
26610099 ) from jsps .
r. f. klie , q. qiao , t. paulauskas , a. gulec , a. rebola , s. t , m. p. prange , j. c. idrobo , s. t. pantelides , s. kolesnik , b. dabrowski , m. ozdemir , c. boyraz , d. mazumdar , and a. gupta , phys .
lett . * 108 * , 196601 ( 2012 ) .
m. w. haverkort , z. hu , j. c. cezar , t. burnus , h. hartmann , m. reuther , c. zobel , t. lorenz , a. tanaka , n. b. brookes , h. h. hsieh , h .- j .
lin , c. t. chen , and l. h. tjeng , phys .
lett . * 97 * , 176405 ( 2006 ) . | we have examined an isovalent rh substitution effect on the transport properties of the thermoelectric oxide ca@xmath0co@xmath1o@xmath2 using single - crystalline form . with increasing rh content @xmath3 , both the electrical resistivity and the seebeck coefficient change systematically up to @xmath4 for ca@xmath0co@xmath5rh@xmath6o@xmath2 samples . in the fermi - liquid regime where the resistivity behaves as @xmath7 around 120 k , the @xmath8 value decreases with increasing rh content , indicating that the correlation effect is weakened by rh @xmath9 electrons with extended orbitals .
we find that , in contrast to such a weak correlation effect observed in the resistivity of rh - substituted samples , the low - temperature seebeck coefficient is increased with increasing rh content , which is explained with a possible enhancement of a pseudogap associated with the short - range order of spin density wave . in high - temperature range above room temperature ,
we show that the resistivity is largely suppressed by rh substitution while the seebeck coefficient becomes almost temperature - independent , leading to a significant improvement of the power factor in rh - substituted samples .
this result is also discussed in terms of the differences in the orbital size and the associated spin state between co @xmath10 and rh @xmath9 electrons .
= 10000 |
the interplay of superconducting and magnetism , two competing orderings , has been long studied .
a new superconducting state may arise , the so - called fflo state , where both the superconducting order parameters and the magnetic ordering are nonuniform @xcite .
the interplay of superconductivity and magnetism can be naturally studied in s / f heterostructures ( s : superconductor , f : ferromagnet ) , where a spatial separation of the two competing orderings can enhance and somehow control the interaction between superconductivity and magnetism .
the physics of s / f heterostructures has been deeply studied @xcite in recent years , and new , interesting aspects were theoretically predicted and in some cases experimentally confirmed . as an example , it has been predicted that the proximity - induced superconducting order parameter does not simply decay in the f layer as in nonmagnetic metals .
instead , it oscillates in the depth of the f layer with a scale given by the coherence length in the f layer , @xmath0 .
this behaviour gives rise to spectacular phenomena , such as the oscillations of @xmath1 with @xmath2 , the thickness of the f layer @xcite , instead of a monotonous depression of @xmath1 with @xmath2 itself . also , the vortex state and pinning properties in s / f structures , or in s films with f particles , dots , pillars , wires etc .
@xcite received much attention . among the many properties that can give important contributions to the understanding of the physical background of superconductors , and thus of s / f heterostructures , stand the microwave response
indeed , the temperature dependence of the penetration depth @xmath3 , giving access to the temperature dependence of the superfluid density @xmath4 , has been measured in some s / f bilayers and s / f / s trilayers : measurements on nb / cumn bilayers @xcite revealed a temperature dependent superfluid fraction largely different from the case of superconducting / normal metal bilayers , suggesting a temperature - independent proximity induced correlation length .
mn impurities in nb / cu bylayers did not affect the estimated zero - temperature @xmath3 @xcite . in nb / ni bilayers @xcite the temperature dependence of the superfluid density and the @xmath2 dependence of @xmath5 were found to be consistent with significant superfluid density inside the f layer .
while there is certainly need for additional measurements of the penetration depth , it is surprising that there seem to be no measurements at all of the vortex state complex resistivity at radio and microwave frequencies , as opposed to conventional superconductors @xcite , diborides @xcite , cuprates @xcite , nanostructured high-@xmath1 superconductors @xcite .
such measurements can be particularly useful since they give direct access to the pinning constant and the free - flux - flow resistivity ( see below ) .
thus , we started an experimental investigation of the microwave properties of s / f / s heterostructures in the vortex state .
microwave resonators are a well - established technique to measure with high resolution the complex resistivity of thin superconducting films in the vortex state @xcite .
however , the resonators lack the capability to perform measurements in an extended ( and continuous ) frequency range .
since no measurements exist of the vortex state microwave resistivity in s / f heterostructures , it would be hazardous to rely on single - frequency measurements to draw conclusions on the usually complicated response in the vortex state .
thus , we applied to s / f / s trilayers the nonresonant , wideband corbino disk technique @xcite : the technique is able to reliably extract the resistivity in a rather wide frequency range , conservatively between 1 and 20 ghz .
it is then possible to compare the measurements to a wide class of models to get the important vortex state parameters . before describing with some detail the technique , we briefly recall the expectations of the frequency - dependent complex resistivity in the vortex state , @xmath6 , as calculated within mean field models with a static magnetic induction field @xmath7 perpendicular to the microwave currents . at microwave frequencies the vortex displacement due to the force exerted by the alternate currents @xmath8 is very small @xcite , and it is customary to write down the equation of forces ( per unit length ) acting on a single vortex line in the local limit as follows @xcite : @xmath9 where @xmath10 is the vortex velocity , @xmath11 is the unit vector along the vortex , @xmath12 is the flux quantum , @xmath13 is the lorentz force exerted on the flux line and the stochastic force @xmath14 takes into account thermal effects , responsible for fluxon jumps between pinning sites .
the viscous drag coefficient @xmath15 , which is also commonly referred to as vortex viscosity , takes into account the power dissipated by moving vortices due to , e.g. , the relaxation processes of quasiparticles and scattering processes due to currents flowing in the vortex cores @xcite .
as such , it bears informations related to the microscopic electronic state .
the effects of pinning are represented by the force @xmath16 where @xmath17 is the spatial function describing the pinning potential .
the appearance of a hall term in ( [ eq : forces ] ) can be neglected in superconductors in the dirty limit . from the basic equation of motion ( [ eq
: forces ] ) , many models can be derived , differing essentially for the impact of thermal effects and elasticity of the vortex lattice @xcite .
nevertheless , it has been shown @xcite that the seemingly large variety of models yield expressions for the vortex resistivity @xmath6 that can be cast in a single , very general expression : @xmath18 where the dimensionless parameter @xmath19 is a measure of the weight of creep phenomenon , @xmath20 is the measuring frequency and @xmath21 is the main characteristic frequency governing the vortex relaxation phenomena . with respect to the well - known depinning frequency @xcite @xmath22
, one has @xmath23 when @xmath24 , that is when thermal effects can be neglected ( one should mention that this is the approximation under which most of microwave measurements in conventional superconductors have been interpreted in the past ) .
the flux - flow resistivity @xmath25 represents the vortex resistivity in absence of any pinning and creep effect , that is the _ ideal _ vortex flow resistivity . as it can be seen , it is possible to obtain the true flux flow resistivity by simply increasing sufficiently the measuring frequency @xmath20 , without the need for ( a ) ultrapure samples and ( b ) high transport currents as in dc , with the related technical problems due to heating .
the frequency dependence of @xmath6 as described by ( [ eq : rhovm ] ) is depicted in figure [ figteo ] as a function of the reduced frequencies @xmath26 in the range @xmath27 . as it can be seen , the real part of the vortex resistivity @xmath28 contains basically the entire information on the vortex parameters , @xmath29 .
thus , measurements of @xmath28 over a sufficiently wide frequency range are of primary importance .
aim of this paper is to show how the swept - frequency corbino disk technique can be a valuable tool in the study of the vortex state resistivity , in particular in s / f heterostructures : the little role played by the models in the analysis of the data , together with the power of swept - frequency data make the corbino disk an ideal probe of new materials where one can not rely on known properties .
we also present the first measurements of the frequency - dependent vortex resistivity in two s / f / s trilayers , and we show that differences emerge with respect to simple nb even at extremely small @xmath2 . as a function of the reduced frequency @xmath26 plotted on the basis of equation [ eq : rhovm ] .
full line : @xmath28 , dashed line : @xmath30 .
the direct determination of the vortex parameters @xmath29 is indicated in the figure.,width=377 ] the paper is organized as follows . in the next section
we introduce the corbino disk technique , with the expressions linking the measured quantities and the complex resistivity of superconducting films in the vortex state . in section [ exp ]
we briefly describe our samples , and we present the experimental data for the complex resistivity taken in a nb film and two nb / pdni / nb trilayers , with different thickness of the f layer .
we summarize the results and suggest further developments in section [ conc ] .
we have developed in the past the corbino disk technique @xcite , following the pioneering work ( in the context of superconductivity ) of the maryland group @xcite .
the technique is conceptually elementary , consisting in the termination of a coaxial cable with the sample under study and the measurement of the complex reflection coefficient @xmath31 . in our particular case , we measured the complex reflection coefficient by means of a vector network analyser , or vna ( we performed most part of the measurements with an agilent - hp 8510c , and a smaller part of the measurements with an anritsu 37269 d ) , connected to a microwave line made up of commercial coaxial cables with physical cutoff frequency at 60 ghz . in principle , from the measured @xmath31 one can extract the real and imaginary part of the complex resistivity of a sample terminating the cable .
the practical realization turns out to be far from being trivial .
the technique is inherently nonresonant , and the reflection coefficient is heavily affected by losses in the cable , phase stability of the cable itself and standing wave in the necessarily long microwave line connecting the sample in the cryostat to the external vna .
the achievable sensitivity of the measurement depends crucially on the calibration and on the characteristics of the vna employed . while very small signals can be in principle detected with a full calibration ( by using three known standard loads , typically an open , a short and a load of known impedance ) ,
this is not feasible in the present case : since both temperature at sample position and temperature gradients along the cable affect the transmission line , the conventional calibration is almost impossible to achieve .
thus , an approximate calibration , with the sample in - situ , has to be used , as detailed below .
we remark that the approximate calibration is the limiting factor in assessing the sensitivity of the present measurements . as a rule of thumb , and with some conservative overestimates , in our present setup and with the chosen calibration procedure no signal below 1% of the normal state of the superconducting sample can be detected with sufficient reliability .
however , the invaluable advantage of swept - frequency measurements compensates for the reduced sensitivity with respect to , e.g. , resonator - based microwave measurements .
we described elsewhere @xcite both the measuring cell and the issues related to the measurement procedure , with particular attention to calibration issues .
we here recall the basic aspects of the technique and the specific calibration method here employed .
the reflection coefficient at the sample surface @xmath32 is related to the impedance of the sample @xmath33 through the relation @xcite @xmath34 where @xmath35 is the characteristic ( wave ) impedance of the dielectric which fills the cable .
( vacuum impedance ) if the last section of the cable is not filled with the dielectric . in practice ,
the device ( launcher ) actually placed at the terminal section of the cable to make the contact between the coaxial cable and the sample , does not include any dielectric but introduces an additional contribution which in general requires to experimentally determine @xmath35 . ] for thin films ( that is , for films whose thickness is lower than the penetration depth ) the sample complex resistivity @xmath36 is related to the sample impedance through the relation @xmath37 , so that @xcite @xmath38 unfortunately , a direct measure of @xmath32 is impossible .
one measures instead ( see figure [ fig : corbinoline ] ) the reflection coefficient at the beginning of the microwave line @xmath39 , which is related to @xmath32 through cable coefficients that should be determined through some calibration procedure .
since those coefficients depend on temperature and temperature gradients of the cable , a full calibration is in general not possible .
this crucial problem can be partially overcome @xcite by making use of a calibration procedure which allows to obtain _ variations _ of @xmath40 with some external parameter ( e.g. , the temperature @xmath41 , the applied magnetic field @xmath42 ) . for instance , by measuring @xmath39 at two different temperatures , it is possible to obtain the sample impedance difference between the two measured temperatures .
this procedure is based on the hypothesis that most of the cable parameters vary only smoothly with temperature .
this hypothesis can be verified through the measured data @xcite . assuming this hypothesis is valid , it can be shown that the measured reflection coefficient @xmath39 can be related to the reflection coefficient @xmath32 at the film surface through the relation @xmath43 the complex coefficient @xmath44 describes in a compact way the residual cable contribution to the measurement . in the present measurements of the vortex state response ,
it is essential that this quantity is field independent : non - magnetic cables and connectors have to be used . by focusing on field sweep measurements ( i.e. measures performed by varying the field at fixed @xmath45 ) ,
it is possible to derive @xmath46 by considering data taken at the two distinct fields @xmath47 and @xmath48 . at @xmath49
, one has @xmath50 , where @xmath51 and @xmath52 are the reflection coefficient and surface impedance of the sample in the normal state . for superconducting samples in the thin film limit , @xmath53 is a frequency independent real quantity ( since the plasma frequency of the material is usually much higher than the measuring frequency ) lower than @xmath35 , which is also real . as a consequence ,
@xmath51 is frequency independent , real and negative . at @xmath47
, one has @xmath54 . at low frequencies and not too close to @xmath1 , @xmath55\ll z_0
$ ] so that @xmath56 . :
the latter involves the additional requirement @xmath57\ll z_0 $ ] .
this is quite useful for the case of thin superconducting films , where usually @xmath58 . ] therefore , at low frequencies the ratio @xmath59 , which is a frequency independent constant .
an example based on the measurements on the nb sample is reported in figure [ fig : g0n ] .
it can be seen that , apart a residual oscillation at high frequency , that remains within 3% , the frequency independence of @xmath51 holds . at various temperatures for sample nb .
@xmath51 can be estimated precisely in the low frequency range @xmath60 ghz , where the measured response is constant as expected . at higher frequencies residual oscillations , within 3% ,
remains after the approximate calibration procedure.,width=377 ] since @xmath51 is a frequency - independent quantity , it is possible to estimate it by using low frequency data only , @xmath60 ghz .
it is then possible to determine @xmath46 , the important frequency - dependent cable coefficient , which contains @xmath51 as a proportionality constant , according to : @xmath61 and then used to extract @xmath62 for each set of measurements corresponding to a field sweep at fixed @xmath41 . finally , by using ( [ zs ] ) the frequency - dependent resistivity can be obtained . in the present measurements
there is the additional advantage that the normal state resistivity @xmath63 at low temperature of our samples is not temperature dependent .
thus , @xmath51 does not depend on temperature either and once obtained at one temperature the value can be used to derive @xmath64 at all @xmath41 .
one should remark that this kind of calibration is feasible only on thin films , where @xmath51 is frequency - independent : the same procedure can not be applied to the study of bulk materials ( e.g. , single crystals ) , being in this case @xmath51 dependent on the frequency . in this case , one should develop an alternative , sample - in - situ , calibration . using ( [ zs ] ) and ( [ extalpha ] ) one finally obtains for the sample complex resistivity @xmath65 : @xmath66 @xmath67 where we have used the approximations @xmath68 and @xmath69 [ where @xmath70 , which hold true since @xmath71 is small .
these expressions clearly show that the real part of the resistivity can be obtained solely by the modulus of the reflection coefficient , whereas the imaginary part strongly depends on the phase . on practical grounds ,
this is a relevant remark : whenever thermal or other instabilities affect the cable response , they impact largely and mainly on the measurement of the phase rather than on the modulus : in these cases , the above expressions show that although the imaginary part of the resistivity would be largely affected , the real part would still remain reliably determined . in the measurements here presented we limit ourself to the extraction of the more reliable real part of the resistivity , @xmath72 .
in this section we present experimental results for the frequency - dependent real part of the resistivity in the mixed state in three superconducting samples : a pure nb film of 20 nm nominal thickness and two nb / pd@xmath73ni@xmath74/nb trilayers .
the samples were fabricated at university of salerno / cnr - spin .
they were grown by ultra - high - vacuum , dc magnetron sputtering .
the films and heterostructures were grown at room temperature , and the thickness was carefully calibrated against deposition rate , following the procedure described in more details elsewhere @xcite . since
microwave measurements can be seriously distorted by the use of semiconducting substrates , as extensively discussed previously @xcite , samples were deposited on crystalline sapphire ( al@xmath75o@xmath76 ) instead of the more common choice of si .
we focused the attention on samples with small @xmath2 , in order to assess whether ferromagnetic layers thinner than the coherence length in the ferromagnet , @xmath0 , can give rise to detectable features in the transport properties , apart the well - known drop in @xmath1 @xcite . since in nb / pdni / nb one estimates @xmath77 3 - 4 nm @xcite , we have considered here the following three samples : a nb thin films ( sample nb ) , with nominal thickness @xmath7820 nm , as a reference , and two trilayers with nominal nb thickness @xmath79 15 nm for upper and lower layers and pdni thickness @xmath80 1 nm and 2 nm ( samples t1 and t2 , respectively ) .
the normal state dc resistivity was in all cases @xmath81 cm .
the main characteristic parameters are reported in the table .
the critical temperature @xmath1 and the upper critical fields @xmath82 were estimated from the disappearance of the microwave signal , and were found in good agreement with the dc data . the magnetic field @xmath42 was applied perpendicular to the film plane . as described in section [ tech ] , by measuring the changes with the applied magnetic field @xmath42 of the frequency - dependent complex reflection coefficient of the samples we derived the field - increase of @xmath83 on the basis of the equation ( [ rho1finale ] ) .
apart from a very narrow region in the @xmath84 phase diagram close to the transition line @xmath85 , we do not expect a detectable contribution of pairbreaking at our working frequency , so that we entirely ascribe the measured signal to the vortex resistivity .
we anticipate that the frequency dependence of the measured real resistivity is well described by the general theoretical expression ( [ eq : rhovm ] ) , with no need for other contributions , thus adding confidence in the assumption .
for this reason , when dealing with experimental quantity we will use the symbol @xmath28 for the experimental field increase of the real resistivity . in order to directly compare different samples
, we report in the following the normalized resistivity , @xmath86 .
measurements have been taken approximately at the same reduced temperature @xmath87 to compare similar thermal properties , and at the same magnetic fields @xmath880.6 and 0.8 t in order to have the same vortex density .
sufficiently large magnetic fields have been chosen in order to examine also the creep factor in the three samples , which becomes negligibly small at lower fields . in figure [ fignb ]
we report @xmath28 in the nb sample taken at @xmath894.5 k. we first notice the appearance of steps in @xmath28 at the larger field .
the resistivity switches toward a more dissipative ( larger @xmath28 ) state and then decays abruptly , then switches again .
we remark that , to the best of our knowledge , such instabilities have not been reported before .
in particular , a time - domain analysis reveals that the switching phenomena here reported have a very long typical time scale , @xmath901s @xcite .
we will not focus further on this anomalous behaviour , that is presently under study .
we mention that it is invariably present in our measurements , although somehow sample - dependent , it is generally more evident approaching the superconducting transition , taken close to @xmath82 , no switches are evident . ] and completely disappears in the normal state ( i.e. , for @xmath91 or @xmath45 ) .
this behaviour has been verified with both vnas .
further details and a characterization of the phenomenon have been given elsewhere @xcite .
we focus instead on the conventional " behaviour , indicated in figure [ fignb ] by full symbols , that we compare to the theoretical expression ( [ eq : rhovm ] ) . as a function of the frequency @xmath20 in the nb sample , at two different magnetic fields and @xmath894.5 k. symbols : experimental data ; open symbols : data in the switched " state ( see text ) .
continuous lines : fits with the general expression ( [ eq : rhovm ] ) , with parameters reported in the table.,width=377 ] directly from the data it can be seen that the vortex characteristic frequency lies well within our measuring range of 1 - 22 ghz , as indicated by the steep rise in @xmath86 . as a consequence , we are able to observe the free - flux - flow plateau in @xmath28 at the high edge of our measuring range . a quantitative fit with ( [ eq : rhovm ] ) is reported in figure [ fignb ] .
the fit is in good agreement with the data , and it gives the value for the vortex parameters reported in the table .
we note the rather high value of the characteristic frequency , @xmath925 and 6 ghz at @xmath930.6 and 0.8 t , respectively .
this result is consistent with the depinning frequency in nb thin films as measured with single - frequency resonators @xcite , and it indicates quite a strong pinning in our nb film .
we also note that the normalized flux - flow resistivity @xmath94 by a significant factor .
this remark is a direct demonstration that the simple bardeen - stephen model @xcite , where the flux flow resistivity @xmath95 with @xmath961 , does not describe well our data . while it should be mentioned that , when a full time - dependent ginzburg - landau calculation is performed , @xmath25 can be much smaller than the bs value @xcite
, it seems that this result , which we recall is independent from a particular model for vortex motion , calls for further studies of the flux flow resistivity in conventional superconductors .
measurements are presently in progress .
finally , we remark that a significant creep exists at the higher measuring field , @xmath97t@xmath98 .
this is an interesting feature , since the role of creep is often neglected in the analysis of high - frequency data in conventional superconductors @xcite , under the assumption that it is relevant only extremely close to the transition .
we notice that recent low - temperature , multifrequency microwave measurements in conventional superconductors revealed some contribution from flux creep @xcite .
it appears that multifrequency or , like in the present work , swept - frequency measurements are a particularly suitable tool to directly uncover features in the vortex state resistivity .
we propose a framework in order to explain the pinning and creep features found in the present measurements .
first , one should consider that at microwave frequencies the flux lines undergo very small displacements around their equilibrium position . in this sense ,
the pinning probed by the microwave response concerns the curvature at the bottom of the potential well , and has ( as a first approximation ) nothing to do with the height of the pinning well itself .
this is a significant difference with respect to conventional dc measurements and pinning , as measured by the critical current density @xmath99 , is essentially dictated by the height of the potential wells .
by contrast , creep is determined by the height of the wells both at microwaves and in dc .
the present measurements can then be explained in terms of rather steep , but not very deep pinning wells : the pinning - related parameter @xmath100 , being rather high , points to steep pinning wells , while the large creep factor indicates that the pinning wells themselves are not particularly deep .
a check of this framework comes from the magnetic field dependence of the vortex parameters : increasing the field should first reduce the height of the pinning wells , and only as a further effect it should affect the shape of the bottom of the well .
this behaviour should result in a quickly increasing creep factor with the field , and a weakly field dependent depinning frequency ( until @xmath82 is approached , where the pinning potential flattens and the characteristic frequency vanishes ) .
in fact , it can be seen in the table that with increasing field @xmath100 does not drop quickly , while @xmath101 increases significantly ( anticipating the results obtained in the trilayers , we observe the same trend , even if @xmath82 and pinning are smaller in sample t2 ) .
the physics at high driving frequency differs from the better known flux motion processes in dc : at low dc currents the motion of flux lines arises as flux bundles @xcite , which require quite a large activation energy .
hence , the small creep factor as seen in dc unless the current is raised .
by contrast , the motion at microwave frequencies is much closer to the single - fluxon motion , as shown by the study of the elastic force matrix for the flux lattice @xcite : with increasing frequency , the range of the transmitted forces between fluxons drops thus yielding the single - vortex regime . in this case , the activation energies related to single vortex creep are smaller @xcite .
having determined the vortex parameters in nb , we turn to the analysis of the data in the trilayers . as a function of the frequency @xmath20 in the trilayer t1 , with 1 nm pdni thickness , at two different magnetic fields and @xmath894.5 k. symbols : experimental data .
continuous lines : fits with the general expression ( [ eq : rhovm ] ) , with parameters reported in the table .
the frequency dependence of @xmath28 is very similar to analogous data for the reference nb sample ( figure [ fignb]).,width=377 ] figure [ figt1 ] reports @xmath86 in sample t1 , together with the fits with ( [ eq : rhovm ] ) .
it turns out that sample t1 , where the pdni thickness is limited to @xmath1021 nm , has a behaviour very close to the nb film for what concerns the vortex resistivity .
the effect of the f layer seems limited to the reduction of @xmath1 by about 1.2 k with respect to a single nb film of thickness 30 nm @xcite .
the vortex parameters are reported in the table .
again , direct indications of strong pinning and significant creep is found , and @xmath94 .
the same considerations already stated for nb are valid .
it seems that @xmath1021 nm is not sufficient to affect the vortex state resistivity in our s / f / s trilayers . as a function of the frequency @xmath20 in the trilayer t2 , with 2 nm pdni thickness , at two different magnetic fields and @xmath893.66 k. symbols : experimental data .
continuous lines : fits with the general expression ( [ eq : rhovm ] ) , with parameters reported in the table .
it can be directly seen from the data that the frequency dependence of @xmath28 is different with respect to the reference nb sample and to the trilayer with @xmath2= 1 nm ( figures [ fignb ] and [ figt1 ] , respectively).,width=377 ] the data for sample t2 are presented in figure [ figt2 ] , again with the fits obtained using ( [ eq : rhovm ] ) . even if the thickness of the f layer is only 2 nm , significant differences with nb and trilayer t1 appear .
first , pinning is significantly reduced , as evident from the much steeper raise in @xmath86 . performing the fits
, one finds that the characteristic vortex frequency is reduced by an appreciable amount with respect to nb and trilayer t1 , while creep does not change much .
however , the most remarkable feature is the value of the flux flow resistivity . in this case
we find @xmath103 , which is _ above _ the bs value .
this finding can not be easily explained , and clearly calls for further investigations .
it should be stressed that the dynamics remains unchanged , meaning with this statement that the frequency dependence is still the same of pure nb and of sample t1 , satisfactorily described by ( [ eq : rhovm ] ) . within the wide class of models represented by ( [ eq : rhovm ] )
, the high values of @xmath104 must find an explanation in the structure of the vortex itself , and/or in the quasiparticle relaxation processes in the superconducting state in presence of a magnetic field , since no pinning or thermal effects can be invoked .
a study of the vortex resistivity as a function of @xmath2 is clearly required , and it will be the subject of ongoing investigation .
@*7lccc & nb & t1 & t2 nb thickness ( nm ) @xmath105 & 20 & 15 & 15 pdni thickness ( nm ) @xmath105 & & 1 & 2 @xmath1 ( k ) @xmath106 & [email protected] & [email protected] & [email protected] measuring @xmath108 ( k ) @xmath109 & [email protected] & [email protected] & [email protected] @xmath110 ( t ) @xmath106 & [email protected] & [email protected] & [email protected] vortex parameters@xmath111 at @xmath930.6 t & & & @xmath104 & [email protected] & [email protected] & [email protected] @xmath21 ( ghz ) & [email protected] & [email protected] & [email protected] @xmath101 & [email protected] & [email protected] & [email protected] vortex parameters@xmath111 at @xmath930.8 t & & & @xmath104 & [email protected] & [email protected] & [email protected] @xmath21 ( ghz ) & [email protected] & [email protected] & [email protected] @xmath101 & [email protected] & [email protected] & [email protected] @xmath105 nominal values , as determined through careful calibration against deposition rate @xcite .
@xmath106 estimated from the disappearance of the microwave signal .
@xmath109 errors report the temperature stability during each measurement .
@xmath111 uncertainties estimated through the sensitivity of the fit quality to each fitting parameter .
in this paper we have presented the first measurements of the frequency dependence of the vortex - state microwave resistivity in s / f / s trilayers , with nb as s layer and pdni as f layer , and on a nb sample .
we have used a wideband technique , the corbino disk technique , capable to obtain reliable measurements in the frequency range 1 - 22 ghz . in nb , significant creep is detected and a previously unobserved switching " behaviour is found approaching the transition line . in trilayers
the experimental data show that , even when the f thickness @xmath2 is as thin as half of the coherence length in the ferromagnet , @xmath0 , the superconducting state changes from the pure superconductor in a way that the vortex state resistivity is affected . by exploiting the wideband technique ,
we have shown that changes in the free - flux - flow resistivity are the most relevant , since one obtains @xmath25 in excess of the bs limit @xmath112 .
in addition , one observes a reduced pinning .
by contrast , the trilayer with @xmath2 = 1 nm does not exhibit any qualitative difference with respect to the nb reference sample .
our observations call for further study of the vortex state resistivity in s / f / s heterostructures , which are presently in progress .
29 ltke - entrup n , plaais b , mathieu p and simon y 1998 _ physica b _ * 255 * 75 ; pautrat a , goupil c , simon c , ltke - entrup n , plaais b , mathieu p , simon y , rykov a and tajima s 2001 _ phys . rev .
b _ * 63 * 054503 | we present wideband microwave measurements ( 1 - 22 ghz ) taken on superconductor / ferromagnet heterostructures with the corbino disk technique .
we apply the technique to nb / pdni / nb trilayers in the vortex state and we compare the results to data taken with the same technique on a nb thin film .
we show that it is possible to directly extract the genuine flux - flow resistivity from our frequency - dependent measurements without overcoming the ( depinning ) critical current .
the characteristic frequency for vortex relaxation can also be estimated , without resorting to a specific model .
we find that the f layer determines a weakening of pinning and an enhancement of the flux - flow resistivity with respect to the nb film . |
the control of the electron spin in nanoscale is an object of great interest in the field of spintronics.@xcite the knowledge of how the spin evolves as the electron goes through a nano - device is a key element for the purpose of using the spin degree of freedom as a mechanism for information transfer and processing .
randomization of the electron spin , however , occurs due to the spin - orbit coupling as the electron wave vector changes during and after scattering processes . between scattering events , the precession frequency of the spin about the local magnetic field
also changes from electron to electron depending on the wave vector . to overcome these detrimental effects ,
many efforts have been made to understand the spin diffusion and the spin relaxation,@xcite in particular after the proposal of the spin - field - effect - transistor ( sfet ) by datta and das more than sixteen years ago.@xcite in such a device , ferromagnetic material is used as source and drain for the injection and the detection of spin - polarized electrons .
the electron spins , moving through a quasi - one - dimensional channel , precess about an effective magnetic field tuned by a gate voltage . in @xmath4-type zinc - blende semiconductors the local magnetic field gives origin to a zeeman - like splitting which , combined with the scattering , can cause a spin relaxation / dephasing known as the dyakonov - perel ( dp ) mechanism.@xcite in gaas , the dresselhaus term@xcite is dominant in the dp wave vector dependent magnetic field , giving rise to a spin splitting due to bulk inversion asymmetry . furthermore in low - dimensional semiconductor structures with asymmetric
confining potential , another term known as the rashba term@xcite contributes to the dp mechanism , giving rise to a spin splitting based on the structure inversion asymmetry . while the rashba term is linear , the dresselhaus term is cubic in the wave vector components .
a few years ago schliemann _ et al . _
@xcite proposed a non - ballistic spin - field - effect transistor based on the competition of the dresselhaus and the rashba terms .
in such a transistor a gate voltage is tuned to give equal strengths to both terms , leading to a very long spin dephasing time for the spin polarization along ( 110 ) direction.@xcite in their work the cubic term is argued to be unimportant .
cheng and wu studied the effect of the cubic term on the spin relaxation time of spins along the ( 110 ) direction by solving the kinetic spin bloch equations , and obtained a finite spin relaxation time.@xcite actually , in spin transport , when spins are injected into a sample , in addition to the direction of spin polarization , the direction of spin diffusion / injection is also important .
the spin relaxation in transport depends on both directions . in the present work ,
the anisotropy of the spin - orbit coupling is taken into account when we study the ideal balance between the rashba and the dresselhaus terms in a quasi - two - dimensional channel of a gaas quantum well .
therefore , both the direction of the spin polarization and the direction of the spin injection are analyzed .
we further show that the cubic term appearing in the dresselhaus effective magnetic field can not be neglected in the competition with the rashba field .
our treatment goes beyond the single particle method in that we use a fully microscopic many - body approach by solving the kinetic spin bloch equations ( ksbe).@xcite as shown by weng and wu , the correlations between the spin - up and spin - down states , _
i.e_. , the off - diagonal terms of the density matrix in spin space , play an essential role in the spin diffusion / transport.@xcite studying the spin polarizations and the oscillations in the direction of diffusion one can calculate the injection length as a function of the direction of injection .
this paper is organized as follows : in sec .
[ so ] we describe the rashba and the dresselhaus spin - orbit couplings and study the effective magnetic field under different injection directions in the quantum well . in sec .
[ ke ] we construct the ksbe and we discuss how they change with the injection direction . in sec .
[ res ] we apply the ksbe to our model and present the results for different parameters .
we conclude in sec .
[ fc ] . ) and ( 001)-axes ] and spin diffusion / injection ( @xmath5-axis).,width=207 ]
the dp mechanism introduces a spin - orbit term in the hamiltonian which can be expressed by a zeeman - like term with a wave - vector - dependent effective magnetic field : @xmath6 for convenience , the effective @xmath7-factor and the bohr magneton are absorbed into the definition of @xmath8 .
for gaas , which lacks an inversion symmetry center , the effective magnetic field contains the dresselhaus term . for electrons in infinite - barrier - height quantum wells with a small well width @xmath9 in the @xmath10 band , this term is:@xcite @xmath11 where @xmath12 represents the spin splitting parameter and @xmath13 .
if a space gradient is applied to the sample , another effect shows up , the rashba effect .
the rashba term in the hamiltonian for electrons in the @xmath10 band is @xmath14 where @xmath15 is a material parameter , and @xmath16 is the electric field determined by the asymmetry.@xcite taking the @xmath17-axis in the ( 001 ) direction ( @xmath17 will be considered throughout this paper as the growth direction of the quantum well ) and @xmath18 , the contribution of the rashba term to the effective magnetic field may be written as @xmath19 where @xmath20 is the rashba parameter .
next , we consider the influence on the effective magnetic field when we change the injection direction to @xmath21 and choose this direction as the @xmath5-axis ( a schematic of the configuration is shown in fig . [ schem ] ) .
the effective magnetic field changes according to @xmath22 , with @xmath23 therefore , we obtain for the dresselhaus field : @xmath24 where the wave vector components are written in the rotated coordinate system . for the rashba field
we obtain : @xmath25 notice that the rashba field is expressed in the new coordinates in the same way as in eq .
( [ dress ] ) . in the following ,
we discuss the diffusion in two different ways : ( i ) we take the identical strengths of @xmath26 with @xmath27 to study the diffusion direction dependence and the spin polarization dependence of the diffusion . in this case
, the spin orbit coupling @xmath28 is written as @xmath29 with the special direction @xmath30 representing the crystal direction @xmath2 in the rotated coordinates .
( ii ) we study the effect of the gate voltage on the spin diffusion .
we treat this issue in sec .
our fully microscopic treatment concerns the calculation of the spin density matrix for the electron with momentum @xmath31 at position @xmath32 : @xmath33 the diagonal elements @xmath34 stand for the electron distribution functions of spin @xmath35 whereas the off - diagonal elements @xmath36 represent the correlations between the spin - up and -down states . by using the non - equilibrium green function method with gradient expansion as well as the generalized kadanoff - baym ansatz,@xcite we construct the ksbe as follows,@xcite assuming the diffusion to take place in the @xmath5-direction : @xmath37 = \hbar \left.\frac{\partial\rho_{\bf k}(x , t)}{\partial t}\right|_{\mathtt{s}}\ .\ ] ] here @xmath38 is the electric potential satisfying the poisson equation @xmath39/(a\kappa_0)\ , \label{possion}\ ] ] with @xmath40 standing for the electron density at position @xmath41 , @xmath42 representing the background positive charge density and @xmath43 being the static dielectric function .
@xmath44 is the hartree - fock term from the coulomb interaction .
the scattering terms @xmath45 include the electron - electron and electron - phonon scattering .
their expressions can be found in ref . .
details of the numerical scheme as well as the material parameters are laid out in ref . .
it has been shown that coulomb scattering plays a very important rule in spin dephasing@xcite and spin diffusion / transport@xcite and therefore can not be neglected in calculating the spin diffusion length .
when the system reaches its steady state , in the absence of an external field , it can be described by the equation @xmath46 = \left.\hbar\frac{\partial\rho_{\mathbf k}(x , t)}{\partial t}\right|_{\mathtt{s}}\ .\end{aligned}\ ] ] by neglecting the hartree - fock and the scattering terms , eq .
( [ eq : beq1 ] ) can be solved analytically.@xcite after dividing both sides of the equation by @xmath47 , we see that along with the diffusion , the spin polarization for each wave vector @xmath31 precesses along @xmath48 .
the fact that spins with different momentums precess with different frequencies is referred to as inhomogeneous broadening.@xcite with any spin - conserving scattering included , the inhomogeneous broadening results in spin dephasing.@xcite for @xmath26 , @xmath49 reads @xmath50 we find that @xmath49 includes two parts : the zeroth - order term ( on @xmath51 ) which is always along the same direction @xmath52 , and the second - order term .
the zeroth - order term includes both the @xmath31-dependent and the @xmath31-independent terms , while the second - order term is always @xmath31-dependent .
according to the previous works,@xcite the @xmath53-dependent term contributes to the inhomogeneous broadening .
we first analyze the spin diffusion without the third - order term of the dresselhaus term ( and hence the second - order term of @xmath54 ) . in this case , the effective magnetic field for each wave vector @xmath31 points to the same direction @xmath55 . from the previous works,@xcite we know that the spin dephasing in the time domain , in the spatial homogeneous case , shows strong anisotropy with respect to the direction of spin polarization : the spin polarization along the direction @xmath55 has an infinite dephasing time , while the ones vertical to this direction have a very short spin dephasing time due to the large spin - orbit coupling .
this is because when the spin polarization direction is the same as the direction of the effective magnetic field , the spin polarization can not precess and hence there is no inhomogeneous broadening .
consequently the spin polarization has an infinite dephasing time .
the same is true for the spin diffusion : when the spin polarization is along the same direction of @xmath49 , _
i.e. _ , @xmath55 ( the crystal direction @xmath2 ) , the spin polarization will not decay _ regardless of the direction of the spin injection _
, even in the presence of scattering , as there is no spin precession and hence no inhomogeneous broadening ; however , when the spin polarization is perpendicular to @xmath55 , the spin diffusion length is very short due to the large inhomogeneous broadening .
nevertheless , in contrast to the spin dephasing in a spatially homogeneous system , the spin diffusion has an extra degree of freedom , _
i.e. _ , the spin diffusion / injection direction ( which should not be confused with the spin polarization direction ) . this degree of freedom introduces an additional level of anisotropy to the spin diffusion / transport . when the spin diffusion / injection direction is along @xmath56 , _
i.e. _ , ( @xmath57 ) direction , the precession frequency @xmath58 becomes @xmath59-independent if the cubic term of the dresselhaus term is neglected .
moreover , even with the scattering included , eq .
( [ eq : beq1 ] ) can also be satisfied by the following solution : @xmath60 where @xmath61 is the fermi distribution with the spin polarization at the left boundary .
we see from eq .
( [ eq : beq2 ] ) that in this special case the oscillation period is the same for all @xmath31 s , with the value @xmath62 .
so , there is no inhomogeneous broadening and the spin has an infinite diffusion length , _ regardless of the direction of the initial spin polarization_. this result is mainly induced by the peculiar precession frequency @xmath63 which is determined by the spin - orbit coupling and the diffusion velocity @xmath47 , so it can not be obtained in the spatially homogeneous case where the precession frequency is only determined by the spin - orbit coupling .
of course , when we include the cubic term in the spin - orbit coupling , the oscillation period becomes wave - vector dependent .
this inhomogeneous broadening , then , results in a finite diffusion length . in brief , we stress the importance of the inhomogeneous broadening caused by the diffusion term as compared to the spin dephasing in the spatially homogeneous case .
consequently the spin polarization shows strong anisotropy not only for the direction of the initial spin polarization but also for the direction of the spin diffusion .
moreover , the cubic dresselhaus term can not be neglected . in what follows
we solve the problem by using the fully microscopic many - body approach of the ksbe with the cubic dresselhaus term and the scattering explicitly included .
( solid curves ) and the inverse of the spin oscillation period @xmath64 ( dashed curves ) for @xmath26 as functions of the injection direction for different spin polarization directions @xmath55 , @xmath65 and @xmath66 at @xmath67 k. it is noted that the scale of the spin oscillation period is on the right hand side of the frame.,height=207 ]
the ksbe are solved numerically for ( 001 ) gaas quantum wells of width @xmath68 nm including the electron - electron and the electron lo - phonon scattering .
the value of the dresselhaus coefficient is as @xmath69 ev@xmath70@xmath71.@xcite the electron density and the temperature are taken as @xmath72 @xmath73 and @xmath67 k separately unless otherwise specified .
all matrix elements of the interactions are given in ref . .
for the spin polarization along direction @xmath74 , the boundary conditions are given by@xcite @xmath75 with the fermi distribution @xmath76^{-1}$ ] ( @xmath77 ) and the chemical potential @xmath78 determined by the polarized electron density .
the diffusion length and the oscillation period are extracted from the spatial evolution of the spin polarization along @xmath79 direction @xmath80 $ ] .
we first fix the spin polarization along the ( 110 ) direction ( @xmath55 ) and study the spin diffusion length as a function of the spin injection direction in the presence of the cubic dresselhaus term . for comparison , we also study the cases when the spin polarizations are perpendicular to the ( 110 ) direction , _
i.e. _ , @xmath65 and @xmath81 and show how the spin injection lengths change as a function of spin diffusion / injection direction .
subsequently we explore the special case when the spin diffusion / injection direction is along ( @xmath57 ) , _ i.e. _ , @xmath56 and show how the spin injection length changes with the spin polarization in the presence of the cubic dresselhaus term . the spin polarization calculated from the ksbe eq .
( [ eq : beq ] ) and the poisson eq .
( [ possion ] ) along these three spin polarizations can be well fitted by @xmath82 where @xmath83 is the spin diffusion length and @xmath84 represents the oscillation period . in fig .
[ fig1 ] the spin diffusion length @xmath83 and the inverse of the spin oscillation period @xmath64 are plotted against the spin diffusion / injection angle @xmath85 for spin polarizations along @xmath86 , @xmath65 and @xmath66 , respectively .
it is interesting to see that with the inclusion of the cubic dresselhaus term , the spin injection length becomes finite but still _ independent _ on the direction of the spin diffusion / injection if the spin polarization is along ( 110 ) ( @xmath87 ) .
this is in agreement with the case without the cubic term where @xmath83 becomes infinite . at the same time
, the spin polarization along this direction has an infinite oscillation period @xmath88 .
this can be understood because the spin polarization is in the same direction as the effective magnetic field given by the zeroth - order term in @xmath49 and , thus , results in no oscillations .
it is noted that the effective magnetic field from the second - order term in @xmath49 is @xmath89-dependent and therefore can not lead to any oscillation at high temperature due to the scattering.@xcite however , this second - order term causes an inhomogeneous broadening which leads to a finite spin diffusion length . by rewriting the second - order term of @xmath49 into @xmath90 with @xmath51 and @xmath91 denoting the magnitude and the direction of the wave vector @xmath31 separately
, one finds that the magnitude of the inhomogeneous broadening does not change with the spin diffusion / injection direction @xmath85 .
consequently the spin diffusion length does not change with the spin diffusion / injection direction .
strong anisotropy is again found for spin polarizations along the directions perpendicular to @xmath86 except for the special case when @xmath56 .
moreover , it is found in the figure that , for the perpendicular directions , both @xmath83 and @xmath84 depend sensitively on the spin diffusion / injection direction ( additional anisotropy ) .
it is further noticed that @xmath83 and @xmath84 are almost identical for both the perpendicular directions @xmath92 and @xmath93 .
these results can be understood again from the inhomogeneous broadening induced by @xmath49 . as @xmath12 in eq .
( [ omega ] ) is much smaller than @xmath94 , the inhomogeneous broadening for spin polarizations @xmath92 and @xmath93 are therefore defined by the zero - th order term of @xmath49 , which is the same for both perpendicular directions and is spin diffusion / injection direction @xmath85 sensitive .
the oscillation period in the diffusion @xmath84 is determined by the @xmath31-independent component of @xmath49,@xcite _ i.e. _ , @xmath95 which is in good agreement with the results in fig .
[ fig1 ] .
now we turn to the case with special fixed spin diffusion / injection direction @xmath56 , and investigate the spin diffusion length with different spin polarizations .
it is interesting to see from fig .
[ fig1 ] that similar to the case without the cubic dresselhaus term , the spin diffusion lengths for the three spin polarizations are almost identical .
therefore the anisotropy disappears for this particular spin diffusion / injection direction .
this is because when the spin diffusion / injection direction is @xmath56 , the @xmath31-dependent component in the zero - th order term of @xmath49 disappears .
this results in the same inhomogeneous broadening for any spin polarization . and spin oscillation period @xmath84 at @xmath56 and @xmath67 k as functions of the electron density .
note the scale of the oscillation period is on the right hand side of the frame.,height=207 ] and spin oscillation period @xmath84 at two different electron densities , @xmath96 @xmath73 and @xmath97 @xmath73 _ vs. _ the temperature .
the spin diffusion / injection direction @xmath56 .
note the scale of the oscillation period is on the right hand side of the frame.,height=207 ] now we investigate the temperature and electron density dependence of the spin diffusion length at the special spin diffusion / injection direction @xmath56 with @xmath26 . as shown in the previous section
the spin diffusion is isotropic with respect to the spin polarization direction .
hence we choose the spin polarization to be along the @xmath17-axis . in fig .
[ fig2 ] the oscillation period and the diffusion length are plotted as functions of the electron density .
one finds that the diffusion length @xmath83 decreases with the density , whereas the oscillation period @xmath84 increases with it .
however , it is noted that the change observed in the oscillation period is almost negligible ( 1 % ) in contrast to the pronounced changes observed in the diffusion length ( of the order of 40 % ) .
we further investigate the temperature dependence of the spin diffusion length and the oscillation period in fig .
[ fig3 ] by taking two typical electron densities , _
i.e. _ , @xmath98 @xmath73 and @xmath96 @xmath73 .
similar behavior is found here compared to the density dependence : the spin diffusion length @xmath83 decreases markedly with the temperature whereas the oscillation period @xmath84 only increases slightly with it .
both dependences above can be again understood by the inhomogeneous broadening . unlike the results in ref .
[ ] where only the dresselhaus term is considered and a very weak temperature dependence is obtained , the diffusion length here has a strong temperature dependence .
this can be understood by the difference in the behavior of the inhomogeneous broadening , which is determined by the anisotropy of the precession frequency @xmath49 . in the previous case ,
the zero - th order term dominates the inhomogeneous broadening which changes little with temperature and electron density .
however , in the present case with @xmath26 and @xmath56 , the @xmath31-dependent zeroth - order term is always zero and the second - order term alone determines the inhomogeneous broadening .
this term increases effectively with the temperature and with the electron density due to the increase of the average value of @xmath99 .
this is the reason for the obtained marked decrease of the spin diffusion length .
the oscillation period @xmath84 is determined by the @xmath31-independent zeroth - order term of @xmath49 , which is independent of the electron density and the temperature .
therefore one observes only a slight change in the oscillation period , originating from the second - order term and the scattering .
( solid curves ) and spin oscillation period @xmath84 ( dashed curves ) as functions of the ratio @xmath100 with the injection direction @xmath56 for different spin polarization directions @xmath55 , @xmath65 and @xmath66 at @xmath67 k. the chain line is used to guide the eyes .
the scale of the oscillation period is on the right hand side of the frame.,height=207 ] we now tune the gate voltage in order to change the relative importance of the dresselhaus and the rashba terms .
the spin diffusion / injection direction is again fixed at @xmath56 .
the calculated spin diffusion length @xmath83 and the spin oscillation length @xmath84 are plotted against @xmath100 in fig .
[ fig4 ] at @xmath67 k , with the spin polarizations being along @xmath86 , @xmath65 and @xmath66 respectively .
it is noted that since @xmath88 when the spin polarization is along @xmath86 [ ( 110 ) ] , we only plot @xmath84 for the two perpendicular directions in the figure .
three main results are obtained : ( i ) differing from the case without the cubic dresselhaus term , the maximum diffusion length occurs interestingly at @xmath101 ( instead of 1 ) for each spin polarization direction ; ( ii ) the isotropy of the spin polarization direction at @xmath102 is suppressed except at @xmath103 , and the diffusion lengths for the spin polarization along @xmath65 and @xmath93 are always identical ; ( iii ) the spin oscillation period @xmath84 decreases with @xmath104 .
these results can also be understood from the inhomogeneous broadening .
for @xmath56 , the spin precession frequency along the diffusion direction is given by @xmath105\hat{\mathbf n}_1 \nonumber \\ & & \mbox { } - [ \beta+\alpha - \frac{\gamma}{2}(k_x^2-k_y^2)]\hat{\mathbf n}_0\ .
\label{3piov4}\end{aligned}\ ] ] at @xmath67 k , @xmath106 with @xmath107 representing the average over the imbalance of the spin - up and -down electrons .
therefore , @xmath108 is almost zero at @xmath101 and does not contribute to the inhomogeneous broadening .
this results in a maximum of the spin diffusion length .
it is further noted that the inhomogeneous broadening term @xmath109 gives a larger effect when the spin polarization is along @xmath65 and @xmath66 than that when the spin polarization is along @xmath55 as the latter coincides with the precession axis .
this explains why one obtains a larger diffusion length when the spin polarization is along @xmath86 at @xmath101 .
finally as the spin oscillation period @xmath84 is determined by the magnitude of the @xmath89-independent term of @xmath49 , _
i.e. _ , @xmath110 , which increases with @xmath100 .
therefore , @xmath84 decreases with @xmath100 .
in conclusion , we study the spin diffusion in ( 001 ) gaas quantum wells with competing dresselhaus and rashba spin - orbit coupling strengthes by solving the ksbe with the electron - phonon and the electron - electron coulomb scattering explicitly included .
it is shown that unlike the spin dephasing in the time domain where strong anisotropy is determined by different spin polarization , for spin diffusion it is also determined by the spin diffusion / injection direction . by neglecting the cubic term of the dresselhaus spin - orbit coupling and with @xmath26 , the ideal case of an infinite diffusion length is obtained _ either _ for the spin polarization along @xmath2 regardless of the spin diffusion direction _ or _ for the spin diffusion direction along @xmath111 regardless of the spin polarization direction .
however , the cubic term can not be neglected , resulting in a finite diffusion length , which in fact is small , about 2 @xmath112 m at the temperature of 200 k when the electron density lies between @xmath97 @xmath73 and @xmath96 @xmath73 .
it is then shown that when @xmath26 , for the spin polarization along @xmath2 , the spin diffusion length changes little for the spin diffusion direction , whereas for the other two perpendicular spin polarization directions @xmath111 and @xmath113 , the spin diffusion length shows strong anisotropy for the spin diffusion / injection direction and has a peak when the spin diffusion / injection direction is along @xmath1 .
when the spin diffusion / injection direction is along @xmath1 , the spin diffusion length is isotropic for spin polarization .
the electron density and temperature dependence of the spin diffusion is closely investigated .
we also tune the gate voltage to show the competing effect of the dresselhaus and the rashba strengthes . it is found that with the cubic dresselhaus term included
, the maximum spin diffusion length appears at @xmath101 instead of 1 where the cubic term is neglected .
the spin diffusion length decreases when the ratio of the rashba and dresselhaus strengths deviates from 0.95 .
these results can be well understood by the inhomogeneous broadening .
this work was supported by the natural science foundation of china under grant no .
10574120 , the national basic research program of china under grant no .
2006cb922205 , the knowledge innovation project of chinese academy of sciences and srfdp .
one of the authors ( i.c.c.l . )
was partially supported by an esn from cnpq from brazil .
we would like to thank d. csontos for his critical reading of this manuscript .
m. w. wu , m. q. weng , and j. l. cheng , to be published in _ physics , chemistry and application of nanostructures _
( world scientific , singapore , 2007 ) ; cond - mat/0611649 , and references therein .
m. q. weng and m. w. wu , phys .
b * 66 * , 235109 ( 2002 ) ; j. appl . phys . *
93 * , 410 ( 2003 ) .
m. q. weng , m. w. wu , and q. w. shi , phys .
b * 69 * , 125310 ( 2004 ) .
m. w. wu and h. metiu , phys .
b * 61 * , 2945 ( 2000 ) ; m. q. weng and m. w. wu , _ ibid . _ * 68 * , 075312 ( 2003 ) .
m. q. weng , m. w. wu , and l. jiang , phys .
b * 69 * , 245320 ( 2004 ) .
n. s. averkiev , l. e. golub , and m. willander , j. phys . : condens .
matter * 14 * , r271 ( 2002 ) .
j. schliemann , j. c. egues , and d. loss , phys .
lett . * 90 * , 146801 ( 2003 ) . j. schliemann and d. loss , phys .
b * 68 * , 165311 ( 2003 ) .
s. i. erlingsson , j. schliemann , and d. loss , phys .
b * 71 * , 035319 ( 2005 ) .
r. winkler , phys .
b , * 69 * , 045317 ( 2004 ) . c. lechner and u. rssler , phys . rev .
b * 72 * , 045311 ( 2005 ) .
x. cartoix , d. z. -y .
ting , and y. -c .
chang , phys .
b * 71 * , 045313 ( 2005 ) .
e. shafir , m. shen , and s. saikin , phys . rev .
b * 70 * , 241302(r ) ( 2004 ) . s. datta and b. das , appl
. phys . lett . *
56 * , 665 ( 1990 ) . m. i. dyakonov and v. i. perel , zh .
* 60 * , 1954 ( 1971 ) [ sov .
jetp * 33 * , 1053 ( 1971 ) ] ; m. i. dyakonov and v. i. perel , fiz .
tela * 13 * , 3581 ( 1971 ) [ sov .
solid state * 13 * , 3023 ( 1972 ) ] ; m. i. dyakonov and v. yu .
kachorovskii , fiz .
. poluprovodn . *
20 * , 178 ( 1986 ) .
r. winkler , _ spin - orbit couplig effects in two - dimensional electron and hole systems _ ( springer , berlin , 2003 ) .
h. haug and a.p .
jauho , _ quantum kinetics in transport and optics of semiconductors _
( springer , berlin , 1996 ) .
j. l. cheng and m. w. wu , j. appl .
* 101 * , 2007 , in press ; cond - mat/0608413 . | aiming at the optimization of the spin diffusion length in ( 001 ) gaas quantum wells , we explore the effect of the anisotropy of the spin - orbit coupling on the competition between the rashba and the dresselhaus spin - orbit couplings by solving the kinetic spin bloch equations with the electron - phonon and the electron - electron scattering explicitly included . for identical strengths of the rashba and the dresselhaus spin - orbit couplings , the spin diffusion length shows strong anisotropy not only for the spin polarization direction but also for the spin diffusion direction .
two special directions are used seeking for the large diffusion length : ( @xmath010 ) and ( 110 ) . without the cubic term of the dresselhaus spin - orbit coupling and with the identical dresselhaus and rashba strengths , infinite diffusion lengths can be obtained _ either _ for the spin diffusion / injection direction along @xmath1 , regardless of the direction of spin polarization , _ or _ for the spin polarization along @xmath2 , regardless of the direction of the spin diffusion / injection . however ,
the cubic dresselhaus term can not be neglected , resulting in a finite spin diffusion length which decreases with the temperature and the electron density .
the anisotropy for the spin diffusion direction and spin polarization direction is maintained .
for the spin diffusion / injection direction along ( @xmath310 ) , the spin diffusion length increases first with the increase of the rashba strength ( from 0 ) which can be tuned by the external gate voltage ; when the rashba strength is slightly smaller than ( in stead of equal to ) the dresselhaus strength , the diffusion length reaches its maximum , followed by a decrease with further increase of the rashba strength . |
a caricature of markets , from the point of view of financial engineering , is the following single period asset pricing framework @xcite : there are only two times @xmath0 ( today ) and @xmath1 ( tomorrow ) .
the world at @xmath2 can be in any of @xmath3 states and let @xmath4 be the probability that state @xmath5 occurs .
there are @xmath6 risky assets whose price is one at @xmath0 and is @xmath7 at @xmath2 , @xmath8 if state @xmath9 materializes .
there is also a risk - less asset ( bond ) which also costs one today and pays one tomorrow , in all states .
prices of assets are assumed given at the outset .
portfolios of assets can be built in order to transfer wealth from one state to the other .
a portfolio @xmath10 is a linear combination with weights @xmath11 , @xmath12 on the riskless and risky assets .
the value of the portfolio at @xmath0 is @xmath13 which is the price the investor has to pay to buy @xmath14 at @xmath0 .
the return of the portfolio , i.e. the difference between its value at @xmath2 and at @xmath0 , is given by @xmath15 the construction of arbitrage pricing theory relies on the following steps summarizing , the logic of financial engineering is : assuming that markets are arbitrage free , the price of any contingent claim , no matter how exotic , can be computed .
this involves some consideration of risk , as long as markets are incomplete .
but if one can assumes that markets are complete , then prices can be computed in a manner which is completely independent of risk .
note indeed that the probability distribution over @xmath9 plays no role at all in the above construction .
it is also worth to remark that asset return @xmath16 do not depend , by construction , on the type of portfolios which are traded in the market . a complete
, arbitrage free market is the best of all possible worlds .
when markets expand in both complexity and volumes , one is tempted to argue that this is indeed a good approximation of real financial markets .
why are markets arbitrage free ? because , according to the standard folklore , otherwise `` everybody would jump in [ ... ] affecting the prices of the security '' @xcite .
now let us assume that prices are affected not only when `` everybody jumps in '' but anytime someone trades , even though the effect is very small , for individual trades . in the simplified picture of the market we shall discuss below
, prices depend on the balance between demand and supply ; if demand is higher than supply return is positive , otherwise it is negative .
demand comes from either an exogenous state contingent process or is generated from the derivative market in order to match contracts . here
derivatives are simply contracts which deliver a particular amount of asset in a given state ( e.g. an umbrella if it rains , nothing otherwise ) .
even if markets are severely incomplete , financial institutions which we shall call banks for short in what follows will match the demand for a particular derivative contract if that turns out to be profitable , i.e. if the revenue they extract from it exceeds a risk premium . competition in the financial industry brings on one side to smaller risk premia and on the other to a wider diversity of financial instruments being marketed in the stock market .
these two effects are clearly related because the increase in financial complexity measured by the number of different financial instruments makes the market less incomplete and hence it reduces the spread in the prices of derivatives .
when financial complexity is large enough the market becomes complete because each claim can be replicated by a portfolio of bond and derivatives . beyond this point
, there is an unique value for the price of each derivative , which is the one computed from apt .
though the story is different , the conclusion seems to be the same : efficient , arbitrage - free , complete markets .
the holy grail of financial engineering . actually , the road to efficient , arbitrage - free
, complete markets can be plagued by singularities which arise upon increasing financial complexity .
we shall illustrate this point within the stylized one period framework discussed above .
take the one period framework as above , and let @xmath4 be the probability that state @xmath5 materializes .
imagine there is a single risky asset and a risk - less asset .
again we shall implicitly take discounted processes , so we set the return of the risk - less security to zero .
the price of the risky asset is @xmath1 at @xmath0 and @xmath27 , in state @xmath9 .
however , rather than defining at the outset the return of the asset in each state , we assume it is fixed by the law of demand and supply .
banks develop and issue financial instruments . in this simplified world , a financial instrument is a pair @xmath28 where @xmath29 is the cost payed to the bank at @xmath0 by the investor and @xmath30 is the amount of risky asset it delivers in state @xmath9 at @xmath2 .
we imagine there is a demand @xmath31 units of derivative @xmath32 if the price is less than @xmath33 and zero otherwise . ] for each of @xmath34 possible financial instruments @xmath35 , @xmath36 .
the return of the asset at @xmath2 in each state @xmath9 is given by @xmath37 where @xmath38 is assumed to arise from investors excess demand whereas the second term is generated by banks hedging derivative contracts ) .
the most transparent one , for our purposes , is that of assuming a finite liquidity of the underlying asset in a market maker setting such as that of ref . @xcite . between the time ( @xmath0 ) when derivatives are signed and the maturity ( @xmath2 ) , banks have to accumulate a position @xmath39 in the underlying for each contract @xmath32 they have sold ( @xmath40 ) . at the same time
, an exogenous excess demand @xmath38 is generated by e.g. fundamental trading or other unspecified investment .
the coefficient between returns and excess demand
e.g. the market depth is assumed to be one here , for simplicity of notation .
market clearing between demand and supply can also be invoked to motivate eq .
( [ return ] ) . though that conforms more closely with the paradigms of mathematical economics , it introduces unnecessary conceptual complication within the stylized picture of the market given here .
likewise , we impose absence of arbitrages in the market by assuming that there exist an equivalent martingale measure @xmath41 such that @xmath42=0 $ ] . for the results presented below ,
this is an irrelevant assumption , so we shall not insist more on this . ] .
the supply @xmath43 of instrument @xmath32 is fixed by banks in order to maximize expected profit .
the expected profit for a unit supply ( @xmath40 ) is given by : @xmath44 notice that , even though @xmath45 depends on the supply @xmath43 of derivative @xmath32 , by eq .
( [ return ] ) , banks in a competitive market neglect this dependence and take prices ( i.e. returns ) as given .
rather , banks will sell derivatives if the expected profit is large enough , and will not sell it otherwise .
specifically , we assume that banks match the demand of investors ( i.e. @xmath46 ) if @xmath47 , whereas @xmath48 if @xmath49 .
when @xmath50 then the supply can take any value @xmath51 $ ] . here
@xmath52 can be interpreted as a risk premium related to the risk aversion of banks . indeed ,
as long as the market is incomplete , there is not an unique price for derivatives .
even if the prices @xmath33 were fixed on the basis of an equivalent martingale measure @xmath18 , as in section [ dynpricing ] , it is not possible , for banks , to eliminate risk altogether .
financial markets are quite complex , with all sorts of complicated financial instruments .
this situation is reproduced in our framework by assuming that demand @xmath38 and derivatives @xmath39 are drawn independently at random .
furthermore we take the limit @xmath53 .
this situation clearly defies analytical approaches for a specific realization .
remarkably , however , it is possible to characterize the statistical behavior of typical realizations of such large complex markets . in order to do that ,
some comments on the scaling of different quantities , with the number of states @xmath3 , is in order .
the interesting region is the one where the number of derivatives @xmath34 is of the same order of the number of states @xmath3 ( @xmath54 ) . in this regime we expect the market to approach completeness . for this reason we introduce the variable @xmath55 . assuming that @xmath45 be a finite random variable in the limit @xmath56 , requires @xmath38 to be a finite random variable
e.g. normal with mean @xmath57 and variance @xmath58 .
likewise , we shall take @xmath39 as random variables with zero average and variance @xmath59 .
indeed , the second term of eq .
( [ return ] ) with @xmath60 is of the order of @xmath61 , which is finite in the limit we are considering to be of order one , but introduce a coefficient @xmath62 in eq .
( [ return ] ) as a finite market depth . ] .
it is convenient , in the following discussion , to introduce the parameter @xmath63.\ ] ] where @xmath64 is the expected price of instrument @xmath32 , at @xmath0 .
the dependence on @xmath3 in the equation above is motivated by the fact that the variance of @xmath65 and of the second term in eq .
( [ utility ] ) , is of order @xmath66 .
this implies that the relevant scale for the r.h.s . of eq .
( [ epsilon ] ) is of order @xmath59 .
the parameter @xmath67 encodes the risk premium which banks for selling derivative @xmath32 . in the next section ,
we take the simplifying assumption that @xmath68 does not depend on @xmath32 , in order to illustrate the generic behavior of the model . in the next section
, we shall discuss the case where @xmath67 depends on @xmath32 .
actually , @xmath67 encodes the risk premium that banks require for trading derivatives and hence should depend on the volatility of the realized returns @xmath45 and on the degree of market completeness . while keeping this in mind
, we will consider @xmath67 as an independent parameter with the idea that it decreases from positive values when the volatility of the market decreases or on the route to market completeness .
we shall return to this point below .
we consider , in this section , the situation where banks chose the supply @xmath43 so as to maximize their profit , considering returns @xmath45 as given .
the statistical properties corresponding to this state the so - called competitive equilibrium can be related to those of the minima of a global function @xmath69 , by the following statement : _
the competitive market equilibrium is given by the solution of the minimization of the function _
@xmath70 _ over the variables @xmath71 , where @xmath45 is given in terms of @xmath43 by eq .
( [ return ] ) .
_ notice that , if @xmath72 is negligible , this results states that a consequence of banks maximizing their utility , is that return s volatility the first term of eq.([h ] ) is reduced .
actually , with @xmath73 , only those derivatives which decrease volatility are traded ( @xmath74 ) .
this is an `` unintended consequence '' of banks profit seeking behavior , i.e. a feature which emerges without agents aiming at it .
a proof of the statement above follows by direct inspection of the first order conditions and the observation that @xmath75 .
we first observe that @xmath69 as a function of any of the @xmath43 is a convex function , @xmath76 hence it has a single minimum , for @xmath71 .
imagine @xmath77 to be the minimum of @xmath69 . if @xmath78 it must be that @xmath79 if @xmath49 , it is not convenient for banks to sell derivative @xmath32 , which is consistent with zero supply ( @xmath78 ) .
likewise , if @xmath80 the first order condition yields @xmath47 , which is the condition under which banks should sell as many derivatives as possible . if instead @xmath81 , then @xmath50 which is consistent with perfect competition among banks . from a purely mathematical point of view
, we notice that @xmath69 is a quadratic form which depends on @xmath34 variables @xmath82 , through the @xmath3 linear combinations given by the returns @xmath45 .
it is intuitively clear that , as @xmath34 increases , the minimum of @xmath69 becomes more and more shallow and , for large @xmath34 , it is likely that there will be directions in the space of @xmath82 ( i.e. linear combinations of the @xmath43 ) along which @xmath69 is almost flat .
such quasi - degeneracy of the equilibria corresponds to the emergence of symmetries discussed in the introduction .
the location of the equilibrium is likely to be very sensitive to perturbation along these directions .
these qualitative conclusions can be put on firmer grounds by the theoretical approach which we discuss next . )
^2\right]$ ] in competitive equilibria ( full lines ) , as a function of @xmath55 , for different values of @xmath72 .
points refer to the variance of @xmath83 computed in numerical simulations of a system with @xmath84 ( @xmath85).,width=340 ] a full characterization of the solution of the minimization of @xmath69 , in the limit @xmath56 with @xmath86 fixed , can be derived with tools of statistical mechanics of disordered systems ( see appendix ) , following similar lines of those in ref .
@xcite . indeed , mathematically the model is quite similar to the grand canonical minority game studied in ref .
the solution can be summarized in the following `` representative '' derivative problem : given a normal random variable @xmath87 with mean zero and unit variance , the supply of the `` representative '' derivative is given by @xmath88 where @xmath89 \label{sp1}\\ \chi & = & \frac{ne_z[s_zz ] } { \sqrt{g+\delta}-ne_z[s_zz]}\label{sp2}\end{aligned}\ ] ] are determined self - consistently in terms of expected values @xmath90 $ ] over functions of the variable @xmath87 . loosely speaking
, @xmath87 embodies the interaction through the market of the `` representative '' derivative with all other derivatives .
any quantity , such as the average supply of derivatives @xmath91\ ] ] can be computed from the solution .
figure [ figh ] plots the volatility @xmath92)^2\right]$ ] as a function of @xmath93 for small values of @xmath72 .
as it can be seen , as the market grows in financial complexity , fluctuations in returns of the risky asset decrease and , beyond a value @xmath94 , they become very small ( of order @xmath95 , for small @xmath72 ) .
the expected return @xmath96=\bar d/(1+\chi)$ ] also decreases , keeping a bounded sharpe ratio @xmath96/\sqrt{\sigma}=\bar d/\sqrt{g+\delta}$ ] .
this situation in the region @xmath97 and @xmath98 resembles closely the picture of an efficient , arbitrage free complete market .
unfortunately this is also the locus of a sharp discontinuity , as shown in fig .
this plots @xmath99 as a function of @xmath93 for different values of @xmath72 .
as it can be seen , as the market grows in financial complexity , it passes from a regime ( @xmath100 ) where its behavior is continuous and smooth with @xmath72 to one ( @xmath97 ) which features a sharp discontinuity at @xmath101 is not the actual value of derivatives traded , but the number of derivatives for which there is a demand . the number of derivatives with @xmath74 , is well approximated by @xmath102 , which is less than @xmath3 , for @xmath73 . ] .
in particular , the discontinuity manifests clearly in the limit @xmath103 when the demand of investors is unbounded . then while for @xmath73 the average supply @xmath99 is finite , for @xmath104 and @xmath105 the supply @xmath106 diverges as @xmath103 . in other words , while in one region ( @xmath73 and @xmath104 for @xmath107 ) the equilibrium is controlled by the supply side , in the other region ( @xmath104 , @xmath105 ) the equilibrium is dominated by the demand side .
this distinction , as we shall see in the next section , applies to a more general model where @xmath67 depends on @xmath32 . in competitive equilibria ( full lines ) for different values of @xmath72 .
points refer to the variance of @xmath83 computed in numerical simulations of a system with @xmath84 ( @xmath85).,width=340 ] the consequences of the singularity at @xmath97 and @xmath101 are not entirely evident in the competitive equilibrium setting .
indeed they manifest strongly in response functions : the market behavior close to the singularity is quite sensitive to small perturbations . for example
a small change in the risk perception of banks ( i.e. in @xmath72 ) can provoke a dramatic change in the volume of trading in the derivative market .
the effects of this increased susceptibility appears dramatically in the case where the interaction is repeated over time and banks learn and adapt to investors demand .
this setting not only allows us to understand under what conditions will banks learn to converge to the competitive equilibrium , but it also sheds light on the emergent fluctuation phenomena .
let us assume that the context outlined above is repeated for many periods ( e.g. days ) , indexed by @xmath108 .
let @xmath109 be the state of the market at time @xmath110 and assume this is drawn independently from the distribution @xmath4 in each period .
accordingly , the returns @xmath111 are still determined by eq .
( [ return ] ) , with @xmath112 and @xmath113 , the supply of instruments of type @xmath32 in period @xmath110 . in order to determine the latter
, banks estimate the profitability of instrument @xmath32 on historical data .
they assign a score ( or attraction ) @xmath114 to each instrument @xmath32 , which they update in the following manner : @xmath115 notice that , by eq .
( [ utility ] ) and ( [ epsilon ] ) , @xmath114 increases ( decreases ) if @xmath47 ( @xmath49 ) .
banks supply instrument @xmath32 according to the simple rule @xmath116 therefore , if instrument @xmath32 provides an expected utility larger than the margin @xmath52 , its score will increase and the bank will sell it more likely .
conversely , an instrument with @xmath117 , on average , has a decreasing score and it will not be offered by banks . it can be checked that , in the stationary state , the average probability @xmath118 that banks issue derivative @xmath32 is again given by the configuration @xmath82 which minimizes @xmath69 in eq .
( [ h ] ) .
indeed , figs .
( [ figh],[figs ] ) show that averages in the stationary state of the process nicely follow the theoretical results derived above .
$ ] in numerical simulations of a system with @xmath84 for different values of @xmath72 ( points ) .
lines refer to the theoretical prediction in the approximation of independent variables @xmath119 .
inset : total volatility @xmath120 in numerical simulations for @xmath121 ( @xmath122 for @xmath123 and @xmath124 for @xmath125 ) and @xmath126 ( @xmath127 for @xmath128 and @xmath129 for @xmath84 ) as a function of @xmath72.,width=340 ] at odds with the competitive equilibrium setting , in the present case the volatility of returns also acquires a contribution from the fluctuations of the variables @xmath119 , which is induced by the random choice of the state @xmath109 .
hence , we can distinguish two sources of fluctuations in returns @xmath130 the first depending on the stochastic realization of the state @xmath109 , the second from fluctuations in the learning dynamics . here
@xmath83 is the average return of the asset in state @xmath9 , and it can be shown that it converges to its competitive equilibrium value , @xmath131 .
indeed the contribution of @xmath83 to the volatility @xmath132)^2\right]$ ] closely follows the theoretical curve in fig .
( [ figh ] ) . the dynamical contribution to the volatility @xmath133 $ ] instead shows a singular behavior which reflects the discontinuity of @xmath99 across the line @xmath101 for @xmath97 .
our theoretical approach also allows us to estimate this contribution to the fluctuations under the assumption that the variables @xmath119 are statistically independent . as fig .
[ figv ] shows , this theory reproduces remarkably well the results of numerical simulations for @xmath100 but it fails dramatically for @xmath97 in the region close to @xmath101 .
the same effect arises in minority games @xcite , where its origin has been traced back to the assumption of independence of the variables @xmath119 .
this suggests that in this region , the supplies of different derivatives are strongly correlated .
this effect has a purely dynamical origin and it is reminiscent of the emergence of persistent correlations arising from trading in the single asset model of ref .
@xcite .
we have considered @xmath72 as a fixed parameter up to now . actually , in a more refined model , @xmath67 should depend on @xmath32 and it should be fixed endogenously in terms of the incentives of investors and banks . at the level of the discussion given thus far , it is sufficient to say that the scale of incentives of banks and investors , and hence of @xmath72 , is fixed by the average return @xmath134 or by the volatility @xmath135 .
both these quantities decrease as @xmath93 increases , so it is reasonable to conclude that @xmath72 should be a decreasing function of @xmath93 , in any model where it is fixed endogenously . in other words , as the financial complexity ( @xmath93 ) increases , the market should follow a trajectory in the parameter space @xmath136 , which approaches the critical line @xmath97 , @xmath101 . in the next section ,
we address this problem in an approximate manner , extending our analysis to the case where @xmath67 depends on @xmath32 .
the model considered above can be improved in many ways . for example , the price of derivatives can be determined according to apt , thus making the market arbitrage free both in the underlying and in derivatives . in this section
we will explore this direction in the simplest possible case where the equivalent martingale measure is given .
in particular we will show that this dynamical pricing mechanism is somehow equivalent to an asset dependent risk premia framework . in order to get some insights
, we will then generalize the model considered in the previous sections to the case of a quenched distribution of risk premia .
this assumes that derivative prices change over much longer time - scales than those of trading . in the previous sections prices of derivatives
were fixed by the exogenous parameter @xmath72 and no role was explicitly played by the risk neutral measure @xmath137 . in this section
we consider the case where prices of derivatives are directly computed according to apt .
the framework is the same as before , but prices are computed through the risk neutral measure @xmath138 as @xmath139=\sum_\omega q^\omega a_i^\omega(1+r^\omega)\ ] ] instead of being fixed by the external parameter @xmath72 . as before ,
the emm @xmath137 is fixed exogenously .
again the return of the risky asset is assumed to depend both on an exogenous demand and on the demand induced by the trading of derivatives @xmath140 the supply of derivatives @xmath141 $ ] is then fixed by banks in order to maximize the utility function @xmath142= \sum_\omega ( q^\omega -\pi^\omega ) a_i^{\omega}(1+r^\omega).\ ] ] as in eq .
( [ utility ] ) , the first term is the price of the derivative they cash from clients and the second is the expected cost of buying the stock .
again , banks will sell derivative @xmath32 only if @xmath143 , where @xmath144 is a risk premium .
the analog of eq .
( [ epsilon ] ) then becomes @xmath145 the appearance of asset - dependent @xmath67 calls for an in - depth comparison with the simplified version studied in the previous sections .
we performed numerical simulations using the same learning scheme as before ( eq . [ dynu ] ) , in order to detect typical features of the distribution of @xmath146 .
we observe that the width of the distribution increases with @xmath93 .
a plot of the variance @xmath147 of the distribution as a function of @xmath93 is shown in figure [ eps_eff ] .
notice that , for @xmath93 large enough , @xmath148 .
figure ( [ volb ] ) shows the volatility @xmath149 as a function of @xmath93 .
as we can see , also the volatility increases as the square of the market complexity @xmath150 , so that @xmath151 . as apt naturally leads to the appearence of a distribution of asset risk premia ,
it is instructive to generalize the above picture to the case of _ quenched _ asset dependent risk premia , where a full analytical solution can be obtained . in this section ,
we consider the case where the @xmath146are taken as quenched asset - dependent random variables drawn from a gaussian distribution with mean @xmath152 and variance @xmath147 . in analogy with the previous treatment ,
the competitive equilibrium is equivalent to finding the ground state of the hamiltonian @xmath153 as before , the solution can be summarized in a representative derivative problem . given a normal random variable @xmath87 with @xmath154 mean and unit variance ,
the supply of the representative derivative is given by @xmath155 where @xmath156,\ ] ] @xmath157(1+\chi)}{\sqrt{g+\delta+\sigma_{\epsilon}^2(1+\chi)^2}}\ ] ] are determined self - consistently .
the main features of this generalized problem are as follows : 1 . as for the homogeneous case
, the fluctuations of returns eventually become very small as the market complexity @xmath93 increases ( see figure ( [ fig : vol ] ) where we plot the volatility of returns @xmath158
^ 2]$ ] ) .
however , we can observe that the value @xmath159 beyond which the volatility approaches zero depends on the width of the risk premium distribution , and it increases with @xmath160 .
the sharp transition previously observed in the behavior of the average supply for large @xmath93 becomes smooth as soon as the risk premium distribution has a finite width ( see figure ( [ fig : eps1 ] ) ) .
the sharpness of the crossover in the behavior of @xmath161 as a function of @xmath162 is enhanced as the demand @xmath31 for derivatives increases .
this signals a transition from a supply limited equilibrium , where the main determinant of the supply @xmath43 of derivatives is banks profits , to a demand limited equilibrium , where @xmath43 is mostly limited by the finiteness of the investors demand . in order to make this point explicit ,
it is instructive to investigate the case of unbounded supply ( @xmath103 ) , because the transition becomes sharp in this limit .
the representative derivative , in this case , is described by @xmath163 and , as before , @xmath156,\ ] ] @xmath157(1+\chi)}{\sqrt{g+\delta+\sigma_{\epsilon}^2 ( 1+\chi)^2}}.\ ] ] in contrast with the bounded case and similarly to the case @xmath164 , the system displays a phase separation in the @xmath165 plane between a region in which the average supply remains finite and a region in which the volume of the traded assets diverges ( see left panel of figure [ fig : phase_space ] ) .
introducing a distribution of risk premia with variance @xmath166 has then very different effects depending on whether the supply is bounded or not . in the first case
@xmath166 acts as a regularizer preventing the occurrence of a sharp phase transition . in the second case ,
namely @xmath103 , @xmath167 entails a deformation of the critical line that tends to flatten along the @xmath168 line as @xmath166 grows ( see right panel of figure [ fig : phase_space ] ) .
this characterization allows to get some insights on the case in which prices are dynamically generated .
+ as we showed before , in such a situation the system is characterized by a distribution of effective risk premia which becomes broader as market increases in complexity . hence , upon increasing @xmath93 , one expects a transition from a situation where volumes are limited by the profitability of derivative trading , to one where the demand is saturated .
we presented here a very stylized description of an arbitrage free market , showing that also whithin the ideal context of apt the uncontrolled proliferation of financial intruments can lead to large fluctuations and instability in the market .
interestingly , within our simplified model , the proliferation of financial instruments drives the market through a transition to a state where volumes of trading rapidly expand and saturate investors demand . in order for these kind of models to be more realistic ,
some improvements certainly are needed .
for instance , the demand for derivatives can be derived from a state dependent utility function for consumption at @xmath2 .
this could also be a way to obtain an endogenous risk neutral measure , one of the main problems in the picture of dynamical prices considered above being that the risk neutral measure was fixed exogenously .
we believe that , while accounting for these effects can make the model more appealing , the collective behavior of the model will not significantly change . indeed , the qualitative behavior discussed above is typical of a broad class of models @xcite and mainly depends on the proliferation of degeneracies in the equilibria of the model .
the focus of the present paper is on theoretical concepts .
their relevance for real markets requires quantitative estimates of the parameters .
given the abstract nature of the model , this appears to be a non - trivial task which is an interesting avenue of future research .
it has been recently suggested that market stability appears to have the properties of a public good @xcite .
a public good is a good _ i ) _ whose consumption by one individual does not reduce its availability for consumption by others ( non - rivalry ) and _ ii ) _ such that no one can be effectively excluded from using the good ( non - excludability ) . at the level of the present stylized description
, the expansion in the repertoire of traded assets introduces an externality which drives the market to unstable states .
this suggests that systemic instability may be prevented by the introduction of a tax on derivative markets , such as that advocated long ago for foreign exchange markets by tobin @xcite , or by the introduction of `` trading permits '' , similar to those adopted to limit carbon emissions @xcite .
the stabilizing effect of a tobin tax has already been shown within a model of a dynamic market which is mathematically equivalent to the one presented here @xcite . in summary
, this paper suggests that the ideal view of the markets on which financial engineering is based is not compatible with market stability . the proliferation of financial instruments makes the market look more and more similar to an ideal arbitrage - free , efficient and complete market .
but this occurs at the expense of market stability .
this is reminiscent of the instability discussed long ago by sir robert may @xcite which develops in ecosystems upon increasing bio - diversity . for ecologies
this result is only apparently paradoxical .
indeed the species which populate an ecosystem can hardly be thought of as being drawn at random , but are rather subject to natural selection .
indeed , on evolutionary time scales stability can be reconciled with bio - diversity , as shown e.g. in ref .
the diversity in the ecosystem of financial instruments has , by contrast , been increasing at a rate much faster than that at which selective forces likely operate .
in contrast with the axiomatic equilibrium picture of apt , on which financial engineering is based , the model discussed here provides a coherent , though stylized , picture of a financial market as a system with interacting units . in this picture ,
concepts such as no - arbitrage , perfect competition , market efficiency or completeness arise as emergent properties of the aggregate behavior , rather than being postulated from the outset .
we believe that such an approach can potentially shed light on the causes of and conditions under which liquidity crises , arbitrages and market crashes occur .
mm gratefully acknowledge discussions with j .- p .
bouchaud , d. challet , a. consiglio , m. dacorogna , i. kondor and m. virasoro .
interaction with c. hommes , specially in relation to ref .
@xcite , has been very inspiring .
this work was supported by the complexmarkets e.u
. strep project 516446 under fp6 - 2003-nest - path-1 .
the problem is then the one of finding the ground state of the hamiltonian @xmath169 is carried out in the usual manner .
let s write down the partition function @xmath170 where we have used the shorthand @xmath171 to indicate integrals on the variables in the index .
here it is understood that all variables @xmath43 are integrated from @xmath154 to @xmath1 and all variables @xmath45 and @xmath172 are integrated over all the real axis , with a factor @xmath173 for each @xmath9 .
the next step is to replicate this , i.e. to write @xmath174-i \sum_i a_i^\omega \sum_a s_{i , a}\xi^\omega_a } \\ & = & \int \!\frac{d\vec{u}}{2\pi } { \rm tr}_{\vec{s},\vec{\xi } } e^{-\beta\sum_{i , a}g(s_{i , a})}\prod_{\omega = 1}^\omega e^{-\frac{1}{2\beta}\sum_a\left(u_aq^\omega+ \xi_a^\omega\right)^2-id^\omega \sum_a \xi^\omega_a -i\sum_i a_i^\omega \sum_a s_{i , a}\xi^\omega_a } \end{aligned}\ ] ] where the sums on @xmath175 runs over the @xmath176 replicas . in the second equation above ,
we have performed the integrals over @xmath177 .
we can now perform the average over the random variables @xmath39 which will be assumed to be normal with zero average and variance @xmath59 .
this yields @xmath178 where the average @xmath179 is taken over gaussian variables with mean @xmath152 and variance @xmath147 and the symbol @xmath180 stands for integration over all the independent entries of the matrix @xmath181 .
in order to evaluate the latter we use a standard delta function identity bringing into play the matrix @xmath182 conjugated to the overlap matrix @xmath183 . the @xmath72-average is evaluated as follows @xmath184}\int_{-\infty}^{+\infty}\prod_i \left[d\epsilon_i e^{- \frac{(\epsilon_i-\overline{\epsilon})^2}{2\sigma_{\epsilon}^2}-\beta\sum_{a}g(s_{i , a } ) } \right]\\ & = & \int \!d \mathbf{r } { \rm tr}_{\vec{s } } e^{\sum_{a\le b } r_{ab } [ \omega g_{ab}-\sum_is_{i , a}s_{i , b } ] -\beta\sum_{i , a}\overline{\epsilon } s_i^a+\frac{\omega \beta^2\sigma_{\epsilon}^2}{2}\sum_{a , b } g_{a , b}},\end{aligned}\ ] ] where the quadratic term @xmath185 arising from the gaussian integration has been replaced by the overlap matrix .
taking the replica symmetric ( rs ) ansatz for @xmath181 and @xmath186 @xmath187 the @xmath188-independent part of the exponent in the integrand takes the form @xmath189 where we have neglected terms of order @xmath190 in view of the @xmath191 limit .
this yields @xmath192 } w_{\overline{\epsilon}}[\nu , r]^n,\ ] ] where we have defined @xmath193=\left[{\rm tr}_{\vec{s } } e^{-\beta\sum_{a}\left[\overline{\epsilon } s_a+\frac{\nu}{2}s^2_a\right ] + \frac{1}{2}\left(\beta r\sum_a s_a\right)^2}\right]^n.\ ] ] in the limit @xmath191 this quantity can be evaluated as follows @xmath194 & = e^{n \log \left [ { \rm tr}_{\vec{s } } e^{-\beta\sum_{a}\left[\overline{\epsilon } s_a+\frac{\nu}{2}s^2_a\right ] + \frac{1}{2}\left(\beta r\sum_a s_a\right)^2}\right]}\\ & = \exp\left\{n \log \left [ { \rm tr}_{\vec{s } } e^{-\beta\sum_{a}\left[\overline{\epsilon } s_a+\frac{\nu}{2}s^2_a\right ] } \big\langle e^{z\beta r
\sum_a s_a}\big\rangle_z \right ] \right\}\end{aligned}\ ] ] where we have performed a hubbard - stratonovich transformation in order to decouple the @xmath195 variables introducing the average @xmath196 over the gaussian variable @xmath87 . clearly : @xmath193 = \exp\left\{n\log\big\langle\left(\int_0 ^ 1 ds e^{-\beta [ s^2+s(\overline{\epsilon}-zr)]}\right)^m\big\rangle_z\right\}.\ ] ] exploiting the usual identity @xmath197 valid for @xmath191 , we finally obtain @xmath198 = \exp\left\{n m\big\langle\log\int_0 ^ 1 ds e^{-\beta [ s^2+s(\overline{\epsilon}-zr)]}\big\rangle_z\right\}.\ ] ] inserting ( [ w ] ) into ( [ iofg ] ) we eventually get @xmath199 + \frac{n}{\beta}\left\langle\log\int_0^\infty\!ds e^{-\beta[\overline{\epsilon } s+\nu s^2/2-rsz ] } \right\rangle_z\right\}}\ ] ] after inserting ( [ iofgfinal ] ) into ( [ zqm ] ) we have to perform the @xmath200 integrals in @xmath201 . for each @xmath9 @xmath202&=\int\!d\vec{\xi } e^{-\frac{1}{2\beta}\sum_a\left(u_aq^\omega+\xi_a\right)^2-id^\omega \sum_a \xi_a -\frac{1}{2}\sum_{a , b}\xi_a \xi_b g_{a , b}}\end{aligned}\ ] ] the matrix @xmath181 , in the rs ansatz has two distinct eigenvalues : @xmath203 therefore the determinant of @xmath204 is clearly @xmath205}.\ ] ] the expression ( [ det ] ) assists performing the gaussian integral in ( [ j ] ) as @xmath206=e^{-\frac{1}{2}\log\left(1+\beta m\frac{g}{1+\chi}\right)-\frac{m } { 2}\log(1+\chi ) + \frac{\beta m}{2(1+\chi+m\beta g ) } ( uq^\omega + id^\omega)^2-\frac{\beta m}{2 } ( uq^\omega)^2}\ ] ] where we have rescaled @xmath207 as @xmath208 and considered for @xmath209 the form @xmath210 @xmath211 .
taking the average over @xmath18 in the limit @xmath191 we use the fact that @xmath212 which in our case yields @xmath213 having taken for @xmath137 an exponential distribution with average @xmath1 . hence @xmath214 \nonumber\\ & ~ & + \frac{n}{\beta}\left\langle\log\int_0^\infty\!ds e^{-\beta[\overline{\epsilon } s+\nu s^2/2-rsz ] } \right\rangle_z.\end{aligned}\ ] ] we first observe that the saddle point equation for @xmath207 yields @xmath215 .
then we take the limit @xmath216 which gives @xmath217 -n \left\langle\min_{s\ge 0 } \left[\overline{\epsilon } s+\frac{\nu}{2 } s^2-rsz\right ] \right\rangle_z\ ] ] the saddle points equations , obtained differentiating ( [ fofbeta ] ) with respect to the order parameters and sending @xmath218 to @xmath219 , read @xmath220 where @xmath221 , since , in our case , the supply is limited to @xmath222 .
the above set of equations can then be recasted in the form @xmath223\\ \chi & = \frac{n{\mathbb{e}}_z[s_zz](1+\chi)}{\sqrt{g+\delta+\sigma_{\epsilon}^2(1+\chi)^2}}.\end{aligned}\ ] ] the above calculation can also be performed , in order to probe the solution , introducing an auxiliary field @xmath224 coupled to the returns @xmath225 .
this allows also to easily compute the average and the fluctuations of returns by deriving with respect to @xmath226 the logarithm of the free energy and then setting @xmath227 : @xmath228
we show here how it is possible to find the critical line .
let us consider the case @xmath229 , so that @xmath230 , with @xmath231 . from the saddle pont equations we get @xmath232i_2(z_0)}{1-ni_2(z_0)}\ ] ] where
we have defined @xmath233 .
inserting now this expression for @xmath234 into equation ( [ r ] ) and explointing @xmath235 we obtain @xmath236 that using ( [ chi ] ) can be writen as @xmath237 where @xmath238 .
if we now look for a solution in the case @xmath239 the above equation reduces to @xmath240 and we also have that @xmath241 .
these equations define the critical line . in particular
it is clear at this level that the dependence on the parameters of the risk premia distribution enters trough the ratio @xmath242 99 w. e. buffet , berkshire hathaway annual report 2002 .
monetary fund , _ global financial stability report _ ( april 2008 ) .
monetary fund , _ global financial stability report _ ( october 2008 ) .
m. m. dacorogna , r. genay , u. mller , r. b. olsen , o. v. pictet , _ an introduction to high - frequency finance _ , chap .
v , academic press , london ( 2001 ) .
bouchaud , j. d. farmer and f. lillo , _ how markets slowly digest changes in supply and demand _ , preprint arxiv:0809.0822 osler , c. l. , 2006 .
macro lessons from microstructure
. international journal of finance and economics 11 , 55 - 80 .
brock , c.h .
hommes , f.o.o .
wagener , _ more hedging instruments may destabilize markets _ cendef working paper 08 - 04 university of amsterdam ( 2008 ) . g. raffaelli , m. marsili , jstat l08001 ( 2006 ) ; m. marsili , g. raffaelli , and b. ponsot submitted to j. ec .
dyn . control ( 2008 ) .
bouchaud , _ economics need a scientific revolution _ , nature , vol 455 ,
( 2008 ) k. r. sircar and g. papanicolaou , appl .
fin . * 5 * , 45 - 82 ( 1998 ) ; m avellaneda , md lipkin , a market - induced mechanism for stock pinning , quantitative finance , * 3 * , 417 - 425 ( 2003 ) ; m. jeannin , g. iori , d. samuel , _ the pinning effect : theory and a simulated microstructure model _
, quantitative finance ( forthcoming , 2008 ) ; s.x.y .
pearson , a.m. poteshman , j. fin .
econ . * 78 * , 48 - 87 ( 2005 ) .
r. c. merton and z. bodie _ design of financial systems : towards a synthesis of function and structure _ , j. of investment management , 3 , ( 2005 ) , 1 - 23 robert j. shiller , _ the subprime solution : how today s global financial crisis happened , and what to do about it _ , princeton university press ( 2008 ) .
a. de martino and m. marsili , j. phys .
a , * 39 * , r465-r540 ( 2006 ) .
d. challet and m. marsili phys .
e 68 , 036132 ( 2003 ) .
t. lux and m. marchesi , nature * 397 * , 498 - 500 ( 1999 ) .
s. r. pliska , _ introduction to mathematical finance : discrete time models _ , ( blackwell , oxford , 1997 ) .
m. marsili , ssrn e - print 141597 , 2009 ( available at http://ssrn.com/abstract=1415971 ) .
m. wyart , j.p .
bouchaud , j. ec . behavior and organization ,
* 63 * , 1 ( 2007 ) . j. d. farmer , s. joshi , j. ec .
behav . and organization , * 49 * , 149 - 171 ( 2002 ) .
r. m. may , nature 238 ( 1972 ) , 413 r. mehrotra , v. soni and s. jain , _ diversity begets stability in an evolving network _ , preprint arxiv : nlin.ao/07051075 .
morris , s. and shin , h.s .
, _ financial regulation in a system context _ , brookings papers on economic activities , washington , * 11 * ( 2008 ) .
tobin j , _ `` the new economics one decade older '' _ , the eliot janeway lectures on historical economics in honour of joseph schumpeter 1972 ( princeton university press , princeton , us ) ; _ eastern economic journal _ * iv * 153 - 159 ( 1978 ) g. bianconi , t. galla , m. marsili and p. pin , _ effects of tobin taxes in minority game markets _ to appear in j. econ . behavior and organization ( 2009 ) .
d. w. montgomery , journal of economic theory * 5 * , 395 - 418 ( 1972 ) . | we contrast arbitrage pricing theory ( apt ) , the theoretical basis for the development of financial instruments , with a dynamical picture of an interacting market , in a simple setting .
the proliferation of financial instruments apparently provides more means for risk diversification , making the market more efficient and complete . in the simple market of interacting traders discussed here ,
the proliferation of financial instruments erodes systemic stability and it drives the market to a critical state characterized by large susceptibility , strong fluctuations and enhanced correlations among risks .
this suggests that the hypothesis of apt may not be compatible with a stable market dynamics .
in this perspective , market stability acquires the properties of a common good , which suggests that appropriate measures should be introduced in derivative markets , to preserve stability .
`` in my view , derivatives are financial weapons of mass destruction , carrying dangers that , while now latent , are potentially lethal '' . at the time of warren buffet warning @xcite markets were remarkably stable and the creativity of financial engineers was pushing the frontiers of arbitrage pricing theory to levels of unprecedented sophistication .
it took 5 years and a threefold increase in size of credit derivative markets @xcite , for warren buffet `` time bomb '' to explode @xcite .
paradoxically , this crisis occurred in a sector which has been absorbing human capital with strong scientific and mathematical skills , as no other industry .
apparently , this potential was mostly used to increase the degree of sophistication of mathematical models , leaving our understanding lagging behind the increasing complexity of financial markets .
recent events raise many questions : * the sub - prime mortgage market was approximately 5% of the us real estate market , and the risk was well diversified across the whole world .
why did such a small perturbation cause so large an effect ? * why are stock prices affected by a crisis in derivative markets ?
* why did correlations between risks grow so large during the recent crisis @xcite , with respect to their _ bona - fide _ estimates before the crisis ? these and other questions find no answer in the theories of mathematical economics @xcite .
indeed , real markets look quite different with respect to the picture which arbitrage pricing theory ( apt ) @xcite gives of them .
the problem is that apt is not merely a theoretical description of a phenomenon , as other theories in natural sciences .
it is the theory on which financial engineering is based .
it enters into the functioning of the system it is describing , i.e. it is part of the problem itself .
the key function of financial markets is that of allowing inter - temporal exchanges of wealth , in a state contingent manner .
taking this view , apt makes it possible to give a present monetary value to future risks , and hence to price complex derivative contracts . in order to do this ,
apt relies on concepts , such as perfect competition , market liquidity , no - arbitrage and market completeness , which allows one to neglect completely the feedback of trading on market s dynamics .
these concepts are very powerful , and indeed apt has been very successful in stable market conditions .
in addition , the proliferation of financial instruments provides even further instrument for hedging and pricing other derivatives .
so the proliferation of financial instruments produces precisely that arbitrage - free , complete market which is hypothesized by apt . in theory . in practice , markets are never perfectly liquid . the very fact that information can be aggregated into prices , requires that prices respond to trading ( see e.g. @xcite for evidence on fx markets or @xcite for equity markets ) . in other words , it is because markets are illiquid that they can aggregate information into prices .
liquidity indeed is a matter of time scale and volume size @xcite .
this calls for a view of financial markets as interacting systems .
in this view , trading strategies can affect the market in important ways .
both theoretical models and empirical research , show that trading activity implied by derivatives affects the underlying market in non - trivial ways @xcite .
furthermore , the proliferation of financial instruments ( arrow s securities ) , in a model with heterogeneous agents , was found to lead to market instability @xcite .
the aim of this paper is to contrast , within a simple framework , the picture of apt with a dynamical picture of a market as an interacting system .
we show that while the introduction of derivatives makes the market more efficient , competition between financial institutions naturally drives the market to a critical state characterized by a sharp singularity .
close to the singularity the market exhibits the three properties alluded to above : _
1 ) _ a strong susceptibility to small perturbations and _ 2 ) _ strong fluctuations in the underlying stock market . furthermore _
3 ) _ while correlations across different derivatives is largely negligible in normal times , correlations in the derivative market are strongly enhanced in stress times , when the market is close to the critical state . in brief
, this suggests that the hypothesis of apt may not be compatible with the requirement of a stable market .
capturing the increasing complexity of financial markets in a simple mathematical framework is a daunting task .
we draw inspiration from the physics of disordered systems , in physics , such as inhomogeneous alloys which are seen as systems with random interactions .
likewise , we characterize the typical properties of an ensemble of markets with derivatives being drawn from a given distribution . our model is admittedly stylized and misses several important features , such as the risks associated with increased market exposure in stress conditions or the increase in demand of financial instruments in order to replicate other financial instruments .
however , it provides a coherent picture of collective market phenomena .
but it is precisely because these models are simple that one is able to point out why theoretical concepts such as efficient or complete markets and competitive equilibria have non - trivial implications .
the reason being that these conditions hold only in special points of the phase diagram where singularity occurs ( phase transitions ) .
it is precisely when markets approach these ideal conditions that instabilities and strong fluctuations appear @xcite . loosely speaking ,
this arises from the fact that the market equilibrium becomes degenerate along some directions in the phase space . in a complete , arbitrage - free market
, the introduction of a derivative contract creates a symmetry , as it introduces perfectly equivalent ways of realizing the same payoffs .
fluctuations along the direction of phase space identified by symmetries can grow unbounded .
loosely speaking , the financial industry is a factory of symmetries , which is why the proliferation of financial instruments can cause strong fluctuations and instabilities . in this respect , the study of competitive equilibria alone can be misleading .
what is mostly important is their stability with respect to adaptive behavior of agents and the dynamical fluctuations they support and generate .
the rest of the paper is organized as follows : the next section recalls the basics of apt in a simple case , setting the stage for the following sections .
then we introduce the model and discuss its behavior .
we illustrate the generic behavior of the model in a simple case where the relevant parameters are the number of different derivatives and the risk premium . in this
setting we first examine the properties of competitive equilibria and then discuss the fluctuations induced within a simple adaptive process , by which banks learn to converge to these equilibria .
then we illustrate a more general model with a distribution of risk premia .
this confirms the general conclusion that , as markets expand in complexity , they approach a phase transition point , as discussed above .
the final section concludes with some perspectives and suggests some measures to prevent market instability . |
the vast majority of stars ( initial mass @xmath58 ) end their lives as white dwarfs ( wd ) , after passing through the asymptotic giant branch ( agb ) and planetary nebula ( pn ) phases , in which they shed much of their mass .
the ejected material enriches the interstellar medium ( ism ) with newly synthesized nuclear products ( mainly he , c , n , and possibly o ) , to different extents , depending on the initial stellar mass and exact evolutionary path ( e.g. marigo 2001 , karakas 2010 ) .
intermediate mass stars are the main providers of carbon and nitrogen , whereas low - mass stars are the most relevant component for the mass budget of stellar remnants in galaxies .
most of the stellar mass is shed in the agb and pn phases , but the evolution through these phases is still subject to considerable uncertainties , in particular regarding mass loss and the efficiency of the third dredge - up .
stars within an intial mass range of @xmath00.8 to 8 end as wds with a narrow mass range , mostly below 0.8 .
it is important to understand how the mass of their precursors relates to the final wd mass , in order to understand the relative contribution of different stars to the chemical enrichment of elements such as he , c , n and o. while the evolution of the wd progenitors in the main sequence phase is fairly well understood and observationally constrained , the hot - wd population is hitherto quite elusive , owing to their small radius , hence low optical luminosity , and extremely hot temperatures , to which optical colors are insensitive ( see e.g. @xcite , @xcite , @xcite ) as well as to their very short lifetimes on the constant - luminosity post - agb phase . to make matters more difficult
, the post - agb luminosity at a given stellar temperature varies significantly according to the stellar mass , making it impossible to infer absolute luminosity from other physical parameters .
the evolutionary time spent on the constant - luminosity post - agb phase and on the cooling track is also a strong function of the mass ( vassiliadis & woods 1994 ) .
therefore , the exact relation between progenitor s initial mass and wd mass ( initial - final mass relation , ifmr ) , remains to date a crucial missing link in our understanding of stellar evolution and chemical enrichment of the galaxy .
a characterization of the population of hot wds in the milky way can reduce these uncertainties , and lead to a better understanding of processes that drive the chemical evolution of galaxies like the milky way .
uv photometry combined with optical measurements significantly increases the sensitivity to the hottest temperatures .
for example , the color difference between a = 50,000k and 20,000k star is about 1.5mag in fuv-_g _ , but @xmath30.4mag in u - b , and @xmath30.15 mag in _ g - r _ which are comparable to photometric errors when large surveys are considered .
the sensitivity gained by extending the measurements to uv wavelengths is more critical for discerning the hottest stars ( see e.g. fig.s 5 - 7 of @xcite and bianchi 2009 ) . the census and characterization of cool compact objects
has significantly improved in recent years thanks to optical and ir surveys .
the @xcite catalog from the sdss fourth data release ( dr4 ) contains 9316 spectroscopically confirmed wds and 928 subdwarfs over an area of 4783deg@xmath6 ; about one fourth ( 2741 ) have @xmath418,000k as estimated by the sdss pipeline automated spectral analysis .
an additional @xmath05 - 6000 wd are expected from dr7 ( @xcite ) .
seven ultracool wd were added to the census by @xcite .
gontcharov et al ( 2010 ) extracted from 2mass , tycho-2 , xpm and ucac3 catalogs combined 34 wds , 1996 evolved ( 11,000@xmath3 @xmath360,000k ) and 7769 unevolved ( @xmath37,000k ) subdwarfs , using multicolor photometry and proper motions of stars with 6@xmath3k@xmath7@xmath314 mag .
the current version ( 2008 ) of the mccook & sion ( 1999 ) catalog of spectroscopically confirmed wds includes 10132 entries ( all types ) , a factor of five increase over a decade , with respect to the original ( 1999 ) version listing 2249 wds . @xcite
give a catalog of 201 da wds , spectroscopically confirmed from the @xmath01,000 h - rich da wds discovered in the 2df qso redshift survey ; croom et al .
( 2004 ) 2df catalog includes 2071 wds over 2000 deg@xmath6 .
wds in binaries from the sdss were catalogued and studied by several authors , see e.g. rebassa - mansergas et al.(2010 ) and references therein , silvestri et al ( 2007 ) , heller et al .
( 2009 , wd - m star binaries ) from spectroscopy ; other authors used optical - ir photometry to search for binaries among the known wds ( e.g. tremblay & bergeron 2007 , watcher et al .
special classes such as cvs are addressed by other works ( e.g. szkody et al . 2009 and references therein ,
gaensike et al . 2009 ) , while others studied in detail the very local population ( e.g. holberg et al ( 2002 ) discuss 122 objects within 20 pc from the sun , which they estimate to be an 80% complete sample within this distance , and which include some double - degenerate systems ) . finding the hottest , smallest stars , however , remained a challenge prior to the galex uv sky surveys , which provide deep sensitivity and large area coverage .
for example , 105 were found in the original euve whole - sky survey , with small subsequent additions obtained by combining euve and rosat : see e.g. dupuis ( 2002 ) for a review and discussion .
the rosat whole - sky survey produced 175 wd in x - rays , mostly das ( fleming et al .
1996 , from rosat pspc ) .
such catalogs of confirmed or candidate wds enable the study of these objects as astrophysical probes , of stellar evolution , of mw structure , of the local neighborhood , etc .
most importantly , comprehensive catalogs enable selection of targets for follow - up spectroscopy , which then provides the detailed physics of these objects , especially when extended to the uv and far - uv wavelengths , as proven by the score of results enabled by iue , fuse , and hst spectrographs .
bianchi ( 2009 , 2007 ) , bianchi et al .
( 2009a , 2007a , 2005 ) have demonstrated the power of far - uv and near - uv measurements , afforded for the first time over large areas of the sky by galex , to unambiguously detect and characterize extremely hot stellar sources .
not only the uv wavelengths are more sensitive to the temperatures of hottest stars , but the combination of uv and optical colors allows a better separation of different classes of astrophysical objects ( e.g. bianchi 2009 and references therein ) .
our present work provides a selection of hot - star candidates from uv photometry ; about 40,000 hot - star candidates with fuv , nuv photometric [email protected] ( about 74,000 with photometric [email protected] ) have also sdss optical photometry .
the majority are likley hot wd with log(g ) between 7 and 9 .
the catalog covers different latitudes and enables a first analysis of this stellar population with milky way models .
larger catalogs of uv sources with and without optical sdss match are also constructed , and made available as online products . in section [ s_data ]
we describe the method used to construct `` clean '' catalogs of unique uv sources from two galex surveys with different depths and coverage , and subsets of these catalogs with matched optical photometry , and we present the catalogs characteristics . in section [ s_catalog ]
we extract samples of hot - star candidates , and we analyze them with milky way models in section [ s_discussion ] .
discussion and conclusions are given in section [ s_conclusions ] .
the galaxy evolution explorer ( galex ) is imaging the sky in far - uv ( fuv , 1344 - 1786 , @xmath8 = 1538.6 ) and near - uv ( nuv , 1771 - 2831 , @xmath8 = 2315.7 ) simultaneously , with a field - of - view of 1.2@xmath9 diameter and a resolution of 4.2/5.3@xmath10 [ fuv / nuv ] ( @xcite ) .
the images are sampled with 1.5 pixels . nested surveys with different depth and coverage
are in progress .
the widest sky coverage is provided by the all sky imaging survey ( ais ) and the medium [ depth ] imaging survey ( mis ) , that reach typical depths of 19.9/20.8mag ( fuv / nuv ) and 22.6/22.7mag ( fuv / nuv ) respectively , in the ab magnitude system . the nearby galaxy survey ( ngs , bianchi et al .
2003 , bianchi 2009 , gil de paz et al .
2007 ) , with over 300 fields at mis depth , targeted nearby , hence fairly extended , galaxies , therefore it has been excluded in the present work , lest some sources from galaxies `` shredded '' by the pipeline intrude on our catalog .
see also @xcite , @xcite , @xcite , @xcite , for a general discussion of the content of the uv sky surveys .
in this paper we use data from the galex fifth data release ( gr5 ) ais and mis surveys , which include a total of 28,269 and 2,161 fields respectively .
the data are taken from the mast archive .
we restrict the catalogs to sources within the central 1@xmath9 diameter of the field ( for good photometry and astrometry , and to exclude edge artifacts ) . with such restriction , and eliminating overlaps
, these surveys cover a total unique area of 21,434.8 ( ais ) and 1,578.6 ( mis ) square degrees ( section [ s_area ] ) .
section [ s_catuv ] describes the construction of the catalog of unique galex sources . in order to separate the uv sources by astrophysical classes , we examine in this work the portions of the galex gr5 ais and mis surveys that are also included in the footprint of the sloan digital sky survey seventh data release ( dr7 ) , which provides five optical magnitudes : _ u g r i z _ in addition to the galex fuv , nuv magnitudes .
the overlap between galex gr5 and sdss dr7 includes 10,316 / 1,655 ( ais / mis ) galex fields , and the area coverage of the overlap is 7,325 ( ais ) and 1,103 ( mis ) square degrees .
details of the area calculation are provided in section [ s_area ] .
the sky coverage of ais and mis in the galex data release gr5 and its overlap with sdss dr7 are shown in fig .
[ f_coverage ] . here
we describe the procedure used to construct the catalog of unique galex sources ( i.e. eliminating repeated measurements ) .
all catalogs described in this paper are made publicly available from our website http://dolomiti.pha.jhu.edu/uvsky , and from mast at http://galex.stsci.edu , and as high - level science products ( http://archive.stsci.edu/hlsp/ , in the `` catalogs '' section ) .
therefore , we provide here information on how they were constructed that will be relevant for potential users , as well as to others interested in constructing future versions of similar samples .
we extracted catalogs of galex sources from stsci mast ( www.mast.stsci.edu ) , at the casjobs sql interface ( www.galex.stsci.edu ) .
sources were extracted from the table `` photoobjall '' , from the mis and ais surveys separately , with the criteria that the source distance from the field center had to be @xmath20.5@xmath9 and the photometric error less than 0.5 mag in nuv . in other words , we initially included in our general galex source catalog all nuv reliable detections , regardless of whether they have also an fuv detection .
for our selection of hot star candidates , we will eventually impose the additional criterion of good fuv photometry .
the additional restriction of err@[email protected] mag significanly reduces the number of sources ( by a factor of up to 10 , see table [ t_stats ] ) , with respect to the total number of nuv detections , and of course introduces a bias in the source catalog , the hottest and bluest sources ( the subject of this paper ) being not affected but the redder sources being progressively eliminated , as discussed in sections [ s_errcuts ] and [ s_stats ] . for more discussions about statistical properties , and biases inherent to sample selections , see @xcite .
a few observations planned as part of the mis or ais surveys actually partly failed , and resulted in one of the detectors ( most often fuv ) not being exposed ( such observations are typically repeated later ) .
the observations with one of the two bands having zero exposure time were eliminated from our catalog , otherwise they will bias the statistics of fuv - detection over number of total nuv detections , and the corresponding fields were not counted in the area calculation .
[ f_exptime ] shows the distribution of exposure times ( fuv is shown , but nuv is generally equal or larger ) , for the ais and mis fields .
the typical exposure time for mis is 1500 sec , which is met or exceeded by the majority of fields .
the ais survey aims at exposure times of the order of @xmath0100 sec .
we retained also fields with exposures shorter than typical . therefore , while the typical depth of the two surveys ( ais and mis ) differ by @xmath02 mag , the exposure level is not strictly homogeneous across each catalog .
[ f_exptime]-bottom panel shows the number of galex sources in each galex field .
we color - coded the fields by three ranges of exposure time for the ais ( two for mis ) , since the number of sources detected above a given error - cut generally increases with exposure time .
this plot is useful to check for fields with overdensities , since the surveys , the broad ais in particular , include also some stellar clusters .
in general the spread in number of sources per field is just about a factor of two for mis .
a few ais fields around latitude -30@xmath9 have an overdensity of almost one dex , however the total number of ais fields in each 10@xmath9-wide latitude bin ( used in our analysis , see next section ) is very large , and a few overdensities do not affect our analysis of stellar counts with milky way models over wide areas . fields with high density of sources , and in particular the stellar clusters included in the surveys , will be separately analyzed elsewhere .
they are included in our catalog for completeness and for possible use by others , although customized photometry is desirable in very crowded fields ( e.g. de martino et al . 2008 ) .
nearby galaxies are generally observed as part of the ngs survey which is excluded from our catalog , but a few large galaxies are also in the footprint of ais and mis and bright knots of galaxies `` shredded '' by the pipeline may enter the catalog as separate sources ( see @xcite ) .
a few such sources may have fuv - nuv@xmath3 - 0.13 and appear pointlike in the sdss catalog , and therefore may enter our catalog of hot - star candidates ( section [ s_catalog ] ) , but they would not affect the statistical results . for other , more specific uses of our catalog ,
they may be removed by checking against a list of nearby galaxies , as we did in @xcite . using the `` child '' flags from the sdss pipeline , that track deblended sources , proved not to be useful in indentifying and weeding out sources from `` shredded galaxies '' .
the searches and download of galex sources with the above criteria were done using the java casjobs command - line tool ( casjobs.jar ) .
one problem we often encountered when using this tool to extract and download query results from the database is that sometimes the download pipe gets broken prematurely , and the results are truncated without any warning being given . in order to verify that the output file contained all the results , a separate count query was run on the sql server and compared to the number of output sources in the downloaded file each time .
it was necessary to subdivide the search in small latitude strips due to various limits set by the casjobs interface ( e.g. query length , output file size ) , as well as to avoid the frequent problem of the long searches being interrupted .
the galex archive contains multiple observations of the same source , when some fields overlap or are repeated .
having all measurements of each source is useful for variability studies ( which will be addressed in a forthcoming work ) , and for chosing the best measurement when several are available . for our present purpose we constructed from the total output a unique - source catalog , in the following manner .
galex sources were considered possible duplicates if they lied within 2.5 of each other . if two galex sources were within this distance , but had the same `` photoextractid '' ( i.e. , they are both from the same observation ) they were both considered unique .
otherwise , they were assumed to be multiple observations of the same source .
we choose , to represent the unique source , the measurement with the longest nuv exposure time . in the case of equal exposure times , the observation where the source was closer to the field center was chosen ( i.e. the source with the smallest `` fov_radius '' value from the `` photoobjall '' table ) , as photometric quality is usually better in the central part of the field .
a portion of the galex survey areas is in the footprint of the sdss dr7 ( see fig.1 - bottom panel ) , and for the uv sources in this area we constructed a catalog of matched optical sources .
we uploaded the coordinates of our galex unique sources into the sdss sql interface ( version 3_5_16 rev 1.70 , at casjobs.sdss.org ) and queried for sdss matches against the `` photoprimary '' table ( sdss source catalog that includes only unique sources ) using an inital search radius of 4.2@xmath10 . a match radius of 3was eventually used in the final catalog , as we shall see .
as with the galex searches , the sdss searches were done in an automated way on 1@xmath9-wide galactic latitude strips using the casjobs command - line tool .
we found that there is currently a bug in the sdss version of this tool ( v0.03 ) that causes the last character of a file to be dropped when extracting query results .
we compensated for this by writing our query to pad the end of the file with an additional character .
a given galex source may have multiple sdss matches within the search radius , given the higher spatial resolution of the sdss ( @xmath01.4 ) . in such cases the casjob search returns multiple lines with the same galex source and the various sdss matches .
we `` ranked '' the multiple sdss matched sources based on the distance , with the closest sdss source being retained as the `` match '' for the galex source and the additional sdss sources being noted .
when using uv - optical colors for source classification , uv sources with multiple optical matches must be excluded , because , even though the closest optical source may be the actual counterpart , the uv photometry at the galex spatial resolution of @xmath05may be a composite measure of two optical sources , and therefore the color would not be meaningful .
the fraction of uv sources with multiple optical matches , eliminated from the analysis sample , is then taken into account when estimating density of astrophysical sources ( # per square degree ) .
this fraction is given in table [ t_stats ] .
sources with multiple optical matches are listed separately in our hot - star catalogs , and are included with `` rank''@xmath40 in the total matched - source catalog .
the relative number of uv sources with multiple optical matches , shown in figure [ f_exptime2 ] , increases towards low galactic latitudes , as expected .
the figure refers to our final catalog with match radius of 3@xmath10 , and pointlike sources only : approximately 10% of the uv sources have more than one optical counterpart , at intermediate and high galactic latitudes , in agreement with our previous work on earlier data releases which covered mainly high galactic latitudes ( bianchi et al . 2005 , 2007a ) .
the fraction increses to slightly over 20% for latitudes @xmath12b@xmath12 @xmath1310 - 20@xmath9 , reflecting the higher density of stars in the mw disk , and is still very uncertain at latitudes @xmath12b@xmath12@xmath310@xmath9 , where we have little area coverage . for a larger match radius ( 4.2@xmath10 ) ,
the fraction of multiple matches increases significantly , relative to the total number of matches , and so does the incidence of spurious matches ( sect .
[ s_spurious ] ) , therefore we adopted a final match radius of 3 in our catalog .
the fraction is also higher if exended sources are included .
( small dots ) , as well as averaged over 5@xmath9 ( blue circles ) .
the increase towards the galactic plane reflects the higher density of optical sources at low mw latitudes .
the plot also shows a slight north - south asymmetry .
the fractions shown are for point - like sources , with error cuts of 0.3 mag .
the fraction increases if a larger match radius is used .
the red dots are the incidence of spurious matches ( see text ) .
[ f_exptime2],width=340 ] in the interest of others who may want to apply the same procedures , we also mention that , due to the large current area coverage and consequent large number of sources , when it is necessary to perform the searches on small contiguous portions of the sky ( e.g. 1@xmath9 strips ) , it may happen that the coordinates of a galex uv source fall in one strip ( close to the latitude limit ) , and the sdss match falls in the next latitude range , due to a small difference in coordinates .
such cases are included in the final catalogs in the latitude range which is appropriate according to the position of the galex source .
care should also be taken when using the galactic coordinates as returned by the galex database , as those coordinates differ by @xmath50.5from those derived by converting the galex ra , dec coordinates to galactic coordinates using standard astronomical packages ( e.g. , wcstools from the smithsonian astrophysical observatory ) .
these differences probably arise because the galex database stores galactic coordinates using real data types , while using double precision data types to store ra and dec values . therefore ,
sorting galex search results based on the galactic coordinates returned by the database may give different results than using wcstools to derive those coordinates from the galex ra and dec .
another contingency that must be tested for , in the above procedures , is whether a given sdss source matches multiple galex sources .
this can occur if a sdss source lies somewhere in between two galex sources which were deemed unique . in this case
, the match query returns a match of each of the galex sources with the same sdss object .
we retain the match with the galex source which is the closest to the sdss source , and eliminate the other .
this occurs rarely .
the probability of spurious matches was estimated as follows .
we randomly selected 30% of the coordinates from our matched galex - sdss catalogs ( in separate 5@xmath9 bins of galatic latitude ) , and searched the sdss database against coordinates offset by 0.5@xmath14 from those of the real sources .
the spurious match rate , shown in fig . [ f_exptime2 ] ( red dots ) , is the number of incidences where one or more matches were found ( within a search radius of 3 " ) divided by the number of uploaded coordinates . as for the multiple matches , it is higher towards the galactic plane , where the stellar density increases and therefore also the probability of random matches within a given radius .
the rate is of the order of a few% at high and intermediate latitudes .
however , matches that are positional coincidences are likely to have odd colors and the fraction may be lower in selected analysis samples .
we point out , in the interest of users of our catalogs , that there are several differences from other basic matched source catalogs posted on mast .
most notably , we include only the central 1@xmath9 diameter portion of the field , therefore we conservatively eliminate all edge sources ( mostly defects or bad measurements but also some good sources ) , but we retain all good portions .
second , we consider all existing measurements for each source , from which the best is chosen .
for example , the matched catalog described by @xcite instead , eliminates overlap by reducing each galex field to an hexagon ( selection of primary ) . in this procedure , the vertices of the hexagons include sources near the edge ( i.e. possibly also sources with poor photometry , and artifacts ) .
moreover , since the hexagons are fixed at nominal positions , but the actual pointings may differ , unnecessary gaps between fields ( which may not exist in the actual observations ) are introduced , as the authors point out . our partition , described in section [ s_area ] , and following @xcite , avoids such problems and enables better and more homogeneous photometry quality , as well as easy and precise calculations of the unique - area coverage for our galex surveys , and of overlap area with any other survey , consistent with the actual source catalog rather than with nominal field centers .
this is preferable for our analysis .
such differences may cause samples selected from different matched catalogs , to differ slightly . in order to derive the density of sources extracted from our catalogs , we computed the total areas of unique coverage , taking into account overlap between fields , in the galex gr5 , and then the area overlapping with sdss dr7 , following the method of @xcite . since we restricted the catalogs to sources within the central 1@xmath9 of the galex field , for each field we considered an effective radius of 0.5@xmath9 .
our code scans the entire sky and calculates the unique area covered by the galex fields , and the portion of this area covered also by the sdss .
first , the whole sky is divided in small , approximately square , tassels , along galactic longitude and latitude .
we used steps in _
l , b _ of 0.05/0.1@xmath9 for mis / ais respectively .
then , we find the distance between the center of each tassel in our whole - sky grid and the center of each galex field , and sum the areas of the tassels that are within half degree of a field center , avoiding to count the same tassel twice , which eliminates overlap between fields .
the error in the estimated area depends on the step used to calculate the grid of tassels ( tassel size ) , and the number of tassels .
the area of each tassel decreases at high latitudes ( the step in _
l , b _ is kept constant over the whole sky ) , hence the statistical error due to the tassels along the field s edge ( i.e. tassels that fall partly inside and partly outside the 1@xmath9 circleof a galex field ) varies with the location of the fields , both with latitude and with the relative location of the field center to a grid step .
such errors cancel out statistically for a large number of fields . for a very small area however , i.e. for a few fields , it is desirable to use smaller tassels .
we estimate the uncertainty by computing the areas several times , each time offsetting the positions of the galex field centers by about half the size of a tassel , in both latitude and longitude directions .
the resulting uncertainty is @xmath51 square degree , for our total area and the areas of each latitude bin . in order to calculate the area of overlap between the galex mis and ais surveys and sdss dr7 , we initially matched the centers of all tassels included in our galex coverage ( mis and ais ) to the sdss dr7 footprint : the areas of tassels deemed within the sdss footprint
were summed to obtain the area coverage of our matched source catalogs .
however , we discovered an issue in the sdss database dr7 footprint that prevented its use for calculating the overlap areas .
the output returned from the sdss database footprint query is currently somewhat incorrect , perhaps due to some sdss fields which appear included in the dr7 footprint in the database ( therefore counting towards area coverage in any estimate ) , however do not have sources in the `` photoprimary '' table or corresponding images because the observation failed or was marked as bad for some reason .
we have consulted several experts of the sdss database , but we found no way so far to identify such fields specifically .
when we plotted on the sky the sdss matched sources ( obtained from the sdss photoprimary " catalog ) , and the sky - tassels deemed by the footprint - query to be within the footprint , we noted some not irrelevant areas of mismatch ( most notably , in the latitude strip 50 - 60@xmath9south we found the largest false positive " area , about 17.5deg@xmath6 in total ; the discrepancies are much smaller at other latitudes ) .
we attempted to perform the footprint query on the sdss database using different syntaxes .
specifically , a basic all - inclusive footprint sql search returns several false - positive `` answers ( i.e. several contiguous areas that the database considers within the footprint , have no sdss sources ) , hence the area based on such outputs would be overestimated with respect to the source catalog ; a more detailed footprint query , confining the search to ' ' primary `` footprint area , returns both false - positives and false - negatives .
we finally could not use the sdss footprint query , and wrote a separate code to independently search , for our entire galex sky coverage , sub - areas with and without sdss sources ; we considered the areas devoided of any sdss sources as outside the sdss footprint .
this procedure was complicated because ( 1 ) in order to decide whether a tassel contained sdss sources , we had to consider sdss sources of any color , not just our hot - star catalog ( which includes the rarest stars in the sky , hence some tassels may actually not have any hot star even though they are located within the footprint ) , and ( 2 ) we had to consider much larger sky - tassels than our original grid , to avoid false - negatives , and then iterate within the ' ' positive " tassels with a progressively finer grid ( down to 0.05deg@xmath6 , about 1/16th of a galex field ) , to confine the uncertainty of the area estimate .
it is important for our analysis , which compares density of sources ( # /square degree ) with model predictions , to estimate areas _ consistent _ with the source catalog used , in this case the primary catalog from the sdss .
we used our area estimates in the following analysis .
we stress , for future reference , that there was no way to detect the inconsistency between catalog and database footprint other than by plotting the distribution of sources and footprint - tassels on sky coordinates in various ways .
this is a desirable test on any such work .
the galex gr5 ais and mis sky coverage , and the galex - sdss overlap area in the matched gr5xdr7 releases , restricting the galex field to 1@xmath9 diameter and eliminating overlaps between galex fields , are given in table [ t_area ] , for the whole coverage and for separate ranges of galactic latitude .
figure [ f_stats ] shows the uv magnitude distribution of our galex unique - source catalog , and the effect of progressive error cuts in nuv , and in both nuv and fuv .
while a progressively stringent error cut in one band ( nuv ) simply truncates the sample to a brighter magnitude limit , an additional error cut in fuv , which effectively raises the faint limit of the sample in this band , causes a relatively higher decrease of redder sources , and the histogram of sources distribution becomes double - peaked .
the same effects are seen ( right panel ) for the matched galex - sdss sources catalog .
this catalog can also be separated in pointlike and extended sources using the sdss spatial information ( @xmath01.4 resolution ) .
such distinction shows that most of the faint - magnitude peak of the source distribution is due to extended sources , which are likely galaxies , as can be expected ( e.g. bianchi 2009 ) . the most restricted sample in the right - panel histograms , the point - like matched sources with error @xmath20.3mag in both fuv and nuv , and fuv - nuv@xmath3 - 0.13
, is the subject of this paper ; it contains mostly hot star candidates ( @xmath4 approximately 18000k ) , with some contamination by qsos at faint magnitudes and red optical colors , discussed in section [ s_purity ] ( see also @xcite ) .
the analysis of the hot - star sample with milky way models ( see later ) shows that the magnitude limit introduced by our error cut of @xmath20.3 mag eliminates mostly halo and thick - disk mw stars from the sample ( between the green solid- and dashed - lines on the right - hand panel ) . on the other hand , including sources with larger photometric errors would introduce significant contamination of the sample by non - stellar objects ( see fig [ f_colcol ] ) .
more statistical analysis of the catalogs , and discussion of potential biases in flux - limited sample selection , are given by @xcite .
the matched galex - sdss sources in our catalogs are shown in color - color diagrams in figure [ f_colcol ] as density plots .
model colors for major classes of astrophysical objects are also shown to guide the eye in interpreting the distribution of sources .
we refer to @xcite and references therein for other similar figures and description of the model colors .
we selected hot stellar candidates by extracting the matched pointlike sources with fuv - nuv @xmath3 -0.13 ( in the ab magnitude system ) , which corresponds to stars hotter than @xmath1318,000 k , the precise depending on the stellar gravity , and on the model atmospheres adopted ( discussed later).=3.1 , and moderate reddening amounts , the galex fuv - nuv color is basically reddening - free .
in fact , the higher selective extinction a@xmath15/ in the fuv range is approximately equalled by the effect of the broad 2175 feature in the nuv range , see fig . 2 , and bianchi ( 2009 ) .
the absorption in each uv band , however , is much higher than in optical bands . ]
this fuv - nuv limit was chosen to largely exclude all extragalactic objects , based on colors derived from canonical templates of qsos and galaxies ( e.g. bianchi 2009 and references therein , see also fig .
[ f_colcol ] ) . as shown by figure [ f_colcol ] in this work , and by fig
. 7 ( lower panels ) @xcite , at ais depth ( @xmath021th mag ) sources with fuv - nuv @xmath3 -0.13 have very little contamination by extragalactic objects , but at fainter magnitudes a number of qsos and galaxies with extremely `` blue '' fuv - nuv colors is found .
@xcite have examined the nature of such uncommon qsos , and their figure 1 and 2 show the location of these objects in the color - color diagrams .
these qsos have extremely `` blue '' fuv - nuv ( more negative than typical qsos at any redshift ) , but optical colors typical of qsos and galaxies .
therefore they overlap , in uv optical color - color diagrams , with the locus of hot stars with a cool companion , i.e. hot stars having very blue ( i.e. negative ) fuv - nuv color but optical colors redder than what their fuv - nuv would imply for a single hot source .
we discuss the qso contamination in section [ s_purity ] .
we restricted the catalog to sources with photometric errors @xmath16mag in both fuv and nuv .
[ f_stats ] , previously discussed , and table [ t_stats ] in this work show the effect of the error cuts on the sample s statistics .
we additionally use one optical band ( _ r _ ) to separate the hot sources in `` single '' star candidates , having color [email protected] , and `` binary '' candidates having [email protected] . in more detail
, the nuv-_r _ boundary between single and binary hot stars has a complex dependence on stellar and radii of the binary pair , which we will address in a different work .
for example , a hot wd with an a - type companion , and pairs with small mass ratios will fall in the color selection of our `` single '' hot - star sample .
the adopted value is a useful cut to eliminate most extragalactic sources contaminating our fuv - nuv@xmath3 - 0.13 sample , as they have [email protected] , however it is obviously an oversimplification for a detailed study of binaries . in sum ,
our so - called ( for simplicity ) `` binary '' sample includes stellar pairs with a hot wd and a cooler star , and some qsos ; the `` single '' sample includes all single stars and some binaries .
we impose no error cut on the _ r _ magnitudes , in order to not limit the sample of the hottest wds , which are faint at optical wavelengths . if the restriction of [email protected] mag were imposed to the _
r-_band , the density of objects in the mis sample would be reduced by about 30% ( galex is ` deeper ' than sdss for very hot stars of low luminosity , see section [ s_additions ] ) ; the loss would be much smaller for the brighter ais sample .
the density of hot - star candidates is shown in fig .
[ f_lat ] and will be discussed later .
the magnitude distribution of the total hot - star candidate sample ( fuv - nuv @xmath3 -0.13 ) is shown with shadowed histograms in figures [ f_strip_few ] and beyond , and the `` single '' hot star candidates ( fuv - nuv @xmath3 -0.13 and [email protected] ) with solid - color histograms . while the sdss depth of 22.3/23.3/23.1/22.3/20.8 ab mag in _ u , g , r , i , z _ provides a fairly complete match to the ais uv source catalog , the hottest , smallest stars detected in the mis may fall below the sdss limit .
this can be guessed by comparing in fig.s [ f_strip_few ] , [ f_alllat ] , [ f_thind ] the green - dashed histogram ( matched galex - sdss sources with fuv - nuv@xmath3 - 0.13 ) with the light - green filled - color histogram ( matched sources with fuv - nuv@xmath3 - 0.13 and [email protected] , i.e. `` single '' hot stars ) .
the hottest single wds will be faintest at optical wavelengths ( e.g. a star with log g=9 .
and = 50,000/100,000 k would have fuv-_r_=-2.33/-2.61 in ab mag , according to our tlusty model grids ) .
therefore , hot wds still detectable in uv at the depth of our mis sample ( see fig .
[ f_stats ] ) may be below the sdss detection limit , while hot wds with an optically - brighter binary companion or qsos , which have redder uv - optical colors , will be detectable also in the sdss imaging .
this explains why the density of mis `` single '' hot matched sources ( light - green ) drops at faint magnitudes earlier than the dashed - green histogram , the difference between the two being larger than what can be ascribed to qso contamination ( section [ s_purity ] ) . in order to estimate the incompletess of faint hot - wd counts in our matched sample , we searched for galex mis sources with fuv - nuv@xmath3 - 0.13 that are within the sdss dr7 footprint , but do not have sdss counterpart .
we found @xmath131,500 such sources , having no optical match in our catalog with a match radius of 3@xmath10 ( section [ s_catmatch ] ) .
we performed a number of tests to verify if these are real sources .
we matched them against the sdss catalog , extending the match radius to 6@xmath10 , and found 399 additional matches , 138 of them are classified as `` pointlike '' sources at the sdss resolution .
these mostly appear to be actual sources in the images , although we note that for faint sources , the sdss classification of `` extended '' and `` pointlike '' is not always reliable , as shown by @xcite .
it is expected that a small number of sources may have optical coordinates differing by more than 3@xmath10 from the uv position ; sometimes this is due to a nearby source not fully resolved . for the remaining 1,100 uv sources with no sdss match out to 6@xmath10 , visual inspection of random subsamples
revealed some to be part of extended cirrus emission , many seem likely sources , a few cases are parts of a shredded galaxy .
the ` kron radius ' from the galex pipeline gives an indication of the spatial extent of the source : 189 sources have kron radius larger than 3.5 , hence are probably not stellar sources , but visual inspection suggests about 75% of them to be real sources .
even visual inspection however is not always conclusive , for faint sources or complicated fields , and better resolution or deeper exposures would be needed for a final sample . in sum ,
a very large fraction of the 1,500 objects unmatched within 3@xmath10 are actual sources , and an undetermined fraction may be actual hot stars . in spite of the large uncertainties , we added all these objects to the `` single '' hot matched sources catalog , and
show the total as filled - dark - green histograms in fig.s [ f_lat ] , [ f_strip_few ] , [ f_alllat ] , [ f_thind ] .
they must be considered a very generous upper limit , and only a reminder that the light - green histograms suffer from incompleteness for hot stars with nuv fainter than @xmath022mag . in order to estimate the probability that the photometrically - selected hot star candidates actually are hot stars , and the possible contamination of the sample by other types of objects
, we examined the subsamples of our catalogs for which sdss spectra exist . out of 9,028
mis matched sources with fuv - nuv@xmath3 - 0.13 and fuv , nuv error @xmath20.3mag , 810 have sdss spectra , 104 of which are sources with [email protected] , i.e. in the `` binaries '' _
locus_. of these `` binary '' candidates with existing spectra , 58 are spectroscopically classified by the sdss pipeline as stars , 4 as galaxies , 42 as qsos .
therefore , almost half of the mis sources with fuv - nuv@xmath3 - 0.13 and [email protected] could be qsos ( as found also by @xcite ) , and half could be hot stars with a cool companion . in more detail ,
qsos numbers increase at fainter magnitudes , and have fuv - nuv colors closer to our limit ( -0.13 ) than the hot wd ( @xcite ) . because stars and qso counts vary with magnitude in different ways , and especially because this statistic can be highly biased by the sdss selection of spectroscopic targets , which is of course unrelated to our present selection , and because the spectroscopic survey does not reach the depth of the mis photometric survey , we refrain from assuming a correction for the fraction of extragalactic objects in the binaries sample based on current data .
a much higher purity is found for the `` single '' hot - star candidates ( fuv - nuv@xmath3 - 0.13 and nuv - [email protected] ) . out of 4,924 in the mis sample , 706 have spectra , 703 of which are classified as stars , and two as galaxies .
only one is classified as qso by the sdss spectroscopic pipeline , but it is actually a hot wd , as shown by @xcite .
this implies a purity of almost 100% for the `` single '' hot - star candidates , down to the magnitude limit of the sdss spectroscopy at least .
again we stress however that the spectroscopic subset is serendipitous for our purpose , but not necessarily unbiased . in the ais sample , out of 28,319 total hot - star candidates ( 21,606 of which with nuv - [email protected] ) , 4448 ( 3737 ) have spectra , classified as 4075 ( 3721 ) stars , 309 ( 9 ) qsos , and 59 ( 7 ) galaxies ; corresponding to 91.6% ( 99.6% for `` single '' hot - star candidates ) purity .
the higher content of stellar sources in the ais sample , even for the `` binaries '' , is due to the brighter magnitude limit .
a crude , more direct indication that there is contamination by extragalactic sources in the `` binary '' hot - star _ locus _ , is that the fraction of `` binaries '' ( [email protected] ) among the hot sources is about 25% for the ais and 45% for the mis .
we note that the ais value of 25% is not reflecting the fraction of actual binaries , because the pairs whose components have similar are included in the `` single '' color cut , and on the other hand , ais `` binaries '' may also contain qsos . in view of the contamination by qsos in the `` binary '' hot - star sample , in the following analysis with milky way models we will consider the counts of `` single '' hot - star candidates , and assume a canonical binary fraction of 30% in the models .
n and 70 - 80@xmath9 n have only 6 and 3 square degree mis coverage in the matched footprint , therefore the statistical significance is much less than in other bins .
[ f_lat ] , width=302 ] the number of hot sources selected ( pointlike , with error @xmath20.3mag in fuv and nuv , and fuv - nuv@xmath3 - 0.13 ; see sect.[s_catalog ] ) , and of the subset with [email protected] ( `` single '' ) , are given in the last two columns of table [ t_stats ] .
stellar counts as a function of magnitude will be analyzed in the following sections .
figure [ f_lat ] shows the density of `` single '' hot - star candidates ( all magnitudes combined ) as a function of galactic latitude .
the density increases by over a factor of seven , from the poles towards the galactic plane . owing to the mw extinction ,
the counts are a lower limit at all latitudes , but especially near the galactic disk . therefore , the variation with latitude shown by this figure is less than the actual one . for the mis ,
the density of hot stars in the matched sample ( light - green histogram ) is incomplete due to the sdss limit ( section [ s_additions ] ) .
the dark - green histogram shows the mis `` single '' hot - star candidates including the hot galex sources without optical match .
sources with [email protected] are not shown due to the qso contamination ( section 4.3 ) ; adding the stellar binaries in this color range would increase the number density but not change the trend . in the next section
we analyze the density of hot stellar candidates as a function of magnitude ; the samples are divided in strips of 10@xmath9 galactic latitude , in order to examine the structure of this mw stellar component .
from the point of view of stellar evolution , the sample presented in this paper includes essentially two kinds of evolved stellar objects : the post - agb stars that have just expelled their envelopes and are crossing the hr diagram towards higher s at constant luminosity , on their way to become planetary nebule nuclei , and the hot white dwarfs which are fading both in and luminosity .
post - agb stars are very elusive because they evolve very fast , on time - scales of the order of 10@xmath17 - 10@xmath18 years ( @xcite ) .
hot white dwarfs evolve with longer timescales , but at significantly fainter luminosities .
both are elusive at all wavelengths except the uv .
our hot - star census based on the uv sky surveys provides the first opportunity to examine a comprehensive , unbiased sample of such objects .
our hot post - agb and wd candidate sample extends over a significant sky coverage , and clearly presents a disk - like distribution , concentrated towards the galactic disk ( fig .
[ f_lat ] ) . such sample will be analyzed in the context of a model for the milky way geometry below . in this section
we analyze the number counts and sky distribution of our hot star candidates , using the trilegal stellar population synthesis code ( girardi et al . 2005 ) .
trilegal creates mock catalogues of stars belonging to the milky way , and then distributes them along the line - of - sight .
it extracts the simulated stars from extended libraries of evolutionary tracks and synthetic stellar spectra , assuming reasonable prescriptions for the distributions of ages , masses and metallicities of the milky way stellar components : thin and thick disks , halo and bulge .
we note here that the bulge component does not contribute almost any source to our hot star sample , and is therefore not shown in the plots of our model results .
the thin disk density , @xmath19 , decreases exponentially with the galactocentric radius , @xmath20 , and as a squared hyperbolic secant function in the direction perpendicular to the plane , @xmath21 : @xmath22 the scale length is set to @xmath23 pc , whereas the scale heigth increases with the population age @xmath24 as @xmath25 such an increase is necessary to describe the observed increase of the velocity dispersion w with age ( see e.g. @xcite ) .
the thick disk follows the same functional form but with both scale parameters fixed as @xmath23 pc and @xmath26 pc .
the constants @xmath27 , for the thin and thick disks , are adjusted so that the surface thin disk density is equal to 59 @xmath28 , whereas the local thick disk density is of 0.0015 @xmath29 .
the halo is modeled as an oblate spheroid following a deprojected @xmath30 law , with a local density of 0.00015 @xmath31 .
finally , the sun is located at @xmath32 pc , @xmath33 pc .
the default parameters for the geometry of the mw components in trilegal are calibrated to reproduce the star counts in a local sample extracted from the hipparcos catalogue , and in several multiband catalogues including the shallow , all - sky 2mass , and a few deep surveys such as eis - deep and dms ( @xcite ) . for regions out of the galactic plane ,
i.e. for @xmath34 , errors in the star counts predicted by trilegal are typically of about 10 to 20% down to @xmath35 ( girardi et al .
, in preparation ) .
the reddening is taken into account in the milky way models as follows .
along a given line - of - sight , the value of @xmath36 at infinity is taken from the @xcite maps .
the total extinction is then distributed along the line - of - sight as if it were generated by an exponential dust layer with scale height @xmath37 ; its default value is @xmath38 pc . in this way , the closest simulated stars are unreddened , whereas those at distances of a few hundred parsecs are reddened by the full amount predicted by the @xcite maps .
we use a new version of trilegal , which has been modified for the purposes of this analysis in several ways ; most of these modifications will be detailed in a subsequent paper ( zabot et al .
, in preparation ) .
post - agb stars , planetary nebulae nuclei and white dwarfs of types da and db have been included using the evolutionary tracks from @xcite and @xcite , together with the @xcite synthetic spectra . in the present version
we have considered only the das , since they are the dominant type among hot wds @xcite .
the fraction of dbs is known to increase significantly for wds cooler than 12000 k , thanks to convection in the he layers @xcite .
such cool wds , however , are not included in our selection of sources . the trilegal code uses the koester ( 2008 ) grids to assign magnitudes to the theoretical stars . in figure [ f_models_fuv_nuv ] we show a comparison with magnitudes calculated from tlusty models , and in figure [ f_strip_few ] we compare stellar counts obatined by selecting model stars by and by synthetic colors , in order to illustrate the sometimes neglected effect of the model - atmosphere uncertanties . following our preliminary findings of discrepancies between predictions and hot star counts based on earlier data releases ( bianchi et al .
2009b ) , the trilegal code has been modified to also allow the choice of initial final mass relation ( imfr ) independent from the prescription adopted for the previous evolutionary phases . in this way
, the mass distribution of the white dwarfs does not need to follow the constraints imposed by the previous tp - agb tracks , which come from marigo & girardi ( 2007 ) .
model predictions for sample galactic latitudes are shown in figure [ f_strip_few ] , computed with default trilegal parameters for the mw geometry and the default initial - final mass relation ( imfr ) from marigo & girardi ( 2007 , hereafter mg07 ) , as well as the @xcite ifmr ( hereafter w2000 ) .
we show the thin disk , thick disk , and halo stellar components , as well as the total predicted counts .
the ifmr from mg07 alrgely overpredicts faint star counts , while the w2000 ifmr produces an overall better match with the observed counts , at all galactic latitudes ( see also fig .
[ f_alllat ] ) , and was therefore adopted in all our calculations that follow .
given the relevance of the ifmr in the context of stellar evolution , we explain in the next section how it affects the hot - star count predictions , which can be tested by our data .
-0.13 corresponds to @xmath13 18,800/17,200k ( log(g)=9.0/7.0 ) in the tlusty models , and to = 18,000k in solar kurucz - model colors with log(g)=5.0 ( not shown ) .
while tlusty and koester model colors with log(g)=7.0 agree quite well for up to 26,000k , the two grids are discrepant by @xmath00.15mag for log(g)=9.0 , where the ly@xmath39 wings are broad enough to enter the galex fuv band when convection is taken into account .
the situation reverses at hotter s . a color cut at fuv - nuv@xmath3 - 0.5 benefits from less scatter among model grids , which we could take as a measure of less uncertainty
, however the resulting sample ( the very hottest stars ) would be drastically reduced , and so the statistical significance .
[ f_models_fuv_nuv ] , width=302 ] while the milky way geometry has been derived by previous studies of low mass stars , the imfr is one of the least constrained factors in our understanding of stellar evolution , in spite of its importance for determining the yield of chemical elements , and ultimately the galaxy chemical evolution .
the reasons for this uncertainty include insufficient statistics of post - agb stars .
the final wd mass is confined to a very small range ( @xmath0 0.5 - 1.0 ) compared to the @xmath00.8 - 8range of their main sequence initial masses , and the post - agb evolutionary time - scale varies steeply within the small wd mass range .
the luminosity remains constant over a large range of ( up to @xmath4100,000k ) , and again depends on the remnant mass , making it much harder to establish their distance and absolute luminosity than for main sequence stars .
therefore , the mismatch between model predictions computed with trilegal s default parameters and observed stellar counts ( figure [ f_strip_few ] , top - left panel ) , also previously noted by @xcite , prompted us to explore different initial - final mass relations in trilegal .
although other factors , such as the wd birthrate , or the assumed mw extinction model , or a different geometry may also affect model predictions , the ifmr is a most critical aspect for modeling hot - wd counts , and was therefore explored in this first analysis .
the geometry is better constrained by low - mass stars ( e.g. girardi et al .
2005 ) , which are more numerous , and their counts based on optical - ir bands are less affected by extinction than a uv - based catalog .
the right - hand panels of fig .
[ f_strip_few ] show that the w2000 ifmr produces a better match of the model calculations with the observed stellar counts than the mg07 ifmr , and this trend is seen at all galactic latitudes .
these two ifmrs are illustrated in fig .
[ f_ifmr ] , together with additional ifmrs recently derived by @xcite and @xcite from extensive high - s / n spectroscopy of white dwarfs in galactic open clusters . despite the uncertainties related to the age - dating of the clusters , their field contamination , and the possible dependences on metallicity , they are somewhat similar to the @xcite semi - empirical ifmr .
the mg07 ifmr , instead , is derived from theoretical evolutionary tracks and , as shown in the figure , yields significantly higher final masses for all @xmath40 . in order to understand why the model - predicted hot - wd counts depend on the ifmr ,
let us first consider the distribution of wd masses in present - day surveys of the solar neighborhood .
empirical determinations have always concluded that the wd mass distribution presents a strong peak at low masses , with a maximum close to 0.55 or 0.65 and a well - defined tail of more massive wds , as well as a faint low - mass tail believed to be either observational errors or the result of binary evolution ( e.g. @xcite ) .
the origin of this peaked mass distribution can be readily derived from basic population synthesis theory ( see * ? ? ?
* ; * ? ? ?
given a volume - limited sample containing stellar populations of all ages between @xmath41 and @xmath42 gyr , the distribution of wd initial masses will be given by @xmath43 \ , \tau_{\rm wd}(m_{\rm ini } ) \ , , \label{eq_mini}\ ] ] where @xmath44 is the imf , @xmath45 $ ] is the star formation rate at the time of stellar birth @xmath46 , and @xmath47 and @xmath48 are the main sequence and wd lifetimes , respectively .
the first 2 terms in the right - hand side of this equation represent the production rate ( or , alternatively , the `` death rate '' ) of evolved stars with different masses , in number of stars per unit time .
considering a @xcite imf , @xmath49 , and a reasonably - constant star formation rate @xmath50 over the galaxy history , the equation above indicates a marked peak of the production rate at the minimum initial mass for the formation of a white dwarf , which is about 1 @xmath51 for a galaxy age of @xmath42 gyr . for higher initial masses ,
the production rate should fall as @xmath52 .
this behaviour will be shared by any galaxy model with nearly - constant star formation , independently of its ifmr .
assuming a constant and monotonic ifmr , @xmath53 , the wd mass distribution is given by @xmath54 therefore , for almost - linear and linear ifmrs like the @xcite , @xcite , and @xcite ones , the wd production rate is still expected to behave like a power - law peaked at the smallest masses .
this smallest mass corresponds to a wd mass of @xmath55 ( see fig .
[ f_ifmr ] ) .
north , and south , for four symmetric longitude quadrants .
model calculations were performed in 1@xmath9 fields evenly distributed in longitude , and combined in 90@xmath9 sections , to illustrate the dependence of the model - predicted hot stars counts on longitude .
the strongest differences are seen between center and anti - center directions , as expected , and the north - south asymmetry is most prominent in directions towards the galactic center .
lines show thin disk ( dotted ) , thick disk ( dash - dotted ) , halo ( dashed ) and total counts ( solid ) .
[ f_long ] , title="fig:",width=302 ] north , and south , for four symmetric longitude quadrants .
model calculations were performed in 1@xmath9 fields evenly distributed in longitude , and combined in 90@xmath9 sections , to illustrate the dependence of the model - predicted hot stars counts on longitude .
the strongest differences are seen between center and anti - center directions , as expected , and the north - south asymmetry is most prominent in directions towards the galactic center .
lines show thin disk ( dotted ) , thick disk ( dash - dotted ) , halo ( dashed ) and total counts ( solid ) .
[ f_long ] , title="fig:",width=302 ] .
[ f_thind ] , width=302 ] the wd lifetime @xmath56 , defined as the total time after the agb in which the star will be hotter than 18,000k , is longer for wds of higher mass , because , although they evolve much faster in the constant - luminosity phase , they reach higher than lower - mass wds , and spend a longer time in their wd cooling track before fading below the teff=18000k limit . models using different ifmrs simply associate the final masses , hence also these @xmath56 lifetimes , to different initial masses , without affecting the wd production rate .
the end result in terms of mass distribution @xmath57 ( see equation 4 ) is that smaller wd masses will weigh the distribution towards shorter @xmath48 , hence decreasing the numbers of observed wds with @xmath418000k .
this effect explains the higher number of hot wds predicted by the mg07 ifmr respect to the w2000 ifmr ( fig .
[ f_strip_few ] ) .
in fact , as can be inferred from fig .
[ f_ifmr ] , the w2000 ifmr gives significantly lower wd final masses ( hence shorter @xmath56 ) for the interval of initial masses between 0.6 and @xmath58 .
because of the ifm being skewed towards lower masses , these objects represent numerically the majority within the whole mass range .
figure [ f_ifmr ] ( right panel ) shows the distribution of initial and final wd masses derived from trilegal for a sample latitude , using either the mg07 or w2000 ifmr .
it shows that most of the hot wds observed by galex are expected to derive from low - mass stars , with initial masses below @xmath58 .
moreover , it indicates that the bulk of observed wds have low masses , typically @xmath59 in the case of the w2000 ifmr .
there is no direct confirmation of this particular prediction , although it is in line with wd mass determinations derived from spectroscopic surveys of more limited samples ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the calculations with mg07 ifmr
the distribution of predicted wd masses is quite different from the @xcite case , with a broad mass distribution extending from 0.55 to 0.65 , followed by a sort of gap , and a second peak of wd masses located slightly above 0.7 .
these features reflect the positive slope of this ifmr up to @xmath60 , and its plateau at @xmath61 for @xmath62 ( see fig . [ f_ifmr ] -left ) .
another particularity of this ifmr is the high masses ( about 0.64 ) predicted for the wds belonging to the halo and thick disk .
this prediction conflicts with recent mass determinations of wds in old star clusters , which indicate values close to 0.53 @xcite .
a similar trend is seen at different latitudes .
overall , an ifmr such as that of w2000 produces predicted counts closer to the observed numbers , at bright magnitudes where our sample is complete , and at the same time mass distributions in agreement with those derived from spectroscopy of nearby wds .
these aspects underscore the difficulties of deriving the ifmr from theoretical evolutionary tracks of agb stars , even when these tracks are directly calibrated with observations of agb stars in the local group , as is the case of mg07 .
this work , and future analyses planned on a wider sky area to explore concurrently the effects of the mw dust geometry , suggests that a deeper investigation of agb evolution is necessary , especially concerning the theoretical prescriptions for mass loss , to reconcile constraints provided by wds in our galaxy and those provided by agb stars in nearby galaxies .
we finally note that even models calculated with w2000 ifmr tend to predict more faint - star counts than our `` single '' wd sample ( fig .
[ f_alllat ] ) .
the mismatch is significant if we consider the red model lines ( model stars selected by their synthetic fuv - nuv color ) , which should be the most consistent with the observed sample selection , to the extent that the model atmospheres used by trilegal are correct .
therefore , we explored also the case of a @xcite ifmr : the predicted counts are lower at faint magnitudes ( especially the halo component fainter than nuv@xmath022 ) , as expected from fig .
[ f_ifmr]-left , while still generally in line with the bright star counts .
because of the incompleteness of our sample at magnitudes fainter than @xmath021 , and other factors that can not be conclusively constrained with the current sample , we defer more quantitative conclusions which rests on a comparison of faint stars to a future work , and we adopt in this work the w2000 ifmr , that matches well the bright star counts , to explore other effects .
as explained above , the default parameters , in particular the mw geometry , in the trilegal code were defined from previous applications of this code to analysis of stellar counts from several surveys . particularly important in this context
were the 2mass star counts , which are sensitive to the low - mass stars ( @xmath30.8 ) .
these low - mass stars do not overlap with the mass range of the hot - wd progenitors ( about 0.8 - 8 ) , however they are very numerous and long - lived , and are little affected by uncertainties in stellar evolutionary models
. therefore we should expect the geometry to be well constrained by previous works , except for the limited extent of the previous surveys .
the agreement between our data and model predictions ( with w2000 ifmr ) is generally good at bright magnitudes , where our sample is complete . in more detail
, we should also explore the dependence of observed and predicted stellar counts with longitude .
the effect is more prominent towards the galactic plane , therefore we examined latitudes between 20 - 30@xmath9 north and south , and computed models with evenly spaced coverage along these strips .
the models predict , as expected , different counts towards the galaxy center and anti - center , shown in fig .
[ f_long ] .
the coverage of our present galex - sdss matched catalog at low latitudes is still too sparse for a conclusive comparison , which will be instead attempted by matching the ais data with gsc2 , and with the increased mis coverage expected in the galex `` extended mission '' phase . finally , because our uv - optical matched samples are complete to uv magnitudes brighter than @xmath020 - 21 , therefore mostly sensitive to the thin disk population ( fig.s [ f_strip_few ] and [ f_alllat ] ) , and because the models with default mw geometry slightly overpredict the stellar counts at low ( especially northern ) galactic latitudes , we explored different geometries of the thin disk with trilegal . in particular , in fig .
[ f_thind ] we show the effect of adopting more strongly peaked density distributions at the galactic plane : either the simple vertical exponential law , @xmath63 ( which has been adopted in many mw models , starting from the classical @xcite s ) , or the simple hyperbolic secant law , @xmath64 .
the results confirm that the default geometry adopted by @xcite and in this work , i.e. the squared hyperbolic secant law , provide the closest prediction to the observed counts .
the same distribution does also provide a good match to star counts of `` normal stars '' in 2mass and in the local hipparcos sample @xcite .
the slight overprediction and underprediction of hot star counts at certain latitudes may also be due to extinction by interstellar dust , which may be either patchy ( as in fact it is ) or have a slightly different distribution than assumed .
these remaining discrepancies and the effects of dust distribution may be addressed with a wider mis coverage in the future .
the plots of stellar density versus galactic latitude presented in the previous sections clearly indicate that we are looking at a population of objects for which the mean distance is larger than the mean scale height on the thin disk . at typical distances of a few hundred parsecs , there will be no parallax measurements for such stars .
the most direct distance information we can find for them is related to their proper motions .
the table `` usno '' in the sdss database contains proper motions for sdss objects made after recalibrating usno - b1.0 ( @xcite ) against sdss astrometry .
the systematic errors are quoted to less than 1 mas / year .
the ucac3 proper motion catalog is strongly incomplete for our sample , owing to its magnitude limit of @xmath65 ( which roughly corresponds to @xmath66 ) .
in fact , we checked and found that just a handful of our hot - wd candidates is present in ucac3 .
figure [ f_pm ] presents the histograms of proper motions , @xmath67 , for the hot stars in the 20 - 30n strip of mis , and for three different cuts in nuv magnitude .
the observed distributions are compared with those predicted from the trilegal models , computed for the same area and for the w2000 ifmr .
the model distributions are derived as follows .
the observed velocity ellipsoids @xmath68 of each galaxy component are taken form the literature .
each simulated star is then given a random space velocity that follows the schwarzschild distribution for this ellipsoid .
space velocities are then corrected for the solar motion ( cf . dehnen & binney 1998 ) and projected on the sky using the transformations derived from hipparcos .
the final results are simulated proper motions and radial velocities .
we adopt an age - dependent velocity ellipsoid for the thin disk , derived from the geneva - copenhagen survey of solar neighbourhood ( holmberg et al .
2008 ) . for the thick disk and halo , velocity ellipsoids are taken from layden et al .
these prescriptions provide a first - order description of the general distributions of proper motions and radial velocities , but are not expected to describe the details of the local velocity streams , nor the effects caused by disk shear in the more distant stars .
moreover , we do not simulate measurement errors in the proper motions .
more details will be given elsewhere .
the upper panel of figure [ f_pm]-right compares the observed and model distributions for a bright wd sample , with a cut at nuv@xmath318mag .
there is quite good agreement both in the numbers and distributions of proper motions .
the agreement in the numbers derives from the fact that the photometry is quite complete , and just 4% of this bright subsample does not have a proper motion listed in usno - b .
this means that both photometry and proper motion information are very complete in this case , and quite similar to the ones expected for a thin disk population .
the middle and bottom panels of fig . [ f_pm]-right compare the observed and simulated @xmath67 distribution for fainter samples , with cuts at [email protected] and [email protected] , respectively .
it shows that these samples become incomplete , first in the proper motions at nuv=19.5 , and then in photometry at magnitudes nuv=21.5 .
note for instance the increasing fraction of stars without proper motion measurement in usno - b , which goes from 13 to 54% from [email protected] to [email protected] . from this comparison
we conclude that our hot wds do follow the kinematics expected for the thin disk , even if the proper motion information is severely incomplete for the faintest wd candidates .
distance distributions predicted by the same models are shown in fig .
[ f_pm]-left .
this plot indicates that the brigthest ( and complete ) mis sample of nuv@xmath318mag is likely made of wds located in the thin disk at typical distances between 0.15 and 1 kpc in the latitude range 20 - 30@xmath9 north .
the fainter sample could instead contain a significant fraction of stars at distances larger than 2 kpc , comprising also a small fraction of thick disk and halo wds .
note however that this distribution of distances is inferred from models , and we presently have no data to verify the distances of our sample stars , but for the proper motions already depicted in fig .
[ f_pm ] ( right ) ; future work will address identification of thick disk and halo objects with follow - up spectroscopy .
the distance distribution is fairly similar at other latitudes .
finally , fig.[f_kine ] presents the expected distributions of radial velocities for the latitude 20 - 30n sample , from the models , and the radial velocities derived from sdss spectra , available for a subsample of our hot star catalog ( section [ s_purity ] ) .
this is a biased subset because the choice of sdss spectroscopic targets is obviously unrelated to our selection of hot stellar objects .
the worse effect however , making the comparison inconclusive , is the large errors in velocity determinations from the sdss pipeline ( see fig . [
f_kine]-right ) .
from galex s data release gr5 we constructed catalogs of unique uv sources ( i.e. eliminating repeated observations ) , and subcatalogs with matched sdss optical photometry .
we extracted catalogs of hot star candidates ( fuv - nuv@xmath3 - 0.13 ) , mostly comprised of hot wds .
over 38,000 such sources with photometric error @xmath20.3mag , have sdss photometry ( @xmath074,000 with uv photometric error @xmath20.5mag ) , excluding uv sources with multiple optical matches , that are between @xmath08 to 30% of the total .
the galex surveys ais and mis cover different latitudes at their respective depths ( @xmath021 and @xmath023 ab mag ) , enabling a first quantitative analysis of the hot - wd stellar population with milky way models .
current descriptions of mw stellar components ( halo , thin and thick disk ) are mostly based on stellar counts of low mass ( @xmath30.8 ) stars ( 2mass , shallow but all - sky , plus a few deep but area - limited surveys ) .
we explored in this work different thin disk geometries and different ifmr .
galex provides an unprecedented census of the evolved descendants of 0.8 - 8stars , and the analysis of our current hot - wd sample places some constraints on the initial - final mass relation ( ifmr ) , one of the crucial ingredients to understand the chemical enrichment of the ism .
the brightest ( and complete ) sample is likely made of hot wds located in the thin disk at typical distances between 0.15 and 1 kpc .
the faintest samples could instead contain a significant fraction of stars at distances larger than 2 kpc , comprising also hot wds in the thick disk and halo .
this work is based on the analysis of matched galex - sdss sources , and the sdss magnitude limit prevents analysis of thick - disk and halo components .
hot wds are however detectable at such larger distances at mis depth ( see dark - green histograms in fig.s [ f_strip_few ] and [ f_alllat ] , and section [ s_additions ] ) , and will be investigated in a future work with deeper optical data .
model predictions of hot star counts at different latitudes , computed with currently accepted mw geometry , and assuming w2000 ifmr , match the data quite well at intermediate latitudes and bright magnitudes , but are somewhat discrepant at low latitudes and faint magnitudes .
there is a slight but systematic north - south asymmetry , and mw models with canonical geometry for stars and dust match better the data at southern low latitudes , and at northern high latitudes , indicating that the dust distribution on the mw disk may also be better constrained by our data than was previously possible .
the dust distribution will be investigated with the future more extended sky coverage from the ongoing galex mission .
however , the observed asymmetry may also be due to the longitude dependence of the stellar populations , as our present matched - sources sample covers unevenly north and south galactic latitudes ( fig .
[ f_coverage ] ) .
our mis matched - sources sample is incomplete at magnitudes fainter than about @xmath021.5 due to the sdss limits for the hottest sources . at all latitudes , and at magnitudes where the samples are complete
, mw model predictions of hot wd counts computed with ifmrs favouring lower final masses match the galex hot wd counts significanly better than the ifmr currently postulated by marigo & girardi ( 2007 ) from agb stars data .
this result contributes an important constraint to stellar evolution in the final phases which drive the yield of chemical elements .
what are the implications of constraining the ifmr for the galaxy chemical evolution ? the ifmr contains the record of the nucleosynthesis history during the tp - agb phase , as well as determines the contribution of agb stars in terms of dark remnants .
the ifmr , and the final yield of chemical elements , are significantly affected by the occurrence ( or not ) of the so - called `` third dredge - up '' . at each dredge - up episode
the core mass is reduced by an amount @xmath69 , where @xmath70 is the classical efficiency parameter , and @xmath71 denotes the mass growth of the core during the preceeding interpulse period . in the extreme case where @xmath72 , no increment of the c - o core is predicted . as a consequence , for low - mass stars
not experiencing the third dredge - up ( @xmath73 ) , the remnant mass scales directly with the duration of the tp - agb lifetime , while for stars in which the third dredge - up takes place ( @xmath74 ) this simple proportionality is replaced by a more complex interplay between core growth , dredge - up , and mass loss .
for such more massive stars ( @xmath75 , where @xmath76 , depending on metallicity and model details ) the final mass and the chemical yields are related as we will briefly discuss below ( see marigo & girardi 2001 for more details ) . for the generic element @xmath77 , the corresponding agb yield is defined as : @xmath78 \dot m(t)\ , d t \label{eq_yield}\ ] ] where @xmath79 is the initial abundance at the epoch of star formation , @xmath80 is the value at time @xmath24 , and @xmath81 is the current mass - loss rate .
let us first consider , in the ifmrs in fig .
[ f_ifmr ] , the part for @xmath82 .
the chemical yields produced by these stars ( mainly @xmath83c , @xmath84n , and @xmath85he ) are determined by the first dredge - up , and the cool - bottom process ( boothroyd & sackmann 1999 ) or thermohaline mixing ( charbonnel & zahn 2007 ) operating during the rgb , as no subsequent dredge - up events take place on the agb according to the standard stellar evolution theory . for these low - mass stars
the agb yield , defined by eq .
( [ eq_yield ] ) , simplifies as @xmath86 \delta m_{\rm agb}^{\rm ej}$ ] , where @xmath87 is the surface abundance left after the rgb phase , and @xmath88 is the total amount of mass ejected on the agb . in turn
, the expelled mass is related to the final mass through @xmath89 , where @xmath90 denotes the stellar mass at the onset of the agb phase .
the final masses of these stars with @xmath82 can be reduced by simply invoking more efficient mass - loss along the agb .
their chemical yields would increase correspondingly , since less envelope mass will be locked in the central core following the outward displacement of the h - burning shell , so that @xmath91 will be larger .
for instance , for an agb mass of @xmath92 , a final mass @xmath93 versus @xmath94 would imply an increase of the agb chemical yield by @xmath95 , as well as a larger total chemical yield , including the rgb wind contribution . for @xmath96 , in addition to the first dredge - up ( and possibly second dredge - up on the early - agb ) , the chemical yields include also the enrichment , mainly in @xmath85he and primary @xmath97c , produced by the he - shell flashes and brought up to the surface by the third dredge - up , and the enrichment of @xmath84n produced by the hot - bottom burning process occurring in the most massive agb stars ( @xmath98 ) .
again smaller final masses are reached when higher mass - loss rates occur , and/or deeper dredge - up ( i.e. larger @xmath70 ) .
these two factors may affect the chemical yields in opposite ways .
on one hand , a high mass loss limits the number of dredge - up episodes , hence lowering the chemical yields ; on the other hand , a deeper dredge - up conveys larger amounts of carbon and helium to the surface layer , which are then ejected .
in addition , mass loss and dredge - up efficiency are actually inter - related , a larger surface c abundance generally favouring stronger dust - driven mass loss in c stars ( e.g. mattsson et al .
2008 ) , while a significant reduction of the mass of the envelope may decrease its penetration ( i.e. lower @xmath70 ) during the third dredge - up ( see , e.g. karakas et al .
2002 ) . in sum
, a higher mass expelled does not directly imply that such material is relatively more enriched of processed elements .
an ifmr similar to weidemann ( 2000 ) weighs the final masses towards lower values than e.g. the ifrm of mg07 in the range @xmath99 , while the two relations essentially agree at higher @xmath100 . for @xmath101 , forcing the theoretical ifmr to converge on a w2000-type ifmr by assuming a higher mass loss would produce a modest increase of the chemical yields from these objects ( because more mass is ejected ) , compared to the mg07 imfr case . as for the initial mass range @xmath102 , a higher mass loss ( necessary to reach a smaller final mass , as suggested by our wd counts ) may reduce the number of dredge - up episodes on the tp - agb and this may lead to a lower yield of processed material .
in other words , constraining the ifmr translates into constraining the total mass lost in the agb phase .
the chemical composition of the ejecta from these stars , however , is also critically sensitive to the depth of the third dredge - up , which can be calibrated by measurements of abundances in planetary nebulae ( marigo et al 2003 ) and by a better estimate of the duration of the agb phase at different metallicity ( marigo et al . in prep . ) . with the current mis coverage ( 1,103 square degrees overlapping with sdss photometry in gr5 ) , we compared with mw models stellar counts summed in 10@xmath9 latitude strips , combining all longitudes in order to obtain acceptable statistics .
in order to disentangle geometry of the stellar populations , dust extinction , and stellar evolution in more detail , we need to analyze also the longitude dependence of stellar counts , as demonstrated by fig.s [ f_alllat ] and [ f_long ] . we will explore dust effects in the mw disk by matching the ais with gsc2 , for a larger sky coverage ( see @xcite ) .
the density of hot stars ( the rarest in nature ) is very low and large area coverage is needed for good statistics . the mw halo and thick disk components , however , become significant at uv magnitudes fainter than about 21 , therefore require a wider mis coverage ( planned in the galex extended mission phase ) to be usefully constrained ( see fig.s [ f_coverage ] , [ f_strip_few ] and [ f_alllat ] ) , and deeper optical surveys .
finally , hot - wds in binaries , again elusive at optical wavelengths , are uniquely and unambiguously revealed and characterized by the galex uv sky survey , matched to optical surveys .
these binaries sample different types of stellar pairs ( different ranges ) than those detected and characterized by optical surveys , thanks to the uv photometry ( e.g. fig.6 of @xcite ) . because our current photometric selection includes contamination by extragalactic objects in the `` binary '' _ locus _ , and on the other hand the binaries do not significantly change the distribution of stellar counts , which was the subject of the present analysis
, the binary candidates will be investigated in a future work .
the present catalog of `` single '' hot star candidates has a high purity , as estimated from the serendipitous ( but not unbiased ) spectroscopic sdss coverage .
while we expect most of the hottest sources to be high gravity objects ( cspn , subdwarfs or wd ) , we did not apply any cut in gravity , therefore main sequence stars and supergiants hotter than @xmath018,000k are also included in our catalogs .
their number is very small compared to the evolved stars , however their inclusion in the catalog makes it more generally useful .
finally , we remind future users of our catalogs that galex uv sources with more than one optical counterpart were excluded from the analysis of the hot - star sample , ( and should be excluded from any sample based on photometric color selection ) , but their fraction can be corrected statistically using table [ t_stats ] .
they are included ( flagged with `` rank''=1 ) in our online catalogs
. a few percent of the matched sources may be spurious matches ( fig .
[ f_exptime2 ] ) but these would mostly have random uv - optical colors .
the hot - star candidates with [email protected] , includes also some extragalactic objects , the relative fraction is magnitude and latitude dependent .
the source catalogs ( galex unique sources , galex - sdss matched sources , hot - stars samples ) are available in electronic form only , from the author s web site ( http://dolomiti.pha.jhu.edu/uvsky ) where files description can be found , and will be also posted on mast ( http://galex.stsci.edu and http://archive.stsci.edu/hlsp/ ) .
a statistical analysis and discussion of properties , useful to potential users of our ( and similar ) catalogs for understanding completeness and biases of any sample selection is given by @xcite .
data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
galex ( galaxy evolution explorer ) is a nasa small explorer , launched in april 2003 .
we gratefully acknowledge nasa s support for construction , operation , and science analysis of the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology . l.b . and j.h .
acknowledge partial support from fuse gi grant h901 ( nasa nnx08ag97 g ) .
a.z . acknowledges financial support from cnpq - mct / brazil . l.g . acknowledges partial funding by from contract asi - inaf i/016/07/0 .
we are very greateful to a. thakar for discussions of very many issues regarding the sdss database .
bianchi , l. 2011 , apss , in press ( special issue `` uv universe 2010 '' , editors b. shustov , a.i .
gomez de castro , and m. sachkov ) bianchi , l. 2009 , apss , 320 , 11 bianchi , l. , herald , j. , efremova , b , et al 2011 , apss , in press ( special issue `` uv universe 2010 '' , editors b. shustov , a.i .
gomez de castro , and m. sachkov ) bianchi , l. , hutchings , j.b .
, efremova , b. et al .
2009a , aj , 137 , 3761 bianchi , l. , efremova , b. , herald , j. et al .
2009b , in `` future directions in ultraviolet spectroscopy '' , eds.van steenberg et al aipc , 1135 , 326 bianchi , l. 2007 , in uv astronomy : stars from birth to death , eds . a. gomez de castro & m. barstow , isbn 978 - 84 - 7491 - 852 - 6 , p. 65
bianchi , l. , et al .
2007a , apjs , 173 , 659 bianchi , l. , et al . 2007b , in `` uv astronomy :
stars from birth to death '' , ed .
a. gomez de castro & m. barstow , isbn 978 - 84 - 7491 - 852 - 6 , p.95 , s. o. , kleinman , s. j. , nitta , a. , koester , d. , castanheira , b. g. , giovannini , o. , costa , a. f. m. , & althaus , l. 2007 , mnras , 375 , 1315 kleinman , s.j , nitta , a. , & koester , d. 2009 , journal of physics : conference series , volume 172 , issue 1 , p. 2020 | we present comprehensive catalogs of hot star candidates in the milky way , selected from galex far - uv ( fuv , 1344 - 1786 ) and near - uv ( nuv , 1771 - 2831 ) imaging .
the fuv and nuv photometry allows us to extract the hottest stellar objects , in particular hot white dwarfs ( wd ) , which are elusive at other wavelengths because of their high temperatures and faint optical luminosities .
we generated catalogs of uv sources from two galex s surveys : ais ( all - sky imaging survey , depth [email protected]/20.8 in fuv / nuv ) and mis ( medium - depth imaging survey , depth @xmath022.6/22.7mag ) .
the two catalogs ( from galex fifth data release ) contain 65.3/12.6 million ( ais / mis ) unique uv sources with error@[email protected] , over 21,435/1,579 square degrees .
we also constructed subcatalogs of the uv sources with matched optical photometry from sdss ( seventh data release ) : these contain 0.6/0.9million ( ais / mis ) sources with errors @xmath20.3mag in both fuv and nuv , excluding sources with multiple optical counterparts , over an area of 7,325/1,103 square degrees .
all catalogs are available online .
we then selected 28,319 ( ais ) / 9,028 ( mis ) matched sources with fuv - nuv@xmath3 - 0.13 ; this color cut corresponds to stellar hotter than @xmath018,000 k ( the exact value varying with gravity ) .
an additional color cut of [email protected] isolates binaries with largely differing s , and some intruding qsos ( more numerous at faint magnitudes ) .
available spectroscopy for a subsample indicates that hot - star candidates with [email protected] ( mostly `` single '' hot stars ) have negligible contamination by non - stellar objects .
we discuss the distribution of sources in the catalogs , and the effects of error and color cuts on the samples .
the density of hot - star candidates increases from high to low galactic latitudes , but drops on the mw plane due to dust extinction .
our hot - star counts at all latitudes are better matched by milky way models computed with an initial - final mass relation ( ifmr ) that favours lower final masses .
the model analysis indicates that the brightest sample is likely composed of wds located in the thin disk , at typical distances between 0.15 - 1kpc , while the fainter sample comprises also a fraction of thick disk and halo stars .
proper motion distributions , available only for the bright sample ( nuv@xmath318 mag ) , are consistent with the kinematics of a thin - disk population .
[ firstpage ] astronomical data bases : catalogues stars : white dwarfs stars : evolution galaxy : stellar content ultraviolet : stars galaxies : milky way |
an important step toward a complete understanding of the formation , structure , and composition of interstellar dust is the study and discrimination of factors that produce changes in the observed wavelength - dependent extinction of stars due to grains .
the large and small magellanic clouds ( lmc and smc , respectively ) are metal - poor relative to the galaxy ; recent estimates imply that the lmc and smc have overall metallicities at levels 0.5 and 0.20.25 , respectively , of the galactic ism @xcite .
dust - to - gas ratios determined for these galaxies , as represented by @xmath4 ( ) , are also reduced ( by factors of about 4 and 10 ; @xcite ) , indicating that lmc and smc dust components and their formation mechanisms are not inconsistent with galactic components and mechanisms .
nevertheless , there are differences in extinction among these three galaxies .
early investigations of the extinction properties of magellanic cloud dust identified several sight lines that exhibited marked differences from curves produced by grains in the milky way ( e.g. , @xcite ) .
it was noted in comparing these curves that the 2175 bump strength was reduced in the lmc relative to the galaxy , and somewhat further diminished for curves characteristic of the smc .
similarly , the fuv rise evident in lmc and smc extinction curves is stronger than that common in the galaxy .
subsequent studies have identified regional variation in magellanic cloud curves , associating a distinctive lmc wavelength dependence with the supershell lmc2 @xcite , a steeper uv curve with the star - forming bar region of the smc , and extinction more closely resembling galactic curves with other portions of the clouds .
recently , @xcite ( hereafter g03 ) completed a comprehensive comparison of galactic , lmc , and smc extinction curves from near - infrared to ultraviolet ( uv ) wavelengths ( @xmath5 1150 or @xmath6 ) ; g03 confirmed that curves characteristic of the lmc2 grouping and the smc bar do not conform to the @xcite ( ccm ) parameterization of galactic extinction based on @xmath7 $ ] .
the strength of their uv extinction is a key feature distinguishing lmc , smc , and galactic curves ; in particular , the steeper short - wavelength slopes associated with magellanic cloud extinction curves can be diagnostic of differences in dust compositions and grain size distributions , given appropriate measures of elemental abundances in their respective interstellar media @xcite .
however , the extension of magellanic cloud extinction curves into the fuv ( @xmath8 8.7 @xmath9 ) is complicated by the requirement that interstellar h@xmath0 absorption be removed from the spectra before an extinction curve can be produced using the pair method .
previous efforts in this area , even for milky way sight lines , have been hampered by a lack of high quality data .
in particular , instrumental issues of scattered light , time - variable sensitivity , and a limited sample ( both target and comparison objects ) restricted the utility of the _ copernicus _ dataset ( cf .
@xcite ; @xcite ) , the _ voyager _ uvs data suffer from low resolution which prevents explicit identification and removal of the molecular hydrogen contribution , and the remaining available data [ e.g. , rocket - @xcite ; _ hopkins ultraviolet telescope ( hut ) _ - @xcite ; _ orbiting retrievable far and extreme ultraviolet spectrometer ( orfeus ) _ - @xcite ] include only a few galactic sight lines .
_ far ultraviolet spectroscopic explorer ( fuse ) _
, however , has revolutionized our ability to study interstellar molecular hydrogen .
its combination of slightly better spectral resolution and much greater detector sensitivity than _ copernicus _ allows h@xmath0 to be detected throughout much larger and more diverse portions of the galaxy and even in the magellanic clouds @xcite .
thus , _ fuse _ data are uniquely suited to investigating the fuv extinction properties of milky way , lmc , and smc sight lines and what any differences reveal about the dust populations in each galaxy .
recently , @xcite used a modified maximum entropy method ( mem ) to fit extinction components to ir through uv curves for stars both in the galaxy and in the clouds ; they demonstrated that the average smc bar extinction curve was best fit using a grain distribution in which small silicate and amorphous carbon grains play a much larger role than in the mean galactic extinction .
however , this result is at odds with @xcite , whose data suggested negligible silicon depletion in several gas cloud components along a sight line probing the smc ism . in order to address issues such as dust composition , particularly for small grains enhancing fuv extinction in the clouds
, it would be helpful to explore lmc and smc extinction curves shortward of the uv range that @xcite were confined to by their _ international ultraviolet explorer ( iue ) _ data ; unfortunately , fuv extinction curves have been published to date for only very few reddened magellanic cloud stars @xcite . in this paper
, we construct new extinction curves for nine pairs of reddened and unreddened lmc and smc stars by supplementing the ir through uv curves published by g03 with recent _ fuse _ observations of 16 magellanic cloud targets .
the _ fuse _ data allow us to extend the @xcite analysis of dust composition and grain size distribution to fuv wavelengths , where extinction is dominated by small grains and the differences between galactic and magellanic cloud curves are particularly prominent .
this is the second in a series of three papers exploring extinction in the fuv .
paper i @xcite dealt with a small set of galactic extinction curves characterized by a broad range of @xmath10-values ; paper iii ( in preparation ) will examine a much larger sample of galactic sight lines in an effort to search rigorously for any trends among observed extinction parameters when fuv data are fully considered .
the present sample of 16 lmc and smc stars includes all stars in the _ fuse _ archive for which the spectral match between the reddened and unreddened stars has been rigorously evaluated and there exist spectral data from ir to uv wavelengths .
each reddened star has previously been studied by g03 ; we use the same photometry sources and procedures to construct the ir to uv portions of the extinction curves for the pairs adopted in this study , with the exception that early 2mass photometry has been superseded in our analysis by data from the recent _ all sky release_. in particular , it should be noted that the @xmath10 values listed in table [ exttab ] , which summarizes the properties of each extinction pair , differ slightly from those published by g03 due to this update of the ir data .
consequently , the values of @xmath11 and @xmath2()/@xmath11 in the table are also affected .
the fuv portions of the extinction curves ( shortward of the ly@xmath12 line ) were derived from the new and archival _ fuse _ observations listed in table [ obstab ] .
all of the raw data were processed using the latest release of calfuse ( v3.0 ) .
the calibrated spectra for each channel were cross - correlated , shifted to a common wavelength scale , and combined based on exposure time in cases where several observations of a star existed . also ,
in cases where the different observations or different _ fuse _ channels produced conflicting flux levels for a given wavelength ( e.g. , due to channel misalignment ) , the data for these observations or channels were linearly adjusted to match the maximum observed flux across the regions of overlap .
general details on _ fuse _ observations have been published by @xcite .
we point out here , however , that each observation produces eight spectra nearly 100 in length with individual dispersion characteristics .
the lif channel spectra ( lif1a , lif1b , lif2a , lif2b ) cover most wavelengths from 9901190 twice over ; the spectral region from 9101110 are similarly sampled by the sic channels ( sic1a , sic1b , sic2a , sic2b ) .
sample spectra of both poor and fine quality appear in figure [ spectra ] , using the data from the reddened star azv 456 and its comparison star azv 70 , respectively .
strong fuv h@xmath0 absorption bands can significantly alter the appearance of stellar spectra , depending upon the column density of molecular gas along the line of sight .
consequently , a molecular hydrogen model was constructed for each sight line using procedures patterned generally after those used by @xcite .
the underlying assumption we made in measuring the interstellar h@xmath0 for each sight line was that the model includes only two velocity components , one corresponding to milky way gas and one associated with h@xmath0 in the appropriate magellanic cloud .
the equivalent width measurements for distinct profiles of each component were entered into separate tables , so that molecular hydrogen measurements could be made individually for each galaxy using a curve - of - growth ( cog ) analysis .
unfortunately , the velocity separation of galactic and magellanic cloud components often resulted in blended absorption profiles at several points in a given spectrum . in general , the construction of a reasonable cog for most of the sight lines required more equivalent width measurements than were available from the unblended and isolated h@xmath0 absorption lines . to remedy this problem , the profile - fitting code fits6p - values , and velocities of input interstellar components .
more details on the algorithm can be found in @xcite . ] and iraf plotting routines were used in conjunction with our idl code @xcite to deconvolve blended profiles and fill out each cog with as many measurements as could reliably be made .
the cog analysis , however , was generally limited to @xmath13 because the @xmath14 lines were saturated to the extent that they possessed broad damping wings that the equivalent width measurement algorithm could not reliably distinguish from the stellar continuum .
thus , an iterative curve - fitting procedure was an additional requirement so that simultaneous determinations of galactic and magellanic cloud column densities for these lower h@xmath0 rotational excitation levels could be made .
the complete molecular hydrogen models , including column densities for the available @xmath15 levels , the associated @xmath16-value , and the derived kinetic and excitation temperatures ( these last properties are discussed in section [ ssection_h2 ] ) , are summarized in tables [ mwmolhydtab ] and [ mcmolhydtab ] for all of our lmc and smc sight lines ; table [ mwmolhydtab ] characterizes the galactic component for each sight line where one was detected , while table [ mcmolhydtab ] details the properties of each magellanic cloud component .
once molecular hydrogen absorption models had been constructed , each spectrum was corrected for this absorption .
all of the _ fuse _ data for each star were combined into a single spectrum by shifting the calibrated channel spectra to a common wavelength reference and averaging the fluxes , weighted by the flux uncertainty at each point in the overlapping regions .
although there are significant differences in the data dispersion properties for each channel , the merged spectrum for each star is of sufficient quality for the construction of a reliable extinction curve .
the `` worm '' problem in _ fuse _ data @xcite was strong only in lif1b .
its appearance in the data for this detector segment was identified by a comparison with the lif2a flux as the data for all detector segments were merged into a single spectrum for each sight line ; the portions of the lif1b spectrum contaminated by the worm were eliminated from the merging process .
a running cross - correlation of the normalized h@xmath0 absorption profile corrected for any wavelength mismatches , and the full final model was divided into the merged _ fuse _ spectrum to complete the `` removal '' of the molecular hydrogen features .
atomic hydrogen was not measured for each sight line individually , but its signature , specifically the ly@xmath12 and ly@xmath17 lines , was removed after the ratio of the reddened and comparison spectra was calculated as a function of wavelength . although _ fuse _ spectra cover wavelengths from 905 through 1187 , data shortward of 1044 ( our ly@xmath17 cutoff ) were not included in this analysis due to limitations in data quality .
in the interest of consistency in instrument response , _ fuse _ flux levels were then manually rescaled to match _ iue _ fluxes over the wavelength range common to both instruments ( 1150 @xmath18 1185 ) when the full ir to fuv spectra were constructed ; given recent improvements to the calibration of _ iue _ data @xcite , flux errors associated with this instrument have not yet been surpassed by the latest _ fuse _ calibration @xcite .
generally , mean fluxes for the data from each instrument agreed to within 20% , although there were a few targets for which the discrepancies exceeded this range .
the scalings applied to each _ fuse _ spectrum ( the ratio of _ iue _ to _ fuse _ flux in the overlap region ) are given in table [ fluxshifts ] .
we have constructed extinction curves using the pair method ( e.g. , @xcite ) , normalized to @xmath19 through @xmath10 ; the pairs of reddened and unreddened stars in the lmc and smc were selected ( g03 ) and are listed in table [ exttab ] . in an effort to minimize uncertainties in each extinction curve , particularly from spectral mismatch , a detailed comparison of the _ fuse _ spectrum for each star with similarly - typed unreddened candidates was performed , and the extinction pairs matched by g03 using _ iue _ were also found to be well - matched in the fuv .
assembly of the extinction curve corresponding to each pair was accomplished using the same procedures and near - infrared through uv data outlined by g03 .
we have also constructed average extinction curves for the subsets of the current sample associated with the lmcavg and lmc2 regions considered previously in the uv ( @xcite ; g03 ) .
each of the individual lmc curves and the average curves have been fit to the @xcite parameterization ; the results are presented in table [ fmtab ] .
the two sight lines in the smc , one representing each of the bar and wing regions , are considered individually .
a quick examination of the individual magellanic cloud extinction curves presented in figure [ allcurves ] reveals that the _ fuse _ portion of each curve is generally consistent with an extension of the data from longer wavelengths .
when the calibrated and h@xmath0-adjusted fluxes were initially compared , however , slight offsets were apparent between the portions of the curve on either side of the ly@xmath12 line . comparing the size of the gaps in the magellanic cloud data with the galactic curves studied in paper i led to the conclusion that the quality of the data for our reddened stars might have compromised the accuracy of the extinction curve construction .
in particular , the s / n of the _ fuse _ data seemed to be inversely proportional to the size of the offset and there were noticeable flux level differences for some spectra between the _ iue _ and the merged _ fuse _ data in the spectral overlap region .
the galactic sight lines had much better s / n values , negligible offsets , and good agreement between _ iue _ and _ fuse _ fluxes , whereas the magellanic cloud data suffered generally poor s / n , noticeable offsets , and sometimes poor flux agreement in the overlap . in order to reduce the offsets , the _ fuse _
data for the cloud sight lines were linearly adjusted ( a wavelength - independent multiplicative factor was applied to the flux ) over the entire fuv bandpass so that their fluxes matched _ iue _ in the instruments overlap region .
an example of one of the more egregious cases is plotted in figure [ fluxmatch ] , including a comparison of the _ fuse _ flux levels both before and after adjustment to the _ iue _ flux level .
the magnitude of each shift was estimated in increments of @xmath20 ; the systematic uncertainty associated with this adjustment has been propagated through all subsequent error calculations .
typically , the shifts improved the appearance of the curves , although the offsets were not eliminated in all cases .
given the general smoothness of the transition between uv and fuv portions of the extinction curve in most cases , it might seem reasonable to introduce further shifts to align these segments ; nevertheless , because the observed offsets fall generally within the error bounds , because we are interested in the detailed shape of the curve at uv and fuv wavelengths , and since we have no independent objective basis for such shifts , no further processing of the individual curves has been performed .
it should also be noted that any residual offsets appear to be biased in the direction of decreased @xmath21 ; thus , abundance requirements derived from the fuv data may be slight underestimates .
as already mentioned , a notable difficulty in deriving extinction curves that extend into the fuv is the existence of strong molecular hydrogen absorption bands longward of the ly@xmath12 line
. nevertheless , stellar observations of sufficient quality to allow the h@xmath0 absorption to be modelled and removed not only permit the construction of an extinction curve but also provide more direct information on the intervening ism .
recently , @xcite completed a survey of molecular hydrogen in the lmc and smc using several _ fuse _ observations of each cloud ; our sample introduces 15 new sight lines into the mix . in considering these new molecular hydrogen measurements ,
however , it should be recognized that the current _ fuse _ data generally have much lower signal - to - noise ratios than galactic observations .
in fact , they are also somewhat lower than the values for previously - observed magellanic cloud targets .
@xcite reported typical 4@xmath22 equivalent width limits of 3040 m , a level similar to values of 2050 m for our comparison stars but much lower than the 50120 m characteristic of the reddened targets .
because of these data limitations , and because the goal of generating a reliable fuv continuum was given a somewhat higher priority than making a rigorous assessment of molecular hydrogen , any column densities smaller than about 10@xmath23 @xmath24 , and a few that are somewhat larger , are subject to sizable uncertainties .
nevertheless , we consider the models listed in tables [ mwmolhydtab ] and [ mcmolhydtab ] to be reasonable tallies of the h@xmath0 along each sight line . in order to more
fully understand how the new sight lines mesh with the @xcite magellanic cloud h@xmath0 analysis , we have adopted the atomic hydrogen measurements previously published by @xcite for the individual sight lines ( see table [ hydtab ] ) ; for the pair listings in the table , we use the values derived directly from the extinction curves by g03 .
it should be noted here that the process of determining atomic hydrogen column densities for the magellanic clouds is complicated by the large quantities detected along each sight line .
the breadth of the ly@xmath12 absorption profile produced by typical interstellar hydrogen distributions with line - of - sight lengths exceeding even a few kiloparsecs is sufficient to mask distinctions between separate velocity components ; in particular , the profiles for atomic hydrogen gas in the lmc or smc are often inextricably blended with their galactic counterparts . @xcite
dealt with this problem by calibrating 21 cm emission measurements to ly@xmath12 from the sight lines in their sample for which the galactic and magellanic cloud 1215 profiles were not severely blended .
atomic hydrogen column densities for the remaining paths in the data set were then estimated using this calibration and the observed 21 cm emission .
g03 assessed magellanic cloud column densities by measuring the ly@xmath12 profile in the reddened - to - comparison ratio spectrum for each extinction pair , under the assumption that the milky way components would cancel each other .
each of these methods is subject to a significant level of uncertainty ; if vertical error bars were to be plotted in figure [ f(h2)comp ] , among the magellanic cloud sight lines they might typically be about 0.3 dex and occasionally larger than 1.0 dex .
nevertheless , differences between the @xcite values for the individual sight lines in each extinction pair generally are well matched by the corresponding column densities derived by g03 .
considered individually , our lmc and smc sight lines and those studied by @xcite exhibit similar ranges in characteristics such as the molecular hydrogen fraction @xmath3(h@xmath0 ) ( @xmath25@xmath26 $ ] ) and the magellanic cloud portion of the color excess @xmath27 ; = @xmath28 , where @xmath29 is 0.075 and 0.037 for lmc and smc paths , respectively . ]
values for the two lmc sight lines studied by @xcite also match these samples .
since the current dataset was selected for its extinction properties , we can not speak to the overall frequency of molecular hydrogen detection in the clouds .
however , as shown in figure [ f(h2)comp ] , the @xmath3(h@xmath0)-values for paths along which we detect a cloud - based h@xmath0 component agree with the levels set by previously observed lmc and galactic sight lines of large @xmath1 .
notably , the sight lines toward azv 456 and sk @xmath18672 possess larger molecular hydrogen fractions than any examined by @xcite ; meanwhile , among the paths directed toward our comparison stars we detect no magellanic h@xmath0 components .
based on the previously - studied sight lines toward sk -6619 and sk -69270 , it was found that while the gas - to - dust ratio @xmath2()/@xmath27 is much larger in the clouds than typically seen in the milky way , the quantity @xmath2(h@xmath0)/@xmath27 has a similar value in each galaxy ( @xcite ; gunderson et al .
this suggests a close relationship between dust and @xmath30 .
similar behavior is evident for our new sight lines , since although the ratios @xmath2()/@xmath27 for these paths are roughly 410 times larger than the galactic mean , comparing h@xmath0 column densities with the same color excesses leads to a range of values around the galactic average . in this way
our new sight lines complement the paths studied by gunderson et al .
( 1998 ) and the more heavily reddened paths of @xcite , implying generally lower values for @xmath3(h@xmath0 ) in the clouds ; nevertheless , the current dataset includes sight lines with unique properties .
plotting the lmc @xmath2()/@xmath27 ratio as a function of @xmath31 ( see figure [ gtdebvplot ] ) , there is a clear distinction between sk -672 and other reddened sight lines : this one lmc path is characterized by an `` atomic '' ratio appropriate to galactic curves , but because of its large molecular hydrogen column density the total gas - to - dust ratio approaches the values for other lmc sight lines . coupled with its milky way - like @xmath2 ( ) ratio
, the agreement exhibited between the gas - to - dust ratios for sk -672 , the sight line with the largest @xmath3(h@xmath0)-value in our sample , and sight lines dominated by atomic gas reinforces the concept that the mechanisms governing dust formation in the galaxy and lmc are similar and that any differences in extinction arise out of how interstellar environmental conditions are manifested in dust grain population characteristics .
the smc sight line azv 456 also possesses a `` galactic '' value for @xmath2()/@xmath27 that gives a @xmath32 ratio more closely resembling the values determined by @xcite for other smc paths .
further details of the interstellar conditions , specifically the temperatures @xmath33 and @xmath34 , in the clouds can be derived from the relative rotational level populations of the individual sight lines ( table [ mcmolhydtab ] ) .
when the cloud density is sufficiently high and h@xmath0 column densities are large enough for the gas to become substantially molecular , ortho - para conversion will bring @xmath33 ( @xmath35 ) close to the kinetic temperature . except for the azv 462 sight line ,
its value is fairly uniform among the lmc and smc components of our h@xmath0 models and we derive a weighted mean temperature of 50@xmath363 k. the magellanic cloud component of h@xmath0 toward azv 462 has the lowest column density among our extragalactic values , and the assumption that this value is consistent with the kinetic temperature breaks down .
the mean temperature we have derived for the magellanic cloud components neglects azv 462 and is somewhat lower than the overall galactic mean of 77@xmath3617 k @xcite , but is consistent with both the value 55@xmath368 k derived from the galactic subsample for which @xmath37 @xcite and temperatures derived by @xcite for magellanic cloud gas along sight lines with some of the largest molecular column densities .
it should be noted that the errors ascribed to each temperature listed in tables [ mwmolhydtab ] and [ mcmolhydtab ] were derived from column density uncertainties , which were often large .
thus , kinetic temperatures with uncertainties larger than about 60k should be regarded as less reliable . in particular , the formal uncertainty is listed in each case , even where it exceeds the calculated temperature , in order to indicate the relative reliability of each determination .
we have also determined @xmath34 for each sight line from the slope in a plot of @xmath38 versus the excitation potential .
is associated with values of t@xmath39 - \log[n(j)/g_{j } ] } = \frac{170 { \rm k } \times ( 2j - 1)}{\ln[\frac{2j+1}{2j-3 } \frac{n(j-2)}{n(j)}]}\;({\rm for}\ , j \ge 2)$ ] , the temperature derived from comparing the populations of successive even or odd @xmath15 levels . ]
this excitation temperature reflects the influence of fluorescent uv pumping and h@xmath0 formation in excited states on the molecular hydrogen rotational level population @xcite . like @xmath33 ,
it is fairly uniform among the lmc and smc sight lines and we calculate a weighted mean of 244@xmath3633 k , similar to previously - determined values for magellanic cloud sight lines @xcite . comparing the column density ratios @xmath2(4)/@xmath2(2 ) and @xmath2(5)/@xmath2(3 ) for our sample with figure 11 from @xcite , we find that these sight lines are reasonably typical of their high @xmath2(h@xmath0 ) paths ; consequently , our measurements support their conclusions that the h@xmath0 formation rate in the clouds is low ( 1040% of galactic rates ) and that at least some regions are illuminated by relatively intense radiation fields ( 10100 times the galactic mean ) .
we note , however , that the proportion of outliers in our sample that do not fit the range of galactic models calculated by @xcite is much smaller than theirs .
values of @xmath33 and @xmath34 were also derived for the galactic components of our h@xmath0 models .
a mean of @xmath33 = 67@xmath364 k was derived for the sight lines with larger column densities and more robust measurements , although higher , and likely to be in part non - thermal , temperatures are implied for several of the more diffuse paths .
excitation temperatures for the galactic components ( 239@xmath3638 k ) are similar to those we have determined for the magellanic clouds . before _ fuse _
, only two reddened sight lines outside the galaxy had been studied in the fuv @xcite .
both regional extinction groups in the lmc are represented by these two stars ; specifically , sk -6619 is associated with lmcavg and sk -69270 with lmc2 ( misselt et al .
the uv extinction curves for these stars are typically non - ccm , but they share the trait of most galactic curves in that their uv extinction slope blends smoothly into the fuv .
these two sight lines have been included for comparison in table [ fmtab ] , which defines the lmc2 and lmcavg subsamples , although they were left out of the average extinction curves for those regions .
@xcite published fuv extinction curves for 3 sight lines in each of the lmc and the smc using _ fuse _ data .
unfortunately , the values of @xmath40(b - v ) for the reddened / unreddened star pairs chosen for that study are too small ( @xmath410.02 - 0.07 mag ) for detailed extinction studies . more specifically , these reddening values are comparable in size to the variation in the foreground @xmath1 and similar to the magnitude of the photometric uncertainties @xcite .
so , within the uncertainties there is no significant reddening difference for any of the star pairs in @xcite , and those sight lines are not included in this study . the new lmc and smc fuv extinction curves presented here seem to follow the trend of sk -69270 and sk -6619 .
despite any small remaining offsets between the _ iue _ and _ fuse _ extinction curves ( see [ ssection_h2modelling ] ) , the new fuv curves closely follow an extrapolation of their fm fits from the uv . out of our sample , only two curves exhibit significant deviations from this overall trend , azv 456 and sk -69228 .
azv 456 is our lone reddened star situated in the smc wing , and the _ fuse _ portion of its extinction curve more closely resembles its fuv ccm curve than an extrapolation of its fm fit in the uv .
coincidentally , the disagreement with the fm fit begins near the onset of molecular hydrogen absorption features ; however , attempting to reconcile the two portions of the _ fuse _ curve by adjusting the h@xmath0 column densities indicates that a single component for the clouds can not reproduce the observed continuum level shift , nor does it appear that a simple combination of overlapping components would accomplish this result .
the effect does not appear to be related to a difference in the sensitivity of the various _ fuse _ channels either , since the continuum across this spectral region is contained in both the lif1b and lif2a detector bands .
similar kinks are apparent to a much smaller degree in the galactic curves for hd 62542 , hd 73882 , and hd 210121 ( paper i ) ; nevertheless , those curves match the fm fits within the formal uncertainties [ approximately 14% in a(@xmath42)/a(v ) across the _ fuse _ band ] and we feel that it would not be appropriate to speculate further about the origin of these features until a larger data sample is compiled .
the fuv extinction curve for sk -69228 is very noisy , which would usually indicate a poor match with its unreddened comparison star , but this same star is a good match through the uv .
the problem seems to arise with a complex of photospheric lines evident in the comparison star spectrum , sk -6515 .
an examination of intermediate continuum segments demonstrates that the reddened and comparison stellar spectra have very similar shapes aside from these lines , and that connecting the lower wavenumber portion of this star s fuv extinction curve with the two highest wavenumber points would be a reasonable approximation .
notably , this procedure results in an fuv curve that is consistent with an extension of the uv fm fit , despite the impression that a small offset is still present after the _ fuse _ flux adjustment .
the curve is still included in further analysis for the sake of completeness and because our comparison sample is limited , but we note that the mean curve derived for the lmc2 group , which includes this path , is consistent with the other curves in the group unaffected by any taint of mismatch . on the whole , the ease with which the uv and fuv extinction curves line up for our magellanic cloud sight lines complements the results for milky way sight lines ; namely , that most galactic sight lines also seem to show fuv extinction that is a good extrapolation of the fm fits made to the extinction curves in the uv . agreement between the ccm relation and fuv extinction ,
however , has proved less ubiquitous .
for instance , @xcite found that the fuv extinction for two galactic sight lines with large ( @xmath43 oph ) and small ( hd 25443 ) values of @xmath10 follow ccm closely , but that there are exceptions ; one notable example is the bright - nebula sight line toward hd 37903 which shows a steeper extinction in the fuv , based on _ copernicus _ data , than would be predicted by ccm .
paper i , which included paths such as hd 210121 and hd 62542 that do not follow ccm in the uv , came to similar conclusions . although the uv extinction curves for each sight line they studied could be smoothly extrapolated into the fuv using an fm fit , only three paths that followed ccm through uv wavelengths were also in accord across the fuv .
the ccm relations for four others from paper i either over or underestimated their fuv extinction , as small discrepancies in the uv became more pronounced at shorter wavelengths , and the two non - ccm curves diverged more strongly from their @xmath10-based curves with increasing wavenumber . because the lmc and smc curves do not have a ccm - like wavelength dependence and because the s / n ratios of the individual extinction curves are relatively low , subtle variations in the fuv slope such as those seen by @xcite can not be ruled out . to increase the s / n for mem modelling purposes , we have constructed average extinction curves for the lmc average and lmc2 regions outlined by misselt et al .
1999 ( see table [ fmtab ] ) ; the curves are plotted in figure [ extcurves ] .
the two sight lines in the smc , which represent the bar and wing regions , are considered individually . using the two average lmc curves and the two smc curves , we have extended the dust - grain population analysis of @xcite into the fuv .
we employ a ( slightly ) modified version of the mem extinction - fitting algorithm @xcite . instead of using the number of grains of a given size to describe the dust population , the algorithm employs the mass distribution in which @xmath44 is the mass of dust grains per h atom in the size interval from @xmath45 to @xmath46 .
thus , the traditional mrn - type model @xcite becomes @xmath47 .
we use a power law with exponential decay ( ped ) as the template function for each component .
the data are examined at 34 wavelengths , and the grain cross sections are computed over the range 0.00252.7 with 50 logarithmically spaced bins .
the shape of the mass distribution is strongly constrained only for data over the interval 0.021 .
below 0.02 , the rayleigh scattering behavior constrains only total mass ; above 1 , the `` grey '' nature of the dust opacity also forces the mem algorithm to simply adjust the total mass , using the shape of the template function to specify the size dependence of the distribution .
the total mass of dust is constrained using both the gas - to - dust ratio and `` cosmic '' abundances ( i.e. , we try not to use more carbon or silicon than is available ) .
the elemental abundance standards used in this analysis are those adopted in paper i for the galaxy ( 358 and 35 atoms per million h for c and si , respectively ) , and by @xcite for the lmc and smc ( 110 and 65 for the lmc and 54 and 11 atoms per million h for the smc ) ; table [ hydtab ] reports the values of the gas - to - dust ratios used here for the lmc and smc .
we consider only three - component models of homogeneous , spherical grains : modified `` astronomical silicate '' @xcite , amorphous carbon @xcite , and graphite @xcite .
while it must be acknowledged that the three grain component system we have used to model these extinction curves is simpler than might be expected of actual interstellar dust , the results can be quite useful for identifying grain population properties that distinguish sight lines from one another .
the same component - specific ped constraints ( e.g. , the onset of the exponential cutoff ) adopted in paper i were utilized for the current modelling .
the mem fits to fm parameterizations of the average magellanic cloud extinction curves listed in table [ fmtab ] are presented in figures [ memfits ] and [ memmass ] .
the first plot shows the amount of extinction provided by the three distinct grain components , as well as the total extinction of the model , compared to the fm fit associated with each extinction curve .
the error bars plotted on these fits are indicative of the mean gap between the fm fit and the underlying extinction curve in each wavelength bin .
the fractions of the adopted elemental abundances available for each grain component that are required by the best mem fit are listed in table [ memtab ] .
figure [ memmass ] depicts the corresponding mass distributions for different sizes of grains belonging to each model component relative to the mass of hydrogen .
the proportions of the available silicon and carbon used in the mem fits to the various fm parameterizations cover a very wide range , as was seen in fits to ir through uv data alone by @xcite .
three general factors determine the fraction of silicon and carbon that any individual sight line will use .
first , the higher the gas - to - dust ratio is , the more metals are available in the gas phase .
second , the higher the abundances of metals are , the more material is available .
finally , high values of @xmath10 imply a greater than average mass fraction in larger grains which are not as efficient per unit mass as smaller grains . for instance
, it can be seen from the mem fits that the smc wing ( azv 456 ) , which has a low gas - to - dust ratio and low elemental abundances relative to the galaxy , uses more than 100% of the available silicon .
however , the lmc2 and the smc bar ( azv 18 ) regions , which also have both low elemental abundances but are characterized by higher gas - to - dust ratios , use less than half the amount of available silicon that the lmcavg and smc wing regions require and also significantly less carbon . a recent observation has suggested that silicon is relatively undepleted in the smc bar @xcite .
however , model fits to the average smc bar extinction indicate that its curve can not be fitted with carbon grains alone @xcite .
figures [ memfits ] and [ memmass ] clearly show that this conclusion is born out in the current fits .
smc extinction for both bar and wing is very steep in the fuv , necessitating the presence of a large population of small grains . yet because the 2175 bump is absent in the bar region , small ( @xmath48 ) carbon grains do not meet the smc fuv extinction requirements ; thus , silicates play a very important role in the models .
even our smc wing sight line , whose extinction curve follows ccm reasonably well , places a much higher demand on silicon reserves than carbon : a typical galactic curve conforming to ccm allows graphite to fulfill a significant part in fuv extinction due to the strength of its uv bump , but the smc wing bump is somewhat weaker than the @xmath10=2.19 ccm curve would imply .
the next steepest sight lines , in the lmc , do not show much difference in demand from a typical galactic curve ( hd 14250 was chosen from paper i for illustration purposes ) , once the gas - to - dust ratio differences are taken into account .
typically , carbon grains are responsible for most of the visible extinction and silicon grains for most of the uv extinction ; therefore , in general , both species of grains are needed along any sight line to get a good fit to the extinction curve .
apart from deficits of large silicate and carbon grains in the smc , the mem analysis of the magellanic cloud fuv extinction produces results similar to those derived from an analysis of several galactic sight lines ( paper i ) .
the galactic sight lines , like the smc wing or lmcavg , are more likely to use more than 100% of the available silicon and/or carbon . among these sight lines ,
the strength of the 2175 bump in relation to the fuv rise is the factor that distinguishes whether the stiffer abundance demand is placed on silicon or carbon .
likewise , the element most demanded by the smc bar sight line and the mean lmc2 curve is determined by this factor , although the minimal presence of the bump in these curves allows the modelling to make more efficient use ( especially in small grains ) of the available elements . the general structure in the size vs @xmath49 curves for extinction out to 8 @xmath9 for the magellanic cloud sight lines does not change much when the _ fuse _ data are included .
figures [ memfits ] and [ memmass ] showing the mem fits and size vs @xmath49 curves can not be directly compared to figure 3 of @xcite because there are differences in the input parameters such as the gas - to - dust ratios .
but the mem models run with the same input parameters for extinction curve that are cutoff at 8 @xmath9 and those shown here , which extend to 10 @xmath9 , are quite similar ; there is at most a few percent difference in the amount of silicon and carbon required by the two sets of models for the four smc and lmc curves .
the structure evident in the mass distributions of figure [ memmass ] particularly identify grain sizes for which each component in our simple model can account for the detailed shape of each extinction curve .
for instance , the two peaks in the amorphous carbon distribution for hd 14250 correspond roughly to maxima appearing in the silicate distribution of @xcite in their silicate - graphite model .
these authors ascribe the peaks to the requirements of fitting the optical and uv portions of the extinction curve using the spectral wavelength dependence appropriate to each of the two grain components in their model . in the case of azv 456 ( the smc wing ) , however , efforts to reproduce the extinction curve using this three - component model appear to require dramatic peaks and dips in the mass distribution as a function of grain size . aside from this curve , however , the scale of structure in the distribution is consistent with levels noted by previous studies ( kim et al .
1994 ; @xcite ) . for azv 456
, the solution requires that the graphite distribution be strongly peaked in order to account for the 2175 bump and provide as much extinction as possible at fuv wavelengths .
since the fuv portion of this curve extends , or even amplifies , the steep rise at the short - wavelength end of the previous ir through uv result @xcite , there is no relaxation of the demand on silicon reserves . specifically , when comparing mem solutions for the smc wing curve with and without the _ fuse _ data , nearly identical proportions of the available silicon and carbon are utilized .
the only notable distinction between the two solutions is that a few percent of the amorphous carbon demand is shifted to graphite when the fuv data are included .
the extinction curves for our other groups also smoothly extend from the uv into fuv wavelengths , and the models derived from ir - to - uv and ir - to - fuv data require almost precisely the same amounts of silicon and carbon be present in dust grains .
the similarity in the mem results for these two sets of models may be due to the efficiency of the small grains in fuv extinction , in that large numbers of such grains are not needed , or perhaps the observed amounts of fuv extinction can be provided by the larger grains which are also important for extinction at longer wavelengths .
this issue will be investigated further as we study the distinction between global and sight - line - specific fuv extinction characteristics of using a much larger number of sight lines , in a future paper .
this paper is the second in a series investigating the fuv characteristics of extinction due to interstellar dust .
the first paper dealt with a small sample of galactic sight lines with a variety of @xmath10 values , including paths whose extinction does not conform to the ccm parameterization .
this paper examined a similarly - sized set of sight lines probing the small and large magellanic clouds , with the goal of determining how the addition of fuv data to their corresponding extinction curves affected models of magellanic cloud dust grain properties . using _ fuse _ observations of four stars in the smc and 12 stars in the lmc
, we were able to construct two extinction curves for the smc and seven representing the lmc . despite the poor quality of the data for some of the reddened stars ,
molecular hydrogen was measured for each sight line and their corresponding absorption features were removed from consideration in the extinction analysis .
flux mismatches between _ iue _ and _ fuse _ data , likely due to poor s / n values in the affected observations , were alleviated by rescaling the entire _ fuse _ spectrum .
comparison of the resulting fuv extinction with the curve constructed only through wavelengths observed by _
iue _ demonstrated that the fuv portions were generally consistent with a smooth extrapolation of the ir - to - uv curve .
nevertheless , we note that a `` kink '' appeared in the azv 456 curve , similar to much smaller features present in curves analysed in paper i. we plan to address the existence and nature of this kink in a subsequent paper examining the variation in properties of fuv extinction .
it was anticipated that including fuv data in the mem analysis of the magellanic cloud dust grain population might give rise to significantly different properties than galactic dust since the fuv rise in the clouds is generally much steeper than it is in the milky way , and the uv bump is smaller .
however , aside from some complex structure likely driven by the simplicity of our grain component model , the sight lines through the clouds exhibitting galactic - like extinction are similar to galactic paths in that they demand 100% or more of the available carbon and/or silicon .
the only real differences between the populations for these galaxies are that the smc wing sight line has a much smaller grain size cutoff for astronomical silicates and amorphous carbon and a generally smaller dust mass .
mem solutions for the sight lines through the clouds that exhibited more distinctly magellanic - cloud - type extinction were able to satisfy their curves using smaller proportions of the elements available to them ; the smc bar solution , like that for the smc wing , is characterized by a smaller grain size cutoff and lower dust mass than galactic or lmc solutions .
the addition of the _ fuse _ data to the analysis does not dramatically alter the properties of each dust population , except to shift small amounts of the carbon demand between amorphous carbon and graphite .
the molecular hydrogen abundances determined in the process of constructing the full ir - to - fuv extinction curves supplement those already appearing in the literature .
although the emphasis has been on eliminating h@xmath0 absorption features from the spectra rather than deriving robust column density measurements , our results are consistent with the larger recent survey of molecular hydrogen in the magellanic clouds by @xcite and we complement their paths by probing several sight lines with larger column densities . among these sight lines , @xmath1/@xmath2 ( ) ratios
are generally reduced relative to galactic values while @xmath1/@xmath2(h@xmath0 ) are roughly the same , implying somewhat lower @xmath3(h@xmath0)-values in the clouds than are typical in the milky way for similar degrees of reddening .
black , j. h. , & dalgarno , a. 1976 , , 203 , 132 bouchet , p. , lequeux , j. , maurice , e. , prvot , l. , & prvot - burnichon , m. l. 1985 , , 149 , 330 browning , m. k. , tumlinson , j. , & shull , j. m. 2003 , , 582 , 810 buss , r. h. , jr . , allen , m. , mccandliss , s. , kruk , j. , liu , j. , & brown , t. 1994 , , 430 , 630 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 ( ccm ) clayton , g. c. , green , j. , wolff , m. j. , zellner , n. e. b. , code , a. d. , davidsen , a. f. , wuppe science team , & hut science team 1996 , , 460 , 313 clayton , g. c. , wolff , m. j. , sofia , u. j. , gordon , k. d. , & misselt , k. a. 2003 , , 588 , 871 diplas , a. , & savage , b. d. 1994 , , 427 , 274 draine , b. t. 2003 , , 41 , 241 fitzpatrick , e. l. 1985a , , 299 , 219 fitzpatrick , e. l. 1985b , , 59 , 77 fitzpatrick , e. l. 1986 , , 92 , 1068 fitzpatrick , e. l. , & massa , d. m. 1990 , , 72 , 163 gordon , k. d. , & clayton , g. c. 1998 , , 500 , 816 gordon , k. d. , clayton , g. c. , misselt , k. a. , landolt , a. u. , & wolff , m. j. 2003 , , 594 , 279 ( g03 ) green , j. c. , snow , t. p. , jr . , cook , t. a. , cash , w. c. , & poplawski , o. 1992 , , 395 , 289 gunderson , k.s . ,
clayton , g.c . , & green , j.c .
1998 , , 110 , 60 hutchings , j. b. , & giasson , j. 2001 , , 113 , 1205 jenkins , e. b. , savage , b. d. , & spitzer , l. 1986 , , 301 , 355 jura , m. 1974 , , 191 , 375 jura , m. 1975 , , 197 , 581 kim , s .- h . , martin , p. g. , & hendry , p. d. 1994 , , 422 , 164 kim , s .- h .
, & martin , p. g. 1996 , , 462 , 296 koornneef , j. , & code , a. d. 1981 , , 247 , 860 koornneef , j. 1982 , , 107 , 247 laor , a. , & draine , b. t. 1993 , , 402 , 441 lewis , n. k. , cook , t. a. , & chakrabarti , s. 2005 , , 619 , 367 longo , r. , stalio , r. , polidan , r. s. , & rossi , l. 1989 , , 566 , 267 massa , d. , & fitzpatrick , e. l. 2000 , , 126 , 517 massa , d. , savage , b.d . , & fitzpatrick , e. l. 1983 , , 266 , 662 massey , p. 2002
, , 141 , 81 misselt , k. a. , clayton , g. c. , & gordon , k. d. 1999 , , 515 , 128 mathis , j. s. , rumpl , w. , & nordsieck , k. h. 1977 , , 217 , 425 nandy , k. , morgan , d. h. , willis , a. j. , wilson , r. , gondhalekar , p. m. , & houziaux , l. 1980 , , 283 , 725 oestreicher , m. o. , gochermann , j. , & schmidt - kaler , t. 1995 , , 112 , 495 rachford , b. l. , snow , t. p. , tumlinson , j. , shull , j. m. , roueff , e. , andr , m. , dsert , j. -m . , ferlet , r. , vidal - madjar , a. , & york , d. g. 2001 , , 555 , 839 rachford , b. l. , snow , t. p. , tumlinson , j. , shull , j. m. , blair , w. p. , ferlet , r. , friedman , s. d. , gry , c. , jenkins , e. b. , morton , d. c. , savage , b. d. , sonnentrucker , p. , vidal - madjar , a. , welty , d. e. , & york , d. g. 2002 , , 577 , 221 rocca - volmerange , b. , prvot , l. , prvot - burnichon , m. l. , ferlet , r. , lequeux , j. 1981 , , 99 , l5 sahnow , d. j. , moos , h. w. , ake , t. b. , andersen , j. , andersson , b .-
g . , andr , m. , artis , d. , berman , a. f. , blair , w. p. , brownsberger , k. r. , calvani , h. m. , chayer , p. , conard , s. j. , feldman , p. d. , friedman , s. d. , fullerton , a. w. , gaines , g. a. , gawne , w. c. , green , j. c. , gummin , m. a. , jennings , t. b. , joyce , j. b. , kaiser , m. e. , kruk , j. w. , lindler , d. j. , massa , d. , murphy , e. m. , oegerle , w. r. , ohl , r. g. , roberts , b. a. , romelfanger , m. l. , roth , k. c. , sankrit , r. , sembach , k. r. , shelton , r. l. , siegmund , o. h. w. , silva , c. j. , sonneborn , g. , vaclavik , s. r. , weaver , h. a. , wilkinson , e. 2000a , , 538 , l7 sahnow , d. j. , gummin , m. a. , gaines , g. a. , fullerton , a. w. , kaiser , m. e. , & siegmund , o. h. 2000b , , 4139 , 149 sasseen , t. p. , hurwitz , m. , dixon , w. , v. , & airieau , s. 2002 , , 566 , 267 savage , b. d. , bohlin , r. c. , drake , j. f. , & budich , w. 1977 , , 216 , 291 savage , b. d. , & mathis , j. s. 1979 , , 17 , 73 schwering , p. b. w. , & israel , f. p. 1991
, , 246 , 231 shull , j. m. , tumlinson , j. , jenkins , e. b. , moos , h. w. , rachford , b. l. , savage , b. d. , sembach , k. r. , snow , t. p. , sonneborn , g. , york , d. g. , blair , w. p. , green , j. c. , friedman , s. d. , & sahnow , d. j. 2000 , , 538 , l73 snow , t. p. , jr . ,
allen , m. m. , polidan , r. s. 1990 , , 359 , l23 snow , t. p. , jr . , & york , d. g. 1975 , , 34 , 19 sofia , u. j. , wolff , m. j. , rachford , b. l. , gordon , k. d. , clayton , g. c. , cartledge , s. i. b. , martin , p. g. , draine , b. t. , mathis , j. s. , snow , t. p. , & whittet , d. c. b. 2005 , , _ accepted _ ( paper i ) tumlinson , j. , shull , j. m. , rachford , b. l. , browning , m. k. , snow , t. p. , fullerton , a. w. , jenkins , e. b. , savage , b. d. , crowther , p. a. , moos , h. w. , sembach , k. r. , sonneborn , g. , & york , d. g. 2002 , , 566 , 857 weingartner , j. c. , & draine , b. t. 2001 , , 548 , 296 welty , d. e. , hobbs , l. m. , & york , d. g. 1991 , , 75 , 425 welty , d. e. , lauroesch , j. t. , blades , c. , hobbs , l. m. , & york , d. g. 2001 , , 554 , l75 york , d. g. , drake , j. f. , jenkins , e. b. , morton , d. c. , rogerson , j. b. , & spitzer , l. 1973 , , 182 , l1 zubko , v. g. , menella , v. , colangeli , l. , & bussoletti , e. 1996 , , 282 , 1321 llcr@@xmath36lr@@xmath36lr@@xmath36lr@@xmath36l azv 18 & azv 462 & a & 0.17 & 0.03 & 2.90 & 0.42 & 0.49 & 0.11 & 0.75 cm 17.27 & 4.30 + azv 456 & azv 70 & b & 0.26 & 0.03 & 2.19 & 0.23 & 0.57 & 0.08 & 7.01 & 1.22 + sk -672 & sk -6635 & a & 0.15 & 0.05 & 3.75 & 0.36 & 0.56 & 0.23 & 1.78 & 0.67 + sk -6826 & sk -6635 & a & 0.19 & 0.03 & 3.45 & 0.27 & 0.64 & 0.16 & 5.46 & 1.08 + sk -68129 & sk -6841 & a & 0.17 & 0.05 & 3.37 & 0.29 & 0.57 & 0.22 & 6.98 & 2.36 + sk -68140 & sk -6841 & a & 0.20 & 0.05 & 3.34 & 0.25 & 0.67 & 0.22 & 5.99 & 1.75 + sk -68155 & sk -67168 & a & 0.20 & 0.05 & 2.81 & 0.22 & 0.56 & 0.18 & 8.88 & 9.21 + sk -69228 & sk -6515 & a & 0.15 & 0.05 & 3.54 & 0.34 & 0.53 & 0.23 & 6.60 & 2.50 + sk -69279 & sk -6563 & a & 0.21 & 0.05 & 3.54 & 0.25 & 0.75 & 0.23 & 5.38 & 1.50 + lccc azv 18 & a1180101000 & 2000 may 29 & 9293 + & b0890101000 & 2001 jun 13 & 45245 + azv 70 & a1180202000 & 2000 oct 03 & 2831 + & a1180203000 & 2000 oct 05 & 1971 + & b0900601000 & 2001 jun 15 & 6152 + azv 456 & q1070101000 & 2000 oct 04 & 3691 + & q1070104000 & 2000 oct 06 & 4903 + & q1070106000 & 2000 oct 09 & 5641 + & q1070102000 & 2000 oct 10 & 2687 + & q1070103000 & 2000 oct 12 & 4352 + & p2210201000 & 2001 jun 14 & 8278 + azv 462 & a1180301000 & 2000 jul 03 & 5550 + sk -6515 & b0861001000 & 2001 nov 17 & 4293 + sk -6563 & a0490701000 & 1999 dec 16 & 5927 + & m1142001000 & 2000 sep 26 & 5581 + & b0861101000 & 2001 sep 22 & 4125 + sk -6635 & b1280101000 & 2001 oct 25 & 4030 + & b0860301000 & 2002 sep 22 & 4480 + sk -672 & b0860101000 & 2001 aug 14 & 6935 + sk -67168 & b0860901000 & 2001 sep 22 & 4102 + sk -6826 & b0860201000 & 2001 sep 17 & 11391 + sk -68129 & b0860501000 & 2001 sep 22 & 6681 + sk -68140 & b0860601000 & 2001 sep 17 & 10712 + sk -68155 & b0860701000 & 2001 sep 22 & 8020 + sk -69228 & b0860401000 & 2001 sep 23 & 9422 + sk -69279 & b0860801000 & 2001 sep 22 & 5991 + lccccccccr@@xmath36lr@@xmath36l + azv 18 & 16.57(0.64)&16.33&16.17&14.82&14.27 & & & 8.1@xmath50 & 66&192&187&296 + azv 70 & 18.55(0.05)&18.42&17.95&15.90&15.46&14.66 & & 4.2@xmath51 & 52 & 6&357&786 + azv 462 & 15.92(0.16)&15.55&15.48&14.88&14.98 & & & 5.9@xmath52 & 72 & 34&419&223 + + sk @xmath186515 & 18.15(0.06)&17.79&17.86&16.75&16.04 & & & 2.4@xmath53 & 84 & 6&165&330 + sk @xmath186563 & 18.00(0.07)&17.75&17.54&16.66&16.56 & & & 1.4@xmath54 & 64 & 8&303&263 + sk @xmath186635 & 17.78(0.46)&17.24&17.60&16.34&15.74&14.08 & & 4.8@xmath55&125&310&210&227 + sk @xmath18672 & 16.83(0.36)&16.56&16.19&15.98&15.81 & & & 6.3@xmath56 & 56 & 90&277&197 + sk @xmath1867168 & 16.38(0.35)&15.60&16.25&15.24&14.74 & & & 3.5@xmath57&200 & 95&200 & 95 + sk @xmath186826 & 16.02(0.35)&15.61&15.32&15.58&14.67 & & & 5.4@xmath58 & 60 & 62&354&519 + sk @xmath186841 & 15.31(0.27)&14.31&15.14&14.39&14.31 & & & 4.3@xmath59&326&230&326&230 +
sk @xmath1868129 & 16.73(0.59)&15.76&16.39&16.15&15.87&15.13 & & 2.9@xmath60&228&868&409&868 + sk @xmath1868140 & 19.01(0.12)&18.54&18.82&17.37&16.28&14.37&14.17 & 2.5@xmath61&110 & 42&227 & 50 + sk @xmath1868155 & 18.35(0.50)&18.26&17.64&15.77&15.36&14.39 & & 4.6@xmath62 & 47&133&330&183 + sk @xmath1869228 & 18.00(0.20)&17.29&17.80&17.22&16.14 & & & 2.5@xmath63&167&189&224&620 + sk @xmath1869279 & 16.17(0.16)&15.70&15.93&14.88&14.71 & & & 8.3@xmath64&102 & 59&277&308 + lccccccccr@@xmath36lr@@xmath36l + azv 18 & 20.36(0.07)&20.13&19.97&17.58&17.79&16.06&15.36 & 8.1@xmath50 & 66 & 9&270 & 52 + azv 456 & 20.93(0.09)&20.85&20.15&17.67&15.62 & & & 6.0@xmath65 & 45 & 4 & 82&175 + azv 462 & 17.65(0.13)&16.65&17.54&16.56&16.16 & & & 2.1@xmath66&&234&564 + + sk @xmath186635 & 19.13(0.19)&19.02&18.46&17.02&15.89&14.78&14.26 & 4.4@xmath67 & 49 & 23&302&152 + sk @xmath18672 & 20.95(0.08)&20.86&20.21&18.41&16.43&15.40 & & 5.6@xmath68 & 46 & 5&159 & 75 + sk @xmath186826 & 20.38(0.08)&20.20&19.90&18.50&17.76&16.43&15.88 & 2.3@xmath69 & 59 & 7&253&260 + sk @xmath1868129 & 20.05(0.10)&20.00&19.08&18.06&16.65&15.87&15.55 & 8.0@xmath70 & 40 & 13&309&157 + sk @xmath1868140 & 19.50(0.13)&19.09&19.25&18.13&17.45&15.97&15.45 & 5.4@xmath71 & 93 & 35&248 & 71 + sk @xmath1868155 & 19.99(0.24)&19.75&19.62&17.21&16.37&15.42&14.84 & 8.2@xmath72 & 68 & 42&341&152 + sk @xmath1869228 & 18.70(0.14)&18.28&18.37&17.79&17.13 & & & 3.1@xmath73 & 86 & 37&249&555 + sk @xmath1869279 & 20.31(0.07)&20.10&19.90&16.46&16.11&15.34 & & 10.0@xmath74 & 64 & 9&389&437 + lclc azv 18 & 1.10&sk @xmath1867168&1.05 + azv 70 & 1.05&sk @xmath186826 & 0.85 + azv 456 & 0.85&sk @xmath186841 & 0.95 + azv 462 & 1.00&sk @xmath1868129&1.00 + sk @xmath186515 & 1.20&sk @xmath1868140&1.10 + sk @xmath186563 & 0.85&sk @xmath1868155&1.20 + sk @xmath186635 & 1.05&sk @xmath1869228&1.25 + sk @xmath18672 & 0.65&sk @xmath1869279&1.30 + lr@@xmath36lr@@xmath36lr@@xmath36lr@@xmath36lr@@xmath36lr@@xmath36l + azv 18 & @xmath184.902&1.036&2.255&0.436&0.165&0.213 & 0.001&0.026 & 4.697&0.078&0.738&0.012 + + azv 456 & @xmath180.419&0.130&0.908&0.094&5.026&1.625 & 0.513&0.100 & 4.770&0.079&1.470&0.025 + + sk @xmath186619 & & & & & & + sk @xmath18672 & @xmath183.479&1.585&1.723&0.648&3.108&1.296 & 0.881&0.313 & 4.573&0.048&0.935&0.016 + sk @xmath186826 & 0.003&0.020&0.937&0.150&2.581&0.513 & 0.203&0.054 & 4.638&0.049&0.898&0.015 + sk @xmath1868129 & @xmath182.174&0.784&1.393&0.443&1.184&0.351 & 0.207&0.082 & 4.567&0.071&0.688&0.041 + average & @xmath181.704&0.253&1.268&0.041&2.902&0.187 & 0.266&0.030 & 4.602&0.023&0.930&0.015 + + sk @xmath1868140 & @xmath181.929&0.820&1.323&0.387&1.093&0.525 & 0.161&0.055 & 4.440&0.074&0.811&0.014 + sk @xmath1868155 & @xmath182.842&0.982&1.615&0.458&0.897&0.336&@xmath180.012&0.013 & 4.611&0.062&0.684&0.028 + sk @xmath1869228 & @xmath182.800&1.219&1.449&0.529&0.641&0.237&@xmath180.418&0.164 & 4.714&0.079&0.609&0.054 + sk @xmath1869270 & & & & & & + sk @xmath1869279 & @xmath182.696&0.857&1.364&0.364&0.853&0.232&@xmath180.124&0.057 & 4.603&0.077&0.664&0.047 + average & @xmath182.487&0.303&1.443&0.054&0.844&0.140&@xmath180.055&0.016 & 4.586&0.058&0.697&0.012 + lr@@xmath75lr@@xmath75lccr@@xmath75lr@@xmath75l + azv 18 & 9.0&@xmath76&2.3&@xmath77&@xmath181.35&0.174&5.2&@xmath78 & 1.3&@xmath76 + azv 456 & 1.5&@xmath76&8.5&@xmath77&@xmath180.28&0.332&4.5&@xmath76 & 2.6&@xmath76 + azv 462 & 6.0&@xmath77&4.5&@xmath79&@xmath182.83&0.007&8.6&@xmath78 & 6.4&@xmath80 + + azv 18/azv 462 & 8.5&@xmath76&2.3&@xmath77&@xmath181.29&0.17&5.0&@xmath78 & 1.4&@xmath76 + azv 456/azv 070 & 4.0&@xmath76&8.5&@xmath77&@xmath180.53&0.26&1.5&@xmath78 & 3.3&@xmath76 + + sk @xmath186635 & 4.0&@xmath77&1.3&@xmath80&@xmath181.20&0.055&7.3&@xmath76 & 2.4&@xmath77 + sk @xmath18672 & 1.0&@xmath76&8.9&@xmath77&@xmath180.19&0.190&5.3&@xmath76 & 4.7&@xmath76 + sk @xmath186826 & 3.5&@xmath76&2.4&@xmath77&@xmath180.92&0.181&1.9&@xmath78 & 1.3&@xmath76 + sk @xmath1868129 & 5.2&@xmath76&1.1&@xmath77&@xmath181.38&0.201&2.6&@xmath78 & 5.5&@xmath77 + sk @xmath1868140 & 5.5&@xmath76&3.0&@xmath80&@xmath181.97&0.215&2.6&@xmath78 & 1.4&@xmath77 + sk @xmath1868155 & 4.2&@xmath76&9.8&@xmath80&@xmath181.35&0.217&1.9&@xmath78 & 4.5&@xmath77 + sk @xmath1869228 & 4.0&@xmath76&4.9&@xmath81&@xmath182.61&0.145&2.8&@xmath78 & 3.4&@xmath80 + sk @xmath1869279 & 3.5&@xmath76&2.0&@xmath77&@xmath180.98&0.221 & 1.6&@xmath78&9.0&@xmath77 + + sk @xmath186619/sk @xmath186983 & 7.0&@xmath76&1.6&@xmath77&@xmath181.64&0.25 & 2.5&@xmath78&6.4&@xmath77 + sk @xmath18672/sk @xmath186635 & 1.0&@xmath76&8.9&@xmath77&@xmath180.19&0.18 & 5.6&@xmath76&5.0&@xmath76 + sk @xmath186826/sk @xmath186635 & 3.5&@xmath76&2.3&@xmath77&@xmath180.94&0.19 & 1.8&@xmath78&1.2&@xmath76 + sk @xmath1868129/sk @xmath186841 & 4.0&@xmath76&1.1&@xmath77&@xmath181.27&0.17 & 2.3&@xmath78&6.6&@xmath77 + lmcavg curve & & & @xmath181.00&&1.1&@xmath78&5.9&@xmath77 + sk @xmath1868140/sk @xmath186841 & 4.0&@xmath76&3.0&@xmath80&@xmath181.83&0.20 & 2.0&@xmath78&1.5&@xmath77 + sk @xmath1868155/sk @xmath1867168 & 5.0&@xmath76&9.8&@xmath80&@xmath181.42&0.20 & 2.5&@xmath78&4.9&@xmath77 + sk @xmath1869228/sk @xmath186515 & 3.5&@xmath76&4.9&@xmath81&@xmath182.56&0.15 & 2.3&@xmath78&3.3&@xmath80 + sk @xmath1869270/sk @xmath186778 & 3.5&@xmath76&0.7&@xmath77&@xmath181.69&0.19 & 1.8&@xmath78&3.7&@xmath77 + sk @xmath1869279/sk @xmath186563 & 4.0&@xmath76&2.0&@xmath77&@xmath181.04&0.21 & 1.9&@xmath78&9.7&@xmath77 + lmc2 curve & & & @xmath181.50&&1.9&@xmath78&3.1&@xmath77 + | we present an extinction analysis of nine reddened / comparison star pairs in the large and small magellanic clouds based on _ far - ultraviolet spectroscopic explorer ( fuse ) _ fuv observations . to date , just two lmc sight lines have probed dust grain composition and size distributions in the magellanic clouds using spectral data for wavelengths as short as 950 .
we supplement these two with data from 4 regions distinguished by their ir through uv extinction curves and grouped as lmcavg , lmc2 , smc bar and smc wing . despite the distinct characters of extinction in the clouds and milky way , our results are generally analogous to those found for galactic curves
namely , that the _ fuse _ portions of each extinction curve are described reasonably well by fitzpatrick & massa curves fitted only to longer wavelength data and lack any dramatic new extinction features , and any deviations from the cardelli , clayton , & mathis ( ccm ) formalism continue into fuv wavelengths .
a maximum entropy method analysis of all of these curves suggests that lmcavg and smc wing sight lines , whose extinction parameters more closely resemble those for galactic paths , require more silicon and/or carbon in dust than current abundance measurements would indicate are available .
the requirements for lmc2 and smc bar sight lines do not fully tax the available reservoirs , in part because large grains contribute less to the extinction in these directions .
an intermediate product of this extinction analysis is the measurement of new h@xmath0 abundances in the magellanic clouds . collectively considering cloud sight lines that possess significant h@xmath0 column densities , @xmath1/@xmath2 ( ) ratios
are reduced by significant factors relative to the galactic mean , whereas the corresponding @xmath1/@xmath2(h@xmath0 ) values more closely resemble their galactic counterpart .
these trends reflect the fact that among these sight lines @xmath3(h@xmath0)-values are lower than those common in the milky way for paths with similar degrees of reddening . |
the radio emission of jets in the hard state of black - hole binaries and of extragalactic parsec - scale radio sources is often flat in the @xmath5 representation , @xmath6 , with @xmath7 , where @xmath8 is the photon energy , e.g. , @xcite , @xcite , @xcite .
the emission of this type of jet is usually interpreted as due to the partially self - absorbed synchrotron process in a continuous conical jet ( @xcite , hereafter bk79 ) .
the emission at a given energy is self - absorbed from its onset up to a certain height of @xmath9 , and it is optically thin at higher @xmath10 .
the jet becomes optically thin at all @xmath11 at energies above the synchrotron break energy , @xmath12 .
the model of bk79 is also assumed in an important method of measuring magnetic fields of extragalactic jets from core shifts , i.e. , angular displacements with frequency of the observed maxima of jet radio emission @xcite .
this method can also be used to measure the jet power @xcite . the angular emissivity pattern of such continuous , partially optically - thick , jets differs from that of optically - thin jets , either steady - state or in the form of moving blobs @xcite .
the case of the angular dependence of the emission from partially self - absorbed jets was considered by @xcite . here ,
we re - examine this problem , and obtain different results at viewing angles close to the jet axis .
the cause of this discrepancy is that the usual approximation that we view the jet from a side in the comoving frame breaks down in that case .
we also consider implications of our results for the ratio of the jet - to - counterjet emission .
we then apply our results to the compact radio jet observed from cyg x-1 in its hard spectral state .
the angles in the comoving frame , @xmath13 , are shown for the jet and counterjet by the solid and dashed curves , respectively .
the top solid and bottom dashed ( red ) and the two middle ( black ) curves are for @xmath14 and 5/3 ( @xmath15 ) , respectively .
we see that at large @xmath3 , the jet - frame viewing angle for most values of @xmath16 is close to @xmath17 . ] from the origin , and at the radial distance @xmath18 from the jet axis .
, width=283 ] we develop here the formalism of @xcite , hereafter zls12 .
we use the notation similar to that of zls12 , except that now the doppler factor is @xmath19 , the dimensionless jet length , @xmath20 , is in units of the distance of the position of the onset of the synchrotron emission , @xmath21 , as in @xcite , and we take into account the effect of cosmological redshift , which we denote as @xmath22 .
also , the jet radius is given now accurately as @xmath23 rather than by @xmath24 , where @xmath11 is the distance from the jet origin and @xmath25 is the half - opening angle .
we consider the model of a conical jet moving with a constant bulk velocity , @xmath26 , and emitting synchrotron radiation ( bk79 ) .
we are concerned with the angular distribution of the emission of such jets .
we observe the jet at an angle , @xmath16 , with respect to the jet axis .
this angle in the jet frame is given by @xmath27 where @xmath3 is the bulk lorentz factor .
the counterjet is viewed at @xmath28 .
examples of the relationship between @xmath16 and @xmath13 are shown in fig .
[ transform ] .
we see that at large @xmath3 , the jet - frame viewing angle for most values of @xmath16 is close to @xmath17 .
the jet is viewed at @xmath29 in the jet frame for @xmath30 , see case ( iii ) in fig .
[ geometry ] . this case is also equivalent to @xmath31 and @xmath32 .
hereafter , we omit primes for other quantities in the comoving frame , and use a prime instead for quantities being integrated .
we note , however , that unlike the case of a moving blob of a constant size , often assumed to model blazars , the comoving frame in our case is not stationary .
both the distance from the origin and the radius increase with time in the comoving frame , and the magnetic field and the electron density decrease with time .
thus , a photon directed at @xmath29 in the comoving frame moves in a changing environment , in spite of its comoving position being constant .
therefore , we perform our calculations in the observer frame , in which the jet is stationary ( though its medium is moving ) .
we do it using the emission and absorption coefficients transformed from the comoving frame , see below .
the steady - state electron distribution per unit volume in the comoving frame at a given point ( defined in the observer frame ) is assumed to be a of a power - law form , @xmath33 where @xmath34 is the normalization and @xmath35 is the electron lorentz factor .
following bk79 , we assume conservation of the electron distribution along the jet , and conservation of the energy flux in the toroidal component of the magnetic field .
we also define the dimensional energy , @xmath8 , in the observer s frame and a dimensionless one , @xmath36 , in the jet frame .
thus , we have @xmath37 where @xmath21 corresponds to the onset of emission . for calculation of the synchrotron emission and absorption , we assume the magnetic field is tangled @xcite , which implies the emission in the local frame is isotropic . the emission coefficient from isotropic relativistic power - law electrons per unit volume at the jet frame can then be written as , @xmath38 where @xmath39 is the critical magnetic field , @xmath40 is the reduced planck constant , @xmath41 is the electron charge , and @xmath42 ( @xmath43 for @xmath44 ) follows from averaging over the pitch angle , see , e.g. , zls12 .
the emission coefficient in the local observer frame at @xmath8 equals @xmath45 ( e.g. , @xcite ) .
the synchrotron self - absorption coefficient averaged over the pitch angle for a power - law electron distribution can be expressed as , @xmath46 where @xmath47 is the fine - structure constant , and the constant @xmath48 ( @xmath49 for @xmath44 ) is given , e.g. , in zls12 .
the absorption coefficient in the local observer frame at @xmath8 equals @xmath50 .
the source function , @xmath51 , is then , @xmath52 which becomes @xmath53 in the observer frame .
in order to obtain the spectrum observed from the jet , we need to calculate the emission towards the observer from a given location of the jet in the observer frame and then integrate it over the jet projected area .
we can do it by integrating the radiative transfer equation over the line of sight , e.g. , equation ( 1.29 ) in @xcite , and then integrate the solution over the projected area . in the conical geometry and for an arbitrary angle , this can be done only numerically , which we do in appendix [ exact ] . however ,
if the viewing angle in the observer frame is @xmath54 , see the case ( iii ) in fig .
[ geometry ] , we can approximate the jet locally as a cylinder , and neglect the variation of @xmath34 and @xmath55 along the line of sight .
the above approximation also implies that the source function is constant along the line of sight . in this approximation
, the radiative transfer equation solves for the observed intensity as @xmath56 $ ] , where @xmath57 is the optical depth to synchrotron self - absorption integrated over the entire line of sight , @xmath58^{1/2 } , \label{tau1}\ ] ] where @xmath59 is the distance perpendicular to both the jet axis and the line of sight . the flux ( for either jet or counterjet )
is then given by ( zls12 ) @xmath60 , \label{int1}\ ] ] where @xmath61 is the luminosity distance .
since uppermost parts of the jet usually contribute weakly to the total flux , we assume here the jet extends to infinity .
hereafter , the redshift terms are included , generalizing the corresponding equations in zls12 .
the unit of energy in the flux is assumed to be the same as the unit of photon energy ( e.g. , both can be in ev or erg ) , which results in the unit of the flux of @xmath62 s@xmath63 . to get the flux in , e.g. , erg and the photon energy in ev , we need to multiply the above formula by erg / ev .
above the synchrotron break energy , @xmath12 , see equation ( [ break1 ] ) below , the entire jet emission is optically thin , with @xmath64 .
below it , the jet is optically thick from @xmath21 up to some distance , and optically thin above it .
this results in the spectral index of the partially optically thick spectrum of @xmath65 regardless of @xmath66 ( as obtained by bk79 ) . in order to get the normalization of this spectrum
, we can solve equation ( [ int1 ] ) .
this is equivalent to substituting equation ( 23 ) in equation ( 22 ) of zls12 , which yields @xmath67 which is valid for @xmath68 .
@xmath69 is given by a dimensionless double integral defined in zls12 , which we integrate here analytically , @xmath70\right\}\\ & = { 2\sqrt{\upi}\ , \gamma\left(5\over 2 p+8\right ) \gamma\left(p-1\over p+4\right)\over ( p+9 ) \gamma\left(p+9\over 2 p+8\right ) } , \end{aligned}\ ] ] where @xmath71 is the gamma function .
example values of @xmath72 are @xmath73 , 2.10 , 1.61 for @xmath74 , 3 , 4 , respectively .
the flux dependence on the angle is @xmath75 , see also @xcite .
this can be compared to the angular dependence of a steady - state optically - thin jet of @xmath76 ( bk79 ; @xcite ) for @xmath65 .
we see that at large angles the two dependencies are relatively similar , having similar powers of @xmath19 , e.g. , 13/6 and 16/7 for the partially optically - thick jet with @xmath74 , 3 , respectively , vs. 2 for an optically thin jet . however , at small viewing angles we have @xmath77 , which leads to a strong reduction of the flux , formally to null at @xmath78
. however , we point out that the approximation of equation ( [ int1 ] ) breaks down when the viewing angle is substantially different from @xmath29 , in particular for @xmath79 or @xmath80 , see fig .
[ geometry ] . this is because a given line of sight goes through regions with variable magnetic field and the electron normalization , and @xmath81 is no longer constant along it .
furthermore , the length along a line of sight is no longer given by that in a cylinder , which is assumed in equation ( [ tau1 ] ) .
generally , the integration of the attenuated source function over the jet volume needs to be done numerically , see appendix [ exact ] .
furthermore , the treatment needs to be changed for @xmath82 , in which case we no longer see the jet side , but instead view it from the top .
an analogous effect occurs for the bottom view .
these cases require a change of the method of the integration of the radiative transfer solution over the projected area ( appendix [ exact ] ) .
thus , the question arises whether the angular dependence dominated by the above power of @xmath83 actually takes place . in order to address this question ,
we consider the limiting approximation in which we view the jet at @xmath78 , see case ( i ) in fig .
[ geometry ] .
a conical jet has divergent velocities , see , e.g. , @xcite , which effect we neglect , and assume that @xmath78 along all the lines of sight through the jet .
we note that the entire emission originating at a given distance , @xmath11 , is now attenuated by the same optical depth , corresponding to the positions @xmath84 .
thus , we can simplify the radiative transfer equation for this case and multiply the source function by the jet cross section , @xmath85 , where @xmath86 is the jet radius .
we can then write the radiative transfer equation directly for the flux , @xmath87 , where @xmath88 is measured from the top of the jet down to the intersection of the line of sight with the jet boundary ( at @xmath89 ) .
uppermost parts of the jet contribute weakly to the total emission , as well as the part closest to the jet origin are strongly self - absorbed , which is the case for photon energies well below the synchrotron break , @xmath12 , see below .
thus we calculate the above integral from 0 to @xmath90 .
( @xmath91 ) .
the blue dotted curve shows the distributions for an optically thin steady - state jet , i.e. , @xmath92 .
the red solid curve shows the exact dependence for a partially optically - thick jet , see appendix [ exact ] , for @xmath74 and @xmath93 .
this dependence reaches the value of the small - angle limit of equation ( [ onaxis ] ) at @xmath78 .
the black dashed curve shows the cylindrical approximation of equation ( [ thick ] ) , which is exact at @xmath29 , but it breaks down at small angles .
the flux is normalized to unity at @xmath94 . ]
we first calculate @xmath57 from a given @xmath10 up to infinity at @xmath95 , @xmath96 where @xmath97 is the doppler factor at @xmath78 .
we note that equation ( [ taus ] ) is equivalent to the limit of @xmath98 of the expression for @xmath57 in the side - view case , see , e.g. , equation ( 19 ) of zls12 .
namely , in that case , @xmath99 .
we have @xmath100 .
thus , @xmath101 , which , for @xmath78 , equals @xmath102 , as above . we also note that although @xmath57 is invariant between different frames , it does depend on @xmath19 .
we can then write the equation for the flux as @xmath103 where @xmath104 follows from equation ( [ taus ] ) and @xmath105 $ ] .
the above integral yields a gamma function , and we obtain a spectrum with @xmath65 , @xmath106^{5\over 4+p}\!\ ! \gamma\left(\frac{p-1}{p+4}\right ) .
\label{onaxis}\ ] ] the flux dependence on the doppler factor is the same as of the side - view spectrum , equation ( [ thick ] ) , namely @xmath107 .
parenthetically , we note that the often - used approximation of calculating the emission from the part with @xmath108 as optically thin and neglecting the contribution from the optically - thick part yields almost the same result , with the gamma function above replaced by inversion of its argument , @xmath109 , which fractional error is less than 13 per cent for @xmath110 .
it is of interest to calculate the ratio of the on - axis emission , equation ( [ onaxis ] ) , to that seen from a side , equation ( [ thick ] ) , @xmath111^{5\over 4+p}\!\!\gamma\left(\frac{p-1}{p+4}\right ) . \label{ratio}\ ] ] we see that , apart from the dependence on the viewing angle , this ratio depends only on the jet opening angle and velocity , and on the electron index , @xmath66 . in order to get a characteristic value of this ratio , we calculate it at @xmath112 , @xmath113 . for that angle , @xmath114 .
then , we express the jet opening angle as a fraction of @xmath115 , @xmath116 . in theoretical models ,
often @xmath117 is assumed ( e.g. , @xcite ) , but observationally @xmath118 is found . in particular , @xcite and @xcite found the average values in their samples of extragalactic jets of @xmath119 , @xmath120 , respectively .
for jets in black - hole binaries , @xcite also found @xmath121 .
we note that the factor @xmath122 can be theoretically connected in some jet models to the magnetization parameter , @xmath123 @xcite . using the above relations
, we obtain @xmath124 we see that this ratio depends only on the fractional jet opening angle , @xmath122 , the electron index , and @xmath125 , which factor can assume values only between 1 and 2 .
we also see that this ratio is large unless @xmath122 is very small . for example , for @xmath126 and @xmath74 , 3 , 4 , it is @xmath127 , @xmath128 , @xmath129 , respectively .
the solid ( red ) curve in fig .
[ angular ] shows an example of the exact angular distributions of a partially optically - thick jet , calculated in appendix [ exact ] , which reaches the values of the analytical formuale at @xmath130 and @xmath17 .
we see that the flux seen on axis is substantially higher than the maximum of the flux in the large - angle approximation , shown by the dashed ( black ) curve .
this appears to be a new result ; e.g. , @xcite derived the maximum flux and the corresponding viewing angle using the large - angle approximation ( as for the dashed curve in fig .
[ angular ] ) , while we find the overall maximum to occur at @xmath78 , with the flux decrease with the decreasing @xmath16 of that approximation being spurious .
we then compare the values of the synchrotron break energy , @xmath12 . for the top view , we solve @xmath131 with equation ( [ taus ] ) , @xmath132^{2\over 4+p}. \label{break0}\ ] ] in the side - view case , we have ( zls12 ) , @xmath133^{2\over 4+p}. \label{break1}\ ] ] with the same assumptions as above , we obtain with equation ( [ break0 ] ) , @xmath134^{-{2\over 4+p}}. \label{eratio}\ ] ] this ratio is typically @xmath1351 , e.g. , it is @xmath1361.5 for @xmath137 , @xmath74 , @xmath93 .
the emission of the counterjet corresponds to the viewing angle of @xmath28 .
we note that @xmath138 , and @xmath139 .
the doppler factor of the counterjet is then @xmath140 equation ( [ thick ] ) then applies to to the counterjet emission with the substitution of @xmath141 .
when we consider the flux ratio between the jet and counterjet in the side - view approximation , the angular dependencies on @xmath83 in equation ( [ thick ] ) cancel each other . thus , we find the jet to counterjet flux ratio under the assumption of both components being intrinsically symmetric of @xmath142 the power - law index of this dependence changes from 2 for @xmath143 to @xmath1363 for @xmath144 , different from the index of 2 for an optically thin jet with @xmath65 .
the total emission is , obviously , the sum of the fluxes from the jet and the counterjet .
we note that equation ( 22 ) of zls12 implies an incorrect ratio , due to the neglect of the different value of @xmath12 between the jet and counterjet .
equation ( [ fratio ] ) can be solved for @xmath145 , @xmath146 we now consider the case of the counterjet emission at @xmath79 . at this approximation
, we have considered the emission from the top of the jet , equations ( [ fi0][onaxis ] ) .
however , this counterjet emission corresponds to that from the jet bottom , @xmath147 , see the case ( ii ) in fig .
[ geometry ] , for which those equations are not applicable .
we can still use the approximation of a ray along the jet axis , but the direction of the emission is opposite .
this also means that we can not multiply the source function by the jet cross section , since this emission will now be attenuated by different @xmath57 depending on the distance from the axis .
using radiative transfer , we write @xmath148 where @xmath88 , given by equation ( [ taus ] ) , corresponds to the intersection of the line of sight with the jet boundary at the radial distance @xmath18 . then the optical depth along the ray emitted at @xmath149 is @xmath150 , i.e. , it is measured from the intersection with the boundary .
we can now change the variable of the outer integration to @xmath57 [ using equation ( [ taus ] ) ] .
the resulting double integral can be calculated analytically , and the final result is @xmath151^{5\over 4+p}\left(\cot{5\upi\over 4+p } - \cot{\upi\over 4+p}\right){\gamma\left(\frac{-4}{p+4}\right)\over \gamma\left(\frac{1}{p+4}\right)}.\label{flux_cj } \end{aligned}\ ] ] we can obtain the jet - to - counterjet flux ratio for this case using equation ( [ onaxis ] ) . we find it is almost the same , within @xmath13510 per cent , as that given by equation ( [ fratio ] ) .
this shows that equation ( [ fratio ] ) is valid quite generally , in spite of the underlying approximation breaking down at @xmath16 substantially different from @xmath29 .
naturally , equation ( [ flux_cj ] ) with a replacement of @xmath152 also gives the jet emission at @xmath147 .
we also note that we have neglected here a possible ( and likely in a range of angles ) obscuration of the counterjet emission at @xmath79 , by the accretion disc and stellar wind .
also , the jet will synchrotron - absorbed reprocess the counterjet synchrotron emission .
@xcite have obtained radio maps of the black - hole binary cyg x-1 using vlba and vla at 8.4 ghz .
they have found no evidence for the presence of a counterjet , and constrained the flux ratio to @xmath153 .
they assumed @xmath154 and optically - thin emission in calculating constraints on the jet velocity .
currently , a lower value of @xmath16 appears to be the inclination of the orbit of cyg x-1 , in particular @xcite found @xmath155 . on the other hand , @xcite , considering also the evolutionary status of the system , found @xmath156 as the most likely range . for this range ,
equation ( [ beta ] ) at @xmath74 gives @xmath157 , corresponding to @xmath158 .
the value of @xmath66 has been constrained by broad - band spectral models of zls12 , @xcite and @xcite , which give the most likely range of @xmath159 .
the resulting limits on @xmath145 are relatively insensitive to the value of @xmath66 , as illustrated in fig .
[ beta_p ] .
a caveat for this result is that the jet is likely aligned with the black - hole rotation axis , which may be misaligned with the normal to the binary plane . in this case
, the jet inclination may be not given by the above constraints .
possible indications for this to be the case are given by the results of @xcite and @xcite , who have fitted x - ray data from observations of cyg x-1 in the soft spectral state ( in which the inner disc is expected to extend to the innermost stable orbit ) by _ nustar _ and _ suzaku _ , and _ nustar _
, respectively .
their fits gave the inclination of the inner disc substantially larger than @xmath160 , namely @xmath161 and @xmath162 , respectively , which would indicate the plane of the inner disc inclined with respect to the orbital axis .
we caution , however , that this may be due to the limitations of the used models .
@xcite used the compton reflection model of @xcite , which averages over all viewing angles .
then , @xcite used the model of @xcite , which calculates angle - dependent reflection , but both models convolve their reflection spectra using the relativistic code of @xcite , which neglects a number of important effects , as discussed in @xcite .
thus , we consider the issue of the misalignment to be open .
other constraints on the jet velocity in cyg x-1 are by @xcite , who claimed @xmath163 or so based on non - detection of short - time scale correlations between radio and x - ray emission , though this limit appears model - dependent
. then @xcite estimated @xmath164 , which also relies on a number of assumptions .
still , if we accept those constraints at face value , the most likely jet velocity in cyg x-1 becomes @xmath165 .
, and the electron index , @xmath66 .
we use the range of the viewing angles found by @xcite and the lower limit on the ratio of the jet - to - counterjet flux ratio obtained by @xcite of @xmath166 .
the horizontal lines correspond to the previous limits given in their table 2 . ]
@xcite also constrained the half - opening jet angle of the projection of the jet on the sky as @xmath167 .
we note here that the actual half - opening angle is the one after de - projection , i.e. , multiplied by @xmath83 ( e.g. , @xcite ) .
given that @xmath168 , @xmath169 .
since the lower limit on @xmath3 is rather low , see above , this upper limit on the half - opening angle implies a very small factor @xmath170 , unless @xmath3 is much higher than the lower limit of @xcite obtained from the absence of an observed counterjet . if @xmath15 ( @xmath91 ) , @xmath171 .
this implies an extremely efficient collimation mechanism , e.g. , that related to a very low jet magnetization in the models of @xcite and @xcite .
also , this coefficient is substantially lower than those typically seen in blazars , where @xmath1720.2 @xcite .
we have studied the extended synchrotron jet model , originally proposed by bk79 .
we have considered three limiting analytical approximations to the flux vs. the viewing angle . in one ,
usually assumed , the jet is viewed sideways in the comoving frame .
this approximation implies that the flux becomes null when the jet is viewed on axis , with @xmath95 , e.g. , @xcite .
however , we point out that the above approximation breaks down in the low-@xmath16 regime , since the jet is no longer viewed sideways in the comoving frame . we have considered another limiting case , of the jet viewed on axis .
we have found an analytical solution of the radiative transfer integrated over the jet cross section in that case .
we have found out that this emission is rather strong , corresponding to the global maximum of the flux as a function of the viewing angle .
we have also calculated the emission corresponding to the emission angle of @xmath80 , which also corresponds to the counterjet emission in the case of @xmath95 .
then , we have obtained the general exact solution at an arbitary viewing angle numerically , described in appendix [ exact ] .
we have applied our results to the black - hole binary cyg x-1 . given
the jet - to - counterjet flux ratio of @xmath050 found observationally @xcite and the current estimate of the inclination of @xmath156 @xcite , we have found @xmath173 , @xmath174 . combining it with other published constraints , the most likely value is @xmath165 .
we have also pointed out that when the projection effect is taken into account , the radio observations imply the jet half - opening angle of @xmath169 , a half of the value given by @xcite . if @xmath3 is not much above the counterjet limit , the opening angle is @xmath175 , and much lower than the values typically observed in blazars .
we thank marek sikora for valuable discussions and the referee for valuable comments and suggestions .
this research has been supported in part by the polish national science centre grants 2012/04/m / st9/00780 , 2013/10/m / st9/00729 and 2015/18/a / st9/00746 .
blandford r. d. , knigl a. , 1979 , apj , 232 , 34 ( bk79 ) cawthorne t. v. , 1991 , in p. a. hughes , ed .
, beams and jets in astrophysics .
cambridge univ . press , cambridge , p. 187
clausen - brown e. , savolainen t. , pushkarev a. b. , kovalev y. y. , zensus j. a. , 2013 , a&a , 558 , a144 t. , wilms j. , reynolds c. s. , brenneman l. w. , 2010 , , 409 , 1534 fender , r. p. , pooley , g. g. , durouchoux , p. , tilanus , r. p. j. , brocksopp , c. , 2000 , mnras , 312 , 853 j. , kallman t. r. , 2010 , , 718 , 695 ghisellini g. , 2000 , in b. casciaro , d. fortunato , m. francaviglia and a. masiello , recent developments in general relativity .
springer , p. 5
, arxiv : astro - ph/9905181 gleissner t. , et al . , 2004 , a&a , 425 , 1061 healey s. e. , romani r. w. , taylor g. b. , sadler e. m. , ricci r. , murphy t. , ulvestad j. s. , winn j. n. , 2007 , apjs , 171 , 61 heinz s. , begelman m. c. , 2000 , apj , 535 , 104 komissarov s. s. , vlahakis n. , knigl a. , barkov m. v. , 2009 , mnras , 394 , 1182 knigl a. , 1981 , apj , 243 , 700 lind k. r. , blandford r. d. , 1985 , apj , 295 , 358 lobanov a. p. , 1998 ,
a&a , 330 , 79 malyshev d. , zdziarski a. a. , chernyakova m. , 2013 , mnras , 434 , 2380 malzac j. , belmont r. , fabian a. c. , 2009 , mnras , 400 , 1512 miller - jones j. c. a. , fender r. p. , nakar e. , 2006 , mnras , 367 , 1432 moderski r. , sikora m. , baejowski m. , 2003 , a&a , 406 , 855 niedwiecki a. , zdziarski a. a. , szanecki m. , 2016 , apj , 821 , l1 orosz j. a. , mcclintock j. e. , aufdenberg j. p. , remillard r. a. , reid m. j. , narayan r. , gou l. , 2011 , apj , 742 , 84 pushkarev a. b. , kovalev y. y. , lister m. l. , savolainen t. , 2009 , a&a , 507 , l33 pushkarev a. b. , hovatta t. , kovalev y. y. , lister m. l. , lobanov a. p. , savolainen t. , zensus j. a. , 2012 , a&a , 545 , a113 ross r. r. , fabian a. c. , 2005 , mnras , 358 , 211 rybicki g. r. , lightman a. p. , 1979
, radiative processes in astrophysics .
wiley - interscience , new york sikora m. , madejski g. , moderski r. , poutanen j. , 1997 , apj , 484 , 108 stirling a. m. , spencer r. e. , de la force c. j. , garrett m. a. , fender r. p. , ogley r. n. , 2001 , mnras , 327 , 1273 tchekhovskoy a. , mckinney j. c. , narayan r. , 2009 , apj , 699 , 1789 tomsick j. a. , et al . , 2014 , apj , 780 , 78 walton d. j. , et al .
, 2016 , apj , 826 , 87 zamaninasab m. , clausen - brown e. , savolainen t. , tchekhovskoy a. , 2014 , nature , 510 , 126 zdziarski a. a. , lubiski p. , sikora m. , 2012 , mnras , 423 , 663 ( zls12 ) zdziarski a. a. , stawarz . , pjanka p. , sikora m. , 2014a , mnras , 440 , 2238 zdziarski a. a. , pjanka p. , sikora m. , stawarz . , 2014b , mnras , 442 , 3243 zdziarski a. a. , sikora m. , pjanka p. , tchekhovskoy a. , 2015 , , 451 , 927 zikowski j. , 2014 , mnras , 440 , l61
as stated in section [ jet ] , we first obtain the intensity emitted by a given point of the jet projection in a given direction , and then integrate it over the projected area . in order to efficiently deal with the geometry of an inclined cone and its projection , we set up two coordinate systems in the observer frame .
one is the usual jet coordinate system , with the @xmath11 axis along the jet axis , and the @xmath59 and @xmath176 axes orthogonal to it .
the other system has the origin at the top of the jet , at the height @xmath177 , at @xmath178 in the jet coordinates , and it is rotated clockwise by the viewing angle , @xmath16 .
this corresponds to an anticlockwise rotation of the jet itself . since the optically - thin jet emission declines fast with the distance , the results are almost independent of the assumed value of @xmath177 as long as it is much larger than the height at which the emission at a given frequency becomes optically thin .
the coordinates in the rotated system are marked by the subscript @xmath16 .
the transformations from the jet system to the rotated one and the reverse one are @xmath179 respectively , where @xmath180 is the three - dimensional clockwise rotation matrix fixing the @xmath59 axis . the way we calculate the intensity depends on the viewing direction . for viewing from either the top or the side , @xmath181 , @xmath57
is taken to be 0 when the ray leaves the jet , and integrated backward to its maximum value at the ray origin .
for viewing the jet from the back , we follow the usual convention of the radiative transfer , and take @xmath57 to be 0 at the origin and integrate it to its maximum value at which the ray leaves the jet .
this corresponds to the expressions for the intensity in the jet frame ( see section [ def ] ) of @xmath182 where @xmath183 is the running length measured along the ray .
we then change the variable of integration from @xmath183 to the height , @xmath11 , measured in the jet coordinates as it is the only geometric quantity that the source and absorption functions vary with , the other quantities being constant .
this gives us @xmath184^{-1 } \int_{z_{1}}^{z_{2 } } j\left(z\right ) { \rm e}^{\tau(z)-\tau \left(z_{1}\right ) } { \rm d}z , & i \in ( \upi-\theta_{\rm j},\upi),\cr \end{cases}\\ & \tau(z)=\frac{2 \alpha_0 z_0^{\frac{p+6}{2 } } \epsilon ^{-\frac{p+4}{2 } } } { ( p+4 ) \delta_{\rm j } } \left(z^{-\frac{p+4}{2}}-z_{2}^{-\frac{p+4}{2}}\right)\times \begin{cases}(\cos i)^{-1 } , & i \in ( 0,\upi-\theta_{\rm j } ) ; \cr [ \cos ( i-\upi)]^{-1 } , & i \in ( \upi-\theta_{\rm j},\upi),\cr \end{cases}\label{tau}\end{aligned}\ ] ] where the boundaries @xmath185 and @xmath186 are derived below . since we are concerned with the emission in the partially optically - thick regime , we calculate the emission of the entire jet , down to @xmath187 , as in sections [ jet ] and [ cjet ] , and do not impose in numerical calculations @xmath188 . from equation ( [ tau ] )
, we can readily determine the value of @xmath11 corresponding to @xmath189 .
given that we calculate the intensity in the rotated coordinate system , we have to find the heights of the intersections as functions of @xmath190 and @xmath191 . for
a given a point , @xmath192 , in the jet coordinates and the viewing angle , @xmath16 , we can define a straight line at that angle that passes through the point .
we need to find the intersections of this line with the cone , given by @xmath193 the intersections with the cone can be found by squaring both sides and solving the resulting quadratic equation .
however , squaring the equation for the cone also gives us a second cone flipped over the @xmath59-@xmath176 plane ( the counterjet ) .
when the viewing angle is less than the jet opening angle , the first intersection is on the second cone and when it is greater than @xmath194 , the second intersection will be on the second cone , so the functions are assigned the value of infinity at these angles so that these cases can be separated .
the equations are then written in terms of the rotated coordinate system using the transformation of equation ( [ rotated ] ) with @xmath195 , since the area will be projected onto the @xmath190-@xmath191 plane .
this yields , @xmath196 the heights of the intersections in the jet coordinates are then given by @xmath197 , on the viewing angle for @xmath198 ( left ) and @xmath199 ( right ; this value is the same as that used in fig .
[ angular ] ) , both with @xmath200 .
the curves show the dependence for the case ( i ) at @xmath82 ( blue ) , case ( iii ) at @xmath201 ( red ) and case ( ii ) at @xmath82 ( green ) .
, title="fig:",width=302]0.5 cm , on the viewing angle for @xmath198 ( left ) and @xmath199 ( right ; this value is the same as that used in fig .
[ angular ] ) , both with @xmath200 .
the curves show the dependence for the case ( i ) at @xmath82 ( blue ) , case ( iii ) at @xmath201 ( red ) and case ( ii ) at @xmath82 ( green ) .
, title="fig:",width=302 ] in the next step , we integrate the intensity over the jet projected area in the rotated coordinates . depending on the orientation of the observer with respect to the jet
, we have three different cases , illustrated in fig . [ cones ] .
( i ) @xmath82 , for which the projected area changes from a circle at @xmath78 but it contracts into an ellipse as @xmath16 gets larger .
( ii ) @xmath202 , which is very similar but the jet is viewed from the back .
( iii ) @xmath201 , in which case the projected area is shaped similarly to an ice cream cone .
two examples of the changes of the projected area with @xmath16 are shown in fig .
[ area ] . in the cases ( i ) and ( ii ) , the boundaries are easy to calculate ; we can just apply the transformation of equation ( [ rotated ] ) to the equation for a circle in the jet coordinates at height @xmath177 .
this yields @xmath203 ^ 2}}^{\sqrt{z^2 -[y_i/\cos ( i-\upi)]^2 } } i { \rm d}x_i { \rm d}y_i,\label{bottom_view}\end{aligned}\ ] ] in the cases ( i ) and ( ii ) , respectively .
we can see that these equations become identical to the flux given by equations ( [ fi0 ] ) and ( [ fcj ] ) in the cases of @xmath204 and @xmath205 , respectively , for @xmath206 . in the case ( iii ) , the integral must be split into two parts in order to accommodate the change of the domain from triangular to a part of an ellipse . directly finding how the edge of the cone transforms
is difficult but it is easy to find where the base of the jet is in the rotated coordinate system .
we also know the shape of the ellipse and that the edge must intersect exactly once tangentially , so we can form the line using these two points .
the intersections at the ellipse are given by @xmath207 and @xmath208 in the @xmath190-@xmath191 plane , where @xmath209 the edge lines , giving some of the integration limits in equation ( [ flux3 ] ) below , are then given by , @xmath210 past the intersection , the domain is the remaining part of the ellipse although now a modulus sign is needed to ensure that we are always integrating to the furthest point on the ellipse even when the viewing angle passes @xmath211 and @xmath212 changes sign .
this leads us to the final equation , @xmath213 this equation becomes identical to the flux of equation ( [ int1 ] ) in the case of @xmath214 for @xmath206 .
[ fluxes ] shows two examples of this solution , and compares it to the cylindrical approximation , equations ( [ int1 ] ) and ( [ thick ] ) . , @xmath215 , @xmath216 , @xmath200 ( which is much larger than the value of @xmath11 corresponding to @xmath189 , which implies the results being virtually independent of @xmath177 ) , and @xmath217 ( left ) and 0.8 ( middle and right ) .
the middle and right panels have the linear and logarithmic vertical axes , respectively .
the upper curves show the exact model for the case ( i ) at @xmath82 ( blue ) , case ( iii ) at @xmath201 ( red ) and case ( ii ) at @xmath82 ( green ) .
the lower ( black ) curves shows the cylindrical model for the flux , equations ( [ int1 ] ) , ( [ thick ] ) .
we see the two regimes connect smoothly in the exact model and the cylindrical approximation becomes more accurate for increasing viewing angle.,title="fig:",width=221]0.01 cm , @xmath215 , @xmath216 , @xmath200 ( which is much larger than the value of @xmath11 corresponding to @xmath189 , which implies the results being virtually independent of @xmath177 ) , and @xmath217 ( left ) and 0.8 ( middle and right ) .
the middle and right panels have the linear and logarithmic vertical axes , respectively .
the upper curves show the exact model for the case ( i ) at @xmath82 ( blue ) , case ( iii ) at @xmath201 ( red ) and case ( ii ) at @xmath82 ( green ) .
the lower ( black ) curves shows the cylindrical model for the flux , equations ( [ int1 ] ) , ( [ thick ] ) .
we see the two regimes connect smoothly in the exact model and the cylindrical approximation becomes more accurate for increasing viewing angle.,title="fig:",width=221]0.01 cm , @xmath215 , @xmath216 , @xmath200 ( which is much larger than the value of @xmath11 corresponding to @xmath189 , which implies the results being virtually independent of @xmath177 ) , and @xmath217 ( left ) and 0.8 ( middle and right ) .
the middle and right panels have the linear and logarithmic vertical axes , respectively .
the upper curves show the exact model for the case ( i ) at @xmath82 ( blue ) , case ( iii ) at @xmath201 ( red ) and case ( ii ) at @xmath82 ( green ) .
the lower ( black ) curves shows the cylindrical model for the flux , equations ( [ int1 ] ) , ( [ thick ] ) .
we see the two regimes connect smoothly in the exact model and the cylindrical approximation becomes more accurate for increasing viewing angle.,title="fig:",width=221 ] | we study the angular dependence of the flux from partially synchrotron self - absorbed conical jets ( proposed by blandford & knigl ) .
we consider the jet viewed from either a side or close to on axis , and in the latter case , either from the jet top or bottom .
we derive analytical formulae for the flux in each of these cases , and find the exact solution for an arbitrary angle numerically .
we find that the maximum of the emission occurs when the jet is viewed from top on - axis , which is contrast to a previous result , which found the maximum at some intermediate angle and null emission on - axis .
we then calculate the ratio of the jet - to - counterjet emission for this model , which depends on the viewing angle and the index of power - law electrons .
we apply our results to the black - hole binary cyg x-1 . given the jet - to - counterjet flux ratio of @xmath050 found observationally and the current estimates of the inclination
, we find the jet velocity to be @xmath1 .
we also point out that when the projection effect is taken into account , the radio observations imply the jet half - opening angle of @xmath2 , a half of the value given before .
when combined with the existing estimates of @xmath3 , the jet half - opening angle is low , @xmath4 , and much lower than values observed in blazars , unless @xmath3 is much higher than currently estimated .
[ firstpage ] acceleration of particles galaxies : jets radiation mechanisms : non - thermal radio continuum : stars stars : individual : cyg x-1stars : jets . |
* theoretical model and simulations . *
we consider the @xmath0-@xmath1-@xmath2 classical heisenberg model on a simple square lattice @xcite .
the hamiltonian can be expressed as @xmath5 where @xmath6 represents the normalized spin at the site @xmath7 , @xmath8 .
@xmath9 , @xmath10 , and @xmath11 run over all the nearest - neighbor ( nn ) , next - nearest - neighbor ( nnn ) , and next - next - nearest - neighbor ( nnnn ) sites in the magnetic layer , respectively .
@xmath12 , @xmath13 , and @xmath14 are the coefficients for the nn , nnn , and nnnn exchange interactions , respectively .
@xmath15 is the magnetic ( zeeman ) field applied along the @xmath16-direction , @xmath17 is the perpendicular magnetic anisotropy ( pma ) constant , @xmath18 represents the ddi , i.e. , the demagnetization .
the total energy of the given system contains the nn exchange energy , the nnn exchange energy , the nnnn exchange energy , the anisotropy energy , the zeeman energy , and the demagnetization energy .
the simulation is carried out with the 1.2a5 release of the object oriented micromagnetic framework ( oommf ) software @xcite .
the standard oommf extensible solver ( oxs ) objects are employed , including the oxs object for the calculation of the nn exchange interaction .
in addition , we have developed the oxs extension modules for the calculation of the nnn and nnnn exchange interactions , which are implemented in our simulation of the @xmath0-@xmath1-@xmath2 classical heisenberg model .
the time - dependent spin dynamics in the simulation is controlled by the landau - lifshitz - gilbert ( llg ) equation @xcite , @xmath19 where @xmath20 is the effective field , @xmath21 is the time , @xmath22 is the gilbert damping coefficient , and @xmath23 is the gyromagnetic ratio . for the simulation including a vertical spin current generated by the spin hall effect ,
the slonczewski - like spin - transfer torque ( stt ) will be added to the right - hand side of equation ( [ eq : llg ] ) , @xmath24 where @xmath25 , @xmath26 is the current density , @xmath27 is the spin hall angle , @xmath28 is the lattice constant , @xmath29 is the saturation magnetization , @xmath30 is the spin - current polarization direction .
besides , we also employed the oommf conjugate gradient ( cg ) minimizer for the spin relaxation simulation , which locates local minima in the energy surface through direct minimization techniques @xcite .
-@xmath13-@xmath14 ferromagnetic thin film . * ( * a * ) phase diagram obtained by relaxing the magnetic thin film with a random initial spin configuration .
( * b * ) the spin configuration of the relaxed sample with @xmath31 , where both skyrmions and antiskyrmions are formed .
the topological number @xmath32 is indicated .
the color scale represents the out - of - plane spin component ( @xmath33 ) , which has been used throughout this paper .
the arrow denotes the in - plane spin components ( @xmath34 , @xmath35 ) .
the simulation is carried out by the oommf cg minimizer .
the model is a square element with the open boundary condition ( obc ) , which has @xmath36 @xmath37 @xmath36 spins .
the fixed parameters are @xmath38 and @xmath39 .
@xmath17 and @xmath40 are varied between @xmath41 , respectively .
, scaledwidth=48.0% ] and the helicity @xmath42 .
* we construct a radial symmetric antiskyrmion as the initial state .
we check whether the initial geometry is preserved or broken after the relaxation .
the model is a square element with @xmath36 @xmath37 @xmath36 spins . the parameters are @xmath38 , @xmath39 , @xmath43 , and @xmath44 . the simulation on the spin relaxation is carried out by the oommf cg minimizer .
, scaledwidth=48.0% ] in this work , the values for the nn , nnn , and nnnn exchange interactions are given as @xmath45 mev , @xmath38 , and @xmath39 , respectively .
@xmath46 , @xmath47 , @xmath48 , @xmath49 m a@xmath50 s@xmath50 , and @xmath51 ka m@xmath50 .
the geometry of the cubic cells generated by the spatial discretization is @xmath52 @xmath37 @xmath52 @xmath37 @xmath52 .
we define the skyrmion number of a spin texture in the continuum limit by the formula @xmath53 the normalized spin field @xmath54 takes a value on the 2-sphere @xmath55 .
the skyrmion number @xmath32 counts how many times @xmath54 wraps @xmath55 as the coordinate @xmath56 spans the whole planar space .
we parametrize the spin texture as @xmath57 with @xmath58 where @xmath59 is the azimuthal angle ( @xmath60 ) and @xmath42 is the helicity defined mod @xmath61 .
namely , the helicity @xmath42 and @xmath62 are identical .
special values of the helicity play important roles , that is , @xmath63 .
we frequently use @xmath64 instead of @xmath65 .
we may index a skyrmion as @xmath66 with the use of the skyrmion number and the helicity . a skyrmion with @xmath67
may also be called an antiskyrmion . * phase diagram . *
we first relax the frustrated magnetic system from a random initial spin configuration for different values of @xmath17 and @xmath40 .
the initial spin configuration is random in three dimensions .
the phase diagram of the relaxed state with respect to @xmath17 and @xmath40 is given in fig . [ fig1](a ) .
it shows that isolated skyrmions and antiskyrmions can be spontaneously formed in the relaxed state where the magnetic system has certain values of @xmath17 and @xmath40 .
small values of @xmath17 and @xmath40 lead to to the formation of distorted skyrmions and antiskyrmion , or large in - plane spin configurations .
figure [ fig1](b ) shows the detailed spin configuration of the relaxed sample with @xmath43 and @xmath68 .
it can be seen that skyrmions with @xmath69 , antiskyrmions with @xmath70 , @xmath71 , and @xmath72 coexist in the relaxed sample .
it is noteworthy that the high - topological - number antiskyrmion with @xmath73 can be seen as a two - antiskyrmion bound state or a bi - antiskyrmion , about which we will discuss later .
* stability of isolated skyrmions and antiskyrmions .
* we study possible static solutions of isolated skyrmions and antiskyrmions by relaxing the sample with a given initial spin configuration .
namely , we construct a number of possible standard skyrmion and antiskyrmion spin structures with different values of @xmath66 as the initial spin configuration of the sample .
after relaxation of the magnetic system , we get the relaxed states of skyrmions and antiskyrmions , which are stable or metastable solutions .
as shown in fig .
[ fig2 ] , we construct a radial symmetric skyrmion as the initial state .
we check whether the initial geometry is preserved or broken after the relaxation . for the initial states of skyrmions with @xmath74 ,
the numbers @xmath66 do not change after the relaxation although the radius shrinks .
however the skyrmion with @xmath75 is unstable and split into two skyrmions with @xmath76 .
the helicity @xmath42 changes during this splitting process from that of the initial state with @xmath4 .
the skyrmion with @xmath77 is also unstable and split into five skyrmions with @xmath78 and different @xmath42 .
the results of the antiskyrmion are similar to those of the skyrmion as shown in the * binary stable skyrmions and antiskyrmions .
* it has been demonstrated in ref . that the skyrmion energy and the antiskyrmion energy are degenerate and independent of the helicity @xmath42 in the absence of the ddi .
we also present a symmetric analysis of this property in methods .
needless to say , the ddi exists in all magnetic materials .
however , it has been customarily considered negligible for a microscopic spin texture .
this is actually not the case .
we show the detail of each contribution to the skyrmion energy in fig .
[ fig3 ] , where the ddi energy is found to be quite large . as indicated in fig .
[ fig4 ] , there are two effects from the ddi .
first , the degeneracy is resolved between the skyrmion and the antiskyrmion .
second , the degeneracy is resolved between various helicities in general .
the bloch - type ( nel - type ) skyrmions with @xmath4 ( @xmath79 ) are found to possess the minimum energy for @xmath78 ( @xmath80 ) .
it is consistent with the fact that the ddi is due to the magnetic charge @xmath81 and prefers the bloch - type structure @xcite . on the other hand ,
the energy is independent of the helicity for @xmath76 .
the helicity dependence of the energy of antiskyrmions is similar but weaker than that of skyrmions , as shown in fig .
[ fig4](b ) .
this is because an antiskyrmion has the antivortex structure , where the contribution of the ddi is smaller than the vortex structure of a skyrmion .
the helicity dependence of the energy is a new feature due to the ddi , which is overlooked in the previous literature @xcite . )
-skyrmion in the presence and absence of the ddi . * the contribution from the ddi energy is found to be quite large . , scaledwidth=45.0% ] for ( * a * ) relaxed skyrmions and ( * b * ) relaxed antiskyrmions as functions of the skyrmion number @xmath32 and the helicity @xmath42 . *
@xmath82 includes the nn exchange , nnn exchange , nnnn exchange , zeeman , pma , and ddi energies .
the lowest energy states are found to be two degenerate bloch states both for the skyrmion ( @xmath83 ) and the antiskyrmion ( @xmath84 ) due to the ddi .
the corresponding spin configurations for relaxed skyrmions and antiskyrmions are given in fig .
[ fig2 ] and , respectively .
, scaledwidth=48.0% ] as a result , all skyrmions ( @xmath78 ) with any @xmath42 relax to the bloch - type skyrmions with @xmath4 since they have the minimum energy .
exceptions are those possessing the precise initial values of @xmath79 .
indeed , we have studied the relaxation of a skyrmion with @xmath85 for different initial values of @xmath42 in the range of @xmath86 . as shown in fig .
[ fig5 ] , the skyrmion with the initial helicity of @xmath63 keeps their helicity during the relaxation .
however , for the skyrmion with other initial helicities , the helicity is relaxed to @xmath4 , justifying that the most stable skyrmions have @xmath4 .
consequently , the stable skyrmion is the bloch - type skyrmion carrying the internal degree of freedom indexed by a binary value ( @xmath4 ) . as functions of the iteration for (
* a * ) skyrmions and ( * b * ) antiskyrmions with varied initial @xmath42 .
* all of them are relaxed to the bloch - type state unless they are initially in the nel - type state .
the model is a square element with @xmath87 @xmath37 @xmath87 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , and @xmath44 .
the simulation on the spin relaxation is carried out by the oommf cg minimizer .
, scaledwidth=48.0% ] * current - induced dynamics . * a skyrmion and an antiskyrmion move rotationally together with a helicity rotation in the frustrated system without the ddi @xcite , as shown in figs . [ fig6](a ) and [ fig6](b ) .
the skyrmion and antiskyrmion motions drastically change once the ddi is introduced .
indeed , the ddi produces the energy difference depending on the helicity as shown in fig .
namely , the helicity rotation costs some energies in the order of @xmath88 j. in fig .
[ fig5 ] , we show the time evolution of the helicity .
the helicity rotates toward that of the bloch - type state ( @xmath89 ) unless the initial helicity is precisely of the nel - type state .
this helicity locking occurs for the small driving current , since the bloch - type skyrmion has the lowest energy in the presence of the ddi .
namely , under the small current injection , the helicity relaxes to that of the bloch - type state , and then the skyrmion moves along a straight line with the helicity fixed , as shown in fig .
[ fig6](c ) and [ fig6](d ) .
this translational motion occurs due to the helicity locking by the ddi , which is highly contrasted to the case without the ddi . however , as shown in fig .
[ fig6](e ) and [ fig6](f ) , the translational motion with the fixed helicity turns out to be the rotational motion together with a helicity rotation once the intensity of the driving current exceeds a certain critical value ( see and ) .
the reason is that the energy injected by the current overcomes the potential difference between the bloch - type and nel - type skyrmions .
note that the current prefers the helicity rotation .
this is the helicity locking - unlocking transition as a function of the intensity of the driving current .
-@xmath13-@xmath14 ferromagnetic thin film .
* trajectories of ( * a * ) skyrmions and ( * b * ) antiskyrmions with different initial helicities in the absence of the ddi .
trajectories induced by ( * c*,*d * ) a small driving current ( @xmath90 a m@xmath91 ) and by ( * e*,*f * ) a large driving current ( @xmath92 a m@xmath91 ) in the presence of the ddi . in the presence of the ddi , the skyrmion motion is along a straight line with the helicity fixed for the small driving current , but it is along a circle together with the helicity rotation for the large driving current .
this is the helicity locking - unlocking transition .
the model is a square element with @xmath93 @xmath37 @xmath93 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , @xmath44 , @xmath48 .
the simulation is carried out by the oommf solver integrating the llg equation including the spin torque .
, scaledwidth=48.0% ] .
the helicity is flipped by a strong @xmath94-ps - long current pulse and becomes @xmath95 ( red curve ) .
such a flip does not occur by applying a small current pulse with the same duration ( blue curve ) .
( * b * ) a similar flip occurs also for an antiskyrmion .
( * c*)-(*f * ) snapshots of the skyrmion during the helicity flip process .
( * g*)-(*j * ) snapshots of the antiskyrmion during the helicity flip process .
the model is a square element with @xmath96 @xmath37 @xmath96 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , @xmath44 , and @xmath48 .
the simulation is carried out by the oommf solver integrating the llg equation including the spin torque . , scaledwidth=48.0% ] * switching process of binary bloch - type skyrmions .
* we have revealed prominent properties of the helicity - orbital coupling in the skyrmion motion in the frustrated magnetic system with the ddi . we may design a binary memory together with a switching process based on these properties .
let us start with a skyrmion at rest .
( i ) it is in one of the two bloch - type states indexed by the helicity @xmath4 ; since the helicity is a conserved quantity , we may use it as a one - bit memory .
( ii ) we can read its value by observing the motion of a skyrmion ; when a small current pulse is applied , a skyrmion begins to move along a straight line toward left or right according to its helicity @xmath97 or @xmath98 [ see fig . [ fig6](c ) and ] ; this is the read - out process of the memory .
( iii ) on the other hand , when we apply a strong current pulse , a skyrmion begins to rotate together with its helicity rotation ; we can tune the pulse period to stop the helicity rotation so that it is relaxed to the flipped value of the initial helicity , as shown in fig .
this is the switching process of the memory .
as shown in , a skyrmion moving rightward is stopped and moves leftward after one rotation under a strong current pulse applied for an appropriate period . .
( * b * ) spontaneous formation of a bi - antiskyrmion by merging two antiskyrmions and different @xmath42 .
the model is a square element with @xmath99 @xmath37 @xmath100 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , @xmath44 , and @xmath48 .
the simulation is carried out by the oommf solver integrating the llg equation .
, scaledwidth=48.0% ] * spontaneous formations of a bi - skyrmion and a bi - antiskyrmion . *
we proceed to investigate the spontaneous formations of a bi - skyrmion and a bi - antiskyrmion , that is , the spontaneous formations of a high - topological - number skyrmion and a high - topological - number antiskyrmion .
we construct one skyrmion with @xmath95 and one skyrmion and @xmath64 , which are placed at the left and right sides of the sample , respectively , with the initial configuration as shown in fig .
the dynamic process of the spontaneous formation of a bi - skyrmion is illustrated in fig .
[ fig8](a ) , where a time - dependent relaxation of the system is simulated .
the two skyrmions first move close to each other , where they both have in - plane spins near each other that are pointing along the @xmath101-direction .
the two skyrmions then smoothly merge into one bi - skyrmion with @xmath102 , when the encountered in - plane spins are rotated into the @xmath16-direction .
finally , a bi - skyrmion with @xmath102 is formed in the sample ( see ) .
a similar spontaneous formation of a bi - antiskyrmion with @xmath73 is possible as given in fig .
[ fig8](b ) ( see ) .
the natural and spontaneous formations of a bi - skyrmion and a bi - antiskyrmion are caused by the skyrmion - skyrmion attraction and the antiskyrmion - antiskyrmion attraction , respectively .
a detailed analysis on the interactions among skyrmions and antiskyrmions is given in .
on the other hand , it is noteworthy that the helicity rotation of the antiskyrmion is caused by the coupling between the translational motion and the helicity , as reported in ref . .
the rotation of the helicity also happens for the skyrmion , although it is not obvious in fig .
[ fig8](a ) as the spacing between the two skyrmions is very short .
* forced separations of a bi - skyrmion and a bi - antiskyrmion . * after showing the spontaneous formations of the bi - skyrmion and the bi - antiskyrmion , we further demonstrate that it is possible to split the bi - skyrmion and the bi - antiskyrmion by applying a driving current .
figure [ fig9 ] demonstrates the forced separation of two skyrmions from a bi - skyrmion .
we first place a relaxed bi - skyrmion at the center of the sample .
then , a driving current of @xmath103 a m@xmath91 is vertically injected to the sample , which can be realized by the spin hall effect in a heavy - metal substrate .
it is found that the bi - skyrmion starts to rotate counterclockwise , and forms a clear peanut - like shape , as shown in fig .
[ fig9](f ) .
when it has rotated almost @xmath104 degrees , two skyrmions are almost generated , as shown in fig .
[ fig9](k ) . as shown in fig .
[ fig9](l ) , the bi - skyrmion is successfully split into two isolated skyrmions ( see ) . this forced separation is possible since the bi - skyrmion is composed of two skyrmions with opposite helicities , which rotate in opposite directions .
the separation process of a bi - antiskyrmion forced by the driving current is similar to that of the bi - skyrmion ( see and ) .
we note that the total skyrmion number conserves during both the spontaneous formation and the forced separation of the bi - skyrmion and the bi - antiskyrmion .
furthermore we found that the bound state with a higher skyrmion number @xmath105 can also be spontaneously generated ( see ) . *
pair annihilation of a skyrmion and an antiskyrmion .
* finally , we demonstrate that the pair annihilation process of a skyrmion and an antiskyrmion in a forced collision event driven by the current , as shown in fig .
[ fig10 ] .
we first place a relaxed skyrmion and a relaxed antiskyrmion at the center of the sample .
there is a certain spacing between the skyrmion and the antiskyrmion .
then , a @xmath106-ps - long driving current pulse of @xmath103 a m@xmath91 is vertically injected to the sample .
the skyrmion and the antiskyrmion move in a circular trajectory toward opposite directions .
the helicity @xmath42 of the skyrmion and the antiskyrmion varies with the position as it is coupled with the orbital motion ( see fig .
[ fig6 ] and ref . ) . as shown in fig .
[ fig10](b ) , the skyrmion meets the antiskyrmion , where the skyrmion has the helicity of @xmath95 while the antiskyrmion has the helicity of @xmath64 , which is the condition for the pair annihilation .
the skyrmion and the antiskyrmion then form a magnetic bubble state with @xmath107 , as shown in fig .
[ fig10](d ) .
the magnetic bubble state , which is not topologically protected , rapidly shrinks in size and is finally evolved into the trivial ground state , as shown in fig .
[ fig10](l ) .
this is the exact process of the pair annihilation of a skyrmion and an antiskyrmion ( see ) .
a forced collision between a skyrmion and an antiskyrmion will lead to the pair annihilation of the skyrmion and the antiskyrmion as long as they do not have identical @xmath42 ( see ) .
it is also worth mentioning that a propagating spin wave is emitted during the pair annihilation , which can be discerned from fig .
[ fig10 ] . .
* ( a ) we first place a relaxed bi - skyrmion .
( b ) it starts to rotate counterclockwise under an injected driving current . after forming a clear peanut - like shape ,
it is separated into two skyrmions as in ( l ) .
the model is a square element with @xmath36 @xmath37 @xmath36 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , @xmath44 , @xmath48 , and @xmath103 a m@xmath91 .
@xmath32 is indicated in ( a ) and ( l ) .
the simulation is carried out by the oommf solver integrating the llg equation including the spin torque . , scaledwidth=48.0% ] ) .
then , by emitting spin waves , it disappears . the spin wave can be discerned from ( * g*)-(*k * ) .
the positive and negative skyrmion number densities are denoted by red and blue colors , respectively .
@xmath32 is indicated in ( * a * ) and ( * d * ) .
the maximum value of the @xmath32 density is indicated in the lower right corner of each snapshot .
the model is a square element with @xmath108 @xmath37 @xmath108 spins .
the parameters are @xmath38 , @xmath39 , @xmath43 , @xmath44 , @xmath48 , and @xmath103 a m@xmath91 . , scaledwidth=48.0% ]
we have studied magnetic skyrmions in a 2d frustrated magnetic system with competing exchange interactions based on the @xmath0-@xmath1-@xmath2 classical heisenberg model on a simple square lattice .
we have demonstrated that the contribution of the ddi energy is significant although the scale of a skyrmion is of the order of nanometer . as a result , the bloch - type skyrmion has the lowest energy and is stabilized .
it has a bistable structure indexed by the helicity @xmath4 .
the role of the helicity - orbital coupling becomes remarkable in the presence of the ddi .
first , a skyrmion moves along a straight line with a fixed helicity under a week driving current .
the moving direction of a skyrmion is opposite when its helicity is opposite .
second , the dynamics of a skyrmion drastically changes from the translational motion to the circular motion as the increase of the driving current , which is the locking - unlocking transition of the helicity .
third , we can flip the helicity of a skyrmion by applying a strong current pulse with a reasonable period .
we have argued that these properties could be used to design a binary memory together with read - out process as well as the switching process .
furthermore , we have shown that magnetic skyrmions with @xmath109 are possible metastable topological non - trivial states in the frustrated magnetic system .
the skyrmion with a high topological number @xmath110 could be seen as a bound state or a cluster of skyrmions . in particular , we have demonstrated the spontaneous formation and the forced separation of a bi - skyrmion .
in addition , we have shown the pair annihilation of a skyrmion and an antiskyrmion by triggering a collision event .
our results reveal the exotic and promising properties of skyrmions and antiskyrmions in the frustrated magnetic system , which may lead to novel spintronic and topological applications based on the manipulation of skyrmions and antiskyrmions .
* symmetry analysis . *
a skyrmion is indexed by the skyrmion number @xmath32 and the helicity @xmath42 .
we study how the helicity is fixed by minimizing the hamiltonian .
first , we examine the hamiltonian equation ( [ eq : hamiltonian ] ) , which is rewritten as @xmath111 in the momentum space , with @xmath112 the hamiltonian has the following three symmetries : it is convenient to define the vorticity number @xmath113 by the winding number of the in - plane spin texture , namely @xmath114_{\varphi=0}^{\varphi=2\pi}. \label{eq : q_v_phi}\ ] ] we consider the transformation @xmath115 and @xmath116 , under which the skyrmion number is inverted , @xmath117 . in this symmetry operation , the spins are mapped to @xmath118 the hamiltonian is invariant with the above transformation . as a result ,
a skyrmion with @xmath119 and @xmath42 and an antiskyrmion with @xmath120 and @xmath121 have the same energy .
moreover , we can show that the energy is independent of @xmath42 analytically .
the taylor expansion of the hamiltonian equation ( [ hamilq ] ) reads @xmath122 we chose the values of @xmath123 and @xmath14 , which satisfy @xmath124 and @xmath125 .
when @xmath126 , the skyrmion is not stabilized since @xmath127 terms do not play an important role .
the condition @xmath125 is necessary to cut the expansion with the order of @xmath127 .
this is independent of @xmath42 , which implies that the energy is independent of the helicity .
furthermore , the hamiltonian is invariant for @xmath130 , which implies the energy between the skyrmion and the antiskyrmion is identical irrespective to the helicity @xmath42 .
f. buttner , c. moutafis , m. schneider , b. kruger , c. m. gunther , j. geilhufe , c. v. k. schmising , j. mohanty , b. pfau , s. schaffert , a. bisig , m. foerster , t. schulz , c. a. f. vaz , j. h. franken , h. j. m. swagten , m. klaui , and s. eisebitt , nat .
phys . * 11 * , 225 ( 2015 ) .
s. woo , k. litzius , b. kruger , m .- y .
i m , l. caretta , k. richter , m. mann , a. krone , r. m. reeve , m. weigand , p. agrawal , i. lemesh , m .- a .
mawass , p. fischer , m. klaui , and g. s. d. beach , nat . mater . *
15 * , 501 ( 2016 ) .
i. kezsmarki , s. bordacs , p. milde , e. neuber , l. m. eng , j. s. white , h. m. ronnow , c. d. dewhurst , m. mochizuki , k. yanai , h. nakamura , d. ehlers , v. tsurkan , and a. loidl , nat .
* 14 * , 1116 ( 2015 ) .
* acknowledgments * x.z . was supported by jsps ronpaku ( dissertation ph.d . ) program .
is grateful to w. jiang for many useful comments .
also thanks s .- z .
lin for discussions on frustrated skyrmions .
y.z . acknowledges the support by the national natural science foundation of china ( project no .
11574137 ) and shenzhen fundamental research fund under grant no .
m.e . acknowledges the support by the grants - in - aid for scientific research from jsps kakenhi ( grant nos .
25400317 and 15h05854 ) , and also the support by crest , jst .
is very much grateful to n. nagaosa for many helpful discussions on the subject . * author contributions * m.e . and y.z .
conceived the idea .
y.z . , x.l . , and m.e .
coordinated the project .
x.z . performed the numerical simulation and data analysis .
j.x . developed the modules for nnn and nnnn exchange interactions .
m.e . carried out the theoretical analysis .
all authors interpreted the results and prepared the manuscript . | an intriguing feature of the magnetic skyrmion in a frustrated magnetic system is its helicity - orbital coupling .
when the magnetic dipole - dipole interaction ( ddi ) is neglected , a skyrmion can show a current - induced rotational motion together with a helicity rotation since the energy is independent of the helicity . here
, we explore the skyrmion dynamics in a frustrated magnetic system based on the @xmath0-@xmath1-@xmath2 classical heisenberg model explicitly by including the ddi .
the skyrmion energy acquires a helicity dependence due to the ddi , resulting in a current - induced translational motion with a fixed helicity .
the lowest energy states are the degenerate bloch - type states with two distinguishable helicities , which could be used as information carriers in the binary memory .
the skyrmion helicity can be identified by applying a small driving current as the direction of the translational motion is dependent of the helicity .
it is found that by increasing the intensity of the driving current , the helicity locking - unlocking transition occurs , where the translational skyrmion motion with a fixed helicity changes to a rotational skyrmion motion together with a helicity rotation .
this new aspect of the helicity - orbital coupling allows us to realize the flip of the helicity between two stable bloch - type states by applying a strong current pulse , which could act as a switching process in the binary memory .
furthermore , we demonstrate that two skyrmions or two antiskyrmions can spontaneously form a bound state .
we also show the separation of the bound state forced by a driving current .
in addition , we show a pair annihilation of a skyrmion and an antiskyrmion accompanied with the emission of a propagating spin wave .
our results indicate the distinctive nature of skyrmions in frustrated magnetic materials , and also reveal unexpected properties of skyrmions which may lead to viable applications of topological spintronics .
he exotic and versatile magnetic skyrmion is a high - profile topological object in magnetic materials , which promises novel applications in spintronic devices @xcite .
the magnetic skyrmion was first experimentally identified in chiral magnetic materials in 2009 @xcite .
after that , it has been experimentally observed , created , and modified in different systems , including magnetic materials @xcite , semiconductors @xcite , multiferroic materials @xcite , and ferroelectric materials @xcite . due to the properties of the topologically protected stability as well as the efficient mobility driven by external forces
, the magnetic skyrmion is anticipated to be employed as the information carrier in future data storage devices @xcite , logic computing devices @xcite , microwave devices @xcite , spin - wave devices @xcite , and transistor - like devices @xcite .
very recently , it is shown that the magnetic skyrmion in the frustrated magnetic system has remarkable physical properties that renovate our understanding of the chiral magnetic object @xcite .
skyrmions in frustrated magnetic materials are stabilized by the quartic differential term , which is a reminiscence of the original mechanism of the dynamical stabilization proposed by skyrme @xcite .
a prominent property is that a skyrmion and an antiskyrmion have the same energy irrespective to the helicity . for instances , the high - topological - number magnetic skyrmion @xcite and the antiskyrmion @xcite with the topological number of @xmath3
are stable / metastable in the anisotropic frustrated magnet @xcite , where the magnetic skyrmion shows coupled dynamics of the helicity and the center of mass @xcite
. it is found that the translational motion of a magnetic skyrmion is coupled with its helicity in the frustrated magnet , resulting in the rotational skyrmion motion @xcite .
these novel properties due to the helicity - orbital coupling are lacking in the conventional ferromagnetic system , where the magnetic skyrmion is stabilized by the dzyaloshinskii - moriya interaction ( dmi ) and the helicity is locked . in this paper , we explore skyrmions and antiskyrmions in a two - dimensional ( 2d ) frustrated ferromagnetic system with competing exchange interactions based on the @xmath0-@xmath1-@xmath2 classical heisenberg model on a simple square lattice @xcite .
we explicitly include the dipole - dipole interaction ( ddi ) , which is neglected in the previous literature @xcite .
we first study the relaxed spin structure and the energy of a skyrmion with different initial values of the topological number and the helicity .
the energies of a skyrmion with different helicity are degenerate in the absence of the ddi , but the degeneracy is resolved by the ddi . a skyrmion has two degenerate lowest energy states , which are the bloch - type states with the helicities @xmath4 .
namely , a bloch - type skyrmion has an internal degree of freedom taking a binary value of @xmath4 .
these two bloch - type skyrmions are clearly distinguishable since they move straight in opposite directions under a weak driving current . on the other hand , when a strong driving current is applied , the helicity is no longer locked , and the skyrmion performs a rotational motion together with a helicity rotation as if the ddi were absent .
we point out that it is possible to design a binary memory with the use of the binary helicity state of the bloch - type skyrmion : ( i ) apply a weak current pulse to a skyrmion at rest , and observe the direction of the motion ( toward right or left ) to which a skyrmion begins to move ; it determines the binary value of the skyrmion helicity : ( ii ) apply a strong current pulse ; it switches the binary value of the skyrmion helicity .
furthermore , we demonstrate a spontaneous formation of the two - skyrmion bound state ( the bi - skyrmion ) as well as the two - antiskyrmion bound state ( the bi - antiskyrmion ) due to the attraction between two skyrmions and two antiskyrmions , respectively .
indeed , it is possible to split a bi - skyrmion as well as a bi - antiskyrmion by applying a driving current .
the reason is that the direction of the skyrmion motion depends on its helicity .
in addition , we demonstrate a pair annihilation of a skyrmion and an antiskyrmion , which emits a propagating spin wave upon the annihilation event .
our results indicate that there are more promising properties and degrees of freedom of skyrmions and antiskyrmions in the frustrated magnetic system , which have great potential to be used in future topological and spintronic applications . |
skew schur polynomials @xmath0 play an important and multi - faceted role in algebraic combinatorics .
for example , they are generating functions for the littlewood - richardson coefficients @xmath1 @xmath2 they are also generating functions for young tableaux of shape @xmath3 @xmath4 where the sum is over all young tableaux @xmath5 of shape @xmath3 and @xmath6 is a monomial associated to @xmath5 .
the relationship between these two very different facets of skew schur polynomials involves the combinatorics of young tableaux ( * ? ? ?
* chapter 7 , appendix 1 ) .
young tableaux can be thought of as increasing labeled chains in young s lattice , with covers labeled by pairs @xmath7 , where @xmath8 is the entry in the box corresponding to the cover , and @xmath9 the content of that box ( that is , the difference between its column and row ) ; labels are ordered lexicographically .
we introduce skew schubert polynomials , which are the schubert polynomial analogs of skew schur polynomials , in the sense that they generalize the two properties mentioned above . in section 1
, we introduce our key concepts of increasing labeled chains in the bruhat order , which generalize young tableaux , and of skew schubert polynomials , @xmath10 for permutations @xmath11 .
in particular , we show that skew schubert polynomials are generating functions both for littlewood - richardson coefficients @xmath12 for schubert polynomials ( the analog of ) , and for increasing chains in the bruhat order ( the analog of ) . our skew schubert polynomials are different from those defined in @xcite ( using skew divided difference operators ) , which do not expand with nonnegative coefficients in the basis of schubert polynomials , in general .
the skew schubert polynomial @xmath13 is the ordinary schubert polynomial @xmath14 , and our formula for @xmath13 in terms of increasing chains is the formula of corollary 5.3 in @xcite , which inspired us ( see the next section for the notation ) . in section 2
, we give a purely combinatorial proof of that formula , relating it to a standard construction @xcite in terms of rc - graphs by giving a content - preserving bijection between rc - graphs and increasing chains .
let @xmath15 be the symmetric group of permutations of @xmath16 . for @xmath17 , the _ length _
, @xmath18 , of @xmath19 is the number of inversions of @xmath19 .
set @xmath20 to be the longest permutation , @xmath21 .
we use basics on schubert and schur polynomials , which may be found in any of @xcite .
the main outstanding problem in the theory of schubert polynomials is the littlewood - richardson problem ( * ? ? ?
* problem 11 ) : determine the structure constants @xmath12 defined by the polynomial identity @xmath22 since every schur polynomial is a schubert polynomial , this problems asks for the analog of the classical littlewood - richardson rule .
the classical littlewood - richardson coefficients @xmath1 have important and intricate combinatorial properties ( * ? ? ?
* chapter 7 , appendix 1 ) .
the littlewood - richardson coefficients @xmath12 for schubert polynomials should be similarly important .
indeed , they are intersection numbers of schubert varieties ; more precisely , @xmath12 enumerates flags in a suitable triple intersection of schubert varieties indexed by @xmath23 , @xmath24 , and @xmath25 .
the cohomology classes @xmath26 ( called schubert classes ) for @xmath17 of schubert varieties form an integral basis for the cohomology ring @xmath27 of the flag manifold . many identities and some formulas for the constants
@xmath12 have been obtained by studying the class @xmath28 in @xmath27 @xcite .
this is because , for @xmath29 , we have the following identity in the cohomology ring @xmath30 and hence , by the duality of the intersection pairing @xmath31 with this motivation , we would like to define the skew schubert polynomial @xmath10 for permutations @xmath11 in @xmath15 to be _ the _ polynomial representative of the class @xmath28 . unfortunately , the cohomology of the flag manifold is isomorphic to the quotient ring @xmath32/ \langle e_i(x_1,\ldots , x_n)\mid i=1,\ldots , n\rangle\,,\ ] ] and so there is no well - defined polynomial representative of @xmath28 . here
@xmath33 is the @xmath34th elementary symmetric polynomial in @xmath35 .
if however we choose the degree reverse lexicographic term order on monomials in the polynomial ring @xmath36 $ ] , with @xmath37 , then every element of this quotient ring has a unique normal form polynomial representative ( * ? ? ?
1.1 ) with respect to this term order .
these normal form representatives are elements of @xmath38 equivalently , if we set @xmath39 , then normal form representatives of cohomology classes are sums of monomials dividing @xmath40 .
fomin , gelfand , and postnikov @xcite observed that for @xmath17 , the schubert polynomial @xmath14 is the normal form representative of the corresponding schubert class @xmath26 .
in fact , this feature of schubert polynomials , that one does not need to work modulo an ideal , is what led lascoux and schtzenberger to their definition of schubert polynomial .
thus we define the _ skew schubert polynomial _ @xmath10 to be the normal form representative of the class @xmath28 .
these skew schubert polynomials are generating functions for the coefficients @xmath12 .
[ theoremone ] for @xmath11 in @xmath15 , we have @xmath41 each side of is the normal form representative of the corresponding side of .
while @xmath10 depends upon @xmath42 , the laurent polynomial @xmath43 does not , as @xmath44 and @xmath45 are each independent of @xmath42 .
the independence of @xmath44 is the familiar stability property of schubert polynomials .
the independence of @xmath45 may be similarly deduced using divided differences .
this also follows from proposition [ prop : schub - constr ] , which we prove in section 2 .
the skew schubert polynomial @xmath10 is also a generating function for certain chains from @xmath23 to @xmath19 in the bruhat order .
the bruhat order is defined by its covers ; @xmath46 if and only if @xmath47 is a transposition @xmath7 with @xmath48 , and for every @xmath49 we have either @xmath50 or @xmath51 .
thus @xmath52 with @xmath53 if @xmath48 and at no position between @xmath8 and @xmath9 does @xmath23 take a value between @xmath54 and @xmath55 .
this implies that the difference in lengths @xmath56 equals 1 .
the _ labeled bruhat order _ is the labeled rseau ( labeled directed multigraph ) obtained from the bruhat order by drawing a directed edge @xmath57 whenever @xmath46 with @xmath58 , where @xmath59 and @xmath60 .
we note that all the constructions and statements still hold if we uniformly set @xmath61 , but it is more convenient to use the first definition .
there will be @xmath62 such directed edges for every cover .
this structure was defined in @xcite and has been crucial in subsequent work on the problem of multiplying schubert polynomials by bergeron and sottile . in @xcite this structure
was called the colored bruhat order . to any ( saturated )
chain @xmath63 in this rseau , we associate a monomial @xmath64 in the variables @xmath65 , where the power of @xmath66 counts how often @xmath34 was the first coordinate of a label in @xmath63 .
a chain @xmath67 in this rseau is _ increasing _ if its sequence of labels is increasing in the lexicographic order on pairs of integers .
these definitions allow a nice reformulation of a combinatorial construction of schubert polynomials given in @xcite .
[ thm : skew ] let @xmath11 be permutations in @xmath15
. then @xmath68 the sum over all increasing chains @xmath63 in the labeled bruhat order from @xmath23 to @xmath19 .
note that @xmath69 .
we recover the formula of bergeron and sottile for @xmath14 @xcite .
denote by @xmath70 the set of all increasing labeled chains from @xmath19 to @xmath71 .
[ prop : schub - constr ] let @xmath72
. then @xmath73 theorem [ thm : skew ] has an immediate enumerative consequence .
the _ type _ @xmath74 of a chain @xmath63 in the labeled bruhat order is the ( weak ) composition @xmath74 whose @xmath34th component counts the number of occurrences of @xmath34 as the first coordinate of an edge label .
thus @xmath75 . for @xmath11 in the bruhat order and a composition @xmath74 of @xmath56 , let @xmath76 count the number of increasing chains from @xmath23 to @xmath19 of type @xmath74 . since @xmath77 , equating coefficients in and using theorem [ thm : skew ] we obtain [ cor : identity ] @xmath78 .
for a composition @xmath74 with @xmath79 parts where @xmath80 coordinatewise , set @xmath81 where @xmath82 is the complete homogeneous symmetric polynomial of degree @xmath83 in the variables @xmath84 .
we let @xmath85 and @xmath86 denote the corresponding cohomology classes in @xmath27 . for @xmath87 ,
set @xmath88 to be the coefficient of @xmath89 in the product @xmath90 .
this gives a linear map @xmath91 .
[ lem : linear ] for any @xmath92 , @xmath88 is the coefficient of the monomial @xmath93 in the normal form representative of @xmath94 .
this follows from two observations .
first , both @xmath95 and the map associating to @xmath94 the coefficient of the monomial @xmath93 in the normal form representative of @xmath94 are @xmath96-linear maps on @xmath27 .
second , the lemma holds on the basis of schubert classes , whose representatives are schubert polynomials .
this was shown by kirillov and maeno @xcite and is also a consequence of the pieri formula @xcite and the construction of schubert polynomials given in proposition [ prop : schub - constr ] .
given a chain @xmath63 in the bruhat order , let end@xmath97 denote its end point .
the pieri - type formula for schubert classes @xcite is @xmath98 the sum over all increasing chains that begin at @xmath23 , have length @xmath83 , and whose labels have first coordinate @xmath8 .
applying lemma [ lem : linear ] to the class @xmath28 shows that the coefficient of the monomial @xmath99 in the normal form representative of @xmath100 is the coefficient of @xmath89 in the triple product @xmath101 , where @xmath74 is the type of the chain @xmath63 .
since the coefficient of @xmath89 in a product @xmath102 is the kronecker delta @xmath103 ( by the duality of the intersection pairing ) , this is the coefficient of @xmath26 in the product @xmath104 .
the theorem now follows by expanding this product and iteratively applying the pieri formula . by the definition of skew schubert polynomials , we have @xmath105 since @xmath106 and @xmath107 .
thus there should exist a natural bijection between @xmath108 and @xmath109 that preserves type .
corollary [ cor : identity ] suggests that , for @xmath11 , this bijection should generalize to a map from @xmath110 to the union @xmath111 such that the cardinality of the inverse image of each chain in @xmath112 is @xmath113 .
this would give a combinatorial interpretation for @xmath113 , and thus solve the littlewood - richardson problem . the skew schubert polynomial @xmath10 also has a geometric interpretation as the representative in cohomology of a skew schubert variety @xmath114 where @xmath115 is the schubert variety whose representative in cohomology is @xmath14 and @xmath116 and @xmath117 are flags in general position .
stanley noted that skew schur polynomials have a similar geometric - cohomological interpretation @xcite .
schubert polynomials @xmath14 when @xmath19 has a single descent at position @xmath8 ( @xmath19 is a grassmannian permutation ) are schur symmetric polynomials in @xmath118 . even if @xmath19 and @xmath23 are grassmannian permutations with the same descent , then it is not necessarily the case that the skew schubert polynomial @xmath10 equals the corresponding skew schur polynomial .
we illustrate this in @xmath119 , using partitions and skew partitions for indices of schur polynomials .
we have the schubert / schur polynomials @xmath120 and also the skew schur polynomial @xmath121 on the other hand , @xmath122 and @xmath123 . by monk s formula ( or the pieri formula ) , we compute @xmath124 to be @xmath125 the first two terms of @xmath124 carry the same information as the two terms of @xmath126 , but there is also a third term . this is because the multiplication of schubert polynomials is richer than the multiplication of schur polynomials . by theorem [ theoremone ] , this expansion of @xmath124 records the fact that @xmath127 is a summand ( with coefficient 1 ) in each of @xmath128 note that the last product ( unlike the first two ) is not symmetric in @xmath129 and @xmath130 even though both @xmath131 and @xmath127 are symmetric in @xmath129 and @xmath130 .
rc - graphs are combinatorial objects associated to permutations @xmath19 in the symmetric group @xmath15 .
they were defined in @xcite as certain subsets of @xmath132 we may linearly order such a subset of pairs by setting @xmath133 let @xmath134 be the @xmath34th pair in this linear order . consider the sequence @xmath135 an rc - graph @xmath136 is _ associated _ to a permutation @xmath17 if @xmath137 is a reduced decomposition of @xmath19 .
we denote the collection of rc - graphs associated to a permutation @xmath19 by @xmath138 , and the permutation corresponding to a given rc - graph @xmath136 by @xmath139 .
we represent an rc - graph @xmath136 as a pseudoline diagram recording the history of the inversions of @xmath19 . to this end , draw @xmath42 pseudolines going up and to the right such that the @xmath34th pseudoline begins at position @xmath140 and ends at position @xmath141 .
( the positions are numbered as in a matrix , as illustrated in example [ exrc ] below . )
the rule for constructing the pseudoline diagram is the following : two pseudolines entering at position @xmath142 cross at that position if @xmath142 is in @xmath136 , and otherwise avoid each other at that position .
note that two pseudolines cross at most once , as @xmath137 is reduced .
conversely , a pseudoline diagram represents a permutation @xmath19 , where @xmath143 is the endpoint of the pseudoline beginning at position @xmath140 .
given a pseudoline diagram where no two pseudolines cross more than once , the crossings give an rc - graph associated to the permutation represented by the pseudoline diagram .
given an rc - graph @xmath136 , define the monomial @xmath146 as shown in @xcite , the schubert polynomial @xmath147 indexed by @xmath19 can be expressed as @xmath148 comparing this expression with proposition [ prop : schub - constr ] suggests that there should be a natural bijection between @xmath138 and @xmath70 , where @xmath149 , when @xmath136 corresponds to @xmath63 .
indeed , we give such a bijection . given an rc - graph @xmath136 in @xmath138 , we greedily construct the sequence of rc - graphs @xmath150 as follows .
given @xmath151 , add the pair @xmath152 to @xmath151 where @xmath152 is the pair not in @xmath151 that is minimal in the lexicographic order on pairs .
we have the following lemma . 1 .
every subset @xmath151 is an rc - graph .
2 . @xmath154 is a saturated chain in bruhat order .
3 . if @xmath155 is obtained from @xmath151 by adding the pair @xmath152 , then in the labeled bruhat order we have the labeled cover @xmath156 4 .
the labeled chain with the labels of @xmath157 is increasing .
we prove the first statement by induction .
suppose that @xmath151 is an rc - graph and we add the pair @xmath152 to obtain @xmath155 , where @xmath152 is the minimal pair not in @xmath151 .
consider the pseudoline diagram of @xmath151 near the position @xmath152 : @xmath159 by construction , @xmath151 contains every crossing in the shaded region , so the two pseudolines at this position never cross , with one connecting @xmath8 to @xmath160 as drawn . adding the crossing gives a new pseudoline diagram for a permutation with exactly one more inversion .
indeed , assume that we create a double crossing between one of the two pseudolines obtained from the drawn ones by adding the extra crossing and another pseudoline .
it is easy to check graphically that this can only happen if the latter pseudoline meets the pseudoline connecting @xmath8 to @xmath160 in @xmath151 twice .
but this is impossible , as @xmath151 is an rc - graph .
these arguments show that @xmath161 , implying the second statement . for the third , note
that @xmath152 is a possible label , and the fourth follows by our greedy algorithm for adding pairs not in @xmath136 .
observe that the increasing chain @xmath162 constructed in lemma [ l : rc - chain ] from an rc - graph @xmath136 has the property that @xmath163 _ a priori _ , given that @xmath164 is a cover , we are only guaranteed that @xmath165 with @xmath166 .
we prove that this property holds for any increasing chain ending in @xmath71 , and also that the map @xmath167 is a bijection from @xmath138 to @xmath70 .
this will immediately imply proposition [ prop : schub - constr ] .
we prove these statements by downward induction on @xmath18 .
they hold trivially when @xmath169 .
consider the first two steps in @xmath63 : @xmath170 since @xmath63 is increasing , we have @xmath171 or @xmath172 and @xmath173 .
since the part of the chain beginning with @xmath24 is an increasing chain , our inductive hypothesis implies that subsequent covers involve only positions of @xmath24 greater than or equal to @xmath8 . thus if @xmath174 , we must have @xmath175 and @xmath176 .
now suppose that @xmath181 and @xmath188 so that @xmath172 and we then have @xmath173 .
then @xmath189 and so the betweenness condition on covers @xmath190 is violated .
this completes the proof of statement 1 . for the second statement , consider the procedure of successively removing labels @xmath152 of covers of @xmath63 from the rc - graph of @xmath71 ( which contains all crossings ) .
suppose that this process applied to the chain above @xmath24 has created an rc - graph @xmath136 associated to @xmath24 .
we argue that removing the crossing @xmath191 from @xmath136 creates an rc - graph associated to @xmath19 .
first , @xmath191 is a crossing of @xmath136 , as we have only removed crossings that exceed @xmath191 in the lexicographic order . consider the pseudoline diagram of @xmath136 near @xmath191 , which we represent on the left . @xmath192 note that @xmath136 contains all crossings in the shaded region .
removing the crossing at @xmath191 creates the picture on the right . arguing as in the proof of lemma [ l : rc - chain ] ( but in reverse )
shows that we obtain an rc - graph associated to @xmath19 .
property of increasing chains from @xmath19 to @xmath71 implies that the branching in the tree of increasing chains from @xmath19 to @xmath71 is quite simple ; it only depends upon the permutation at a node and not on the history of the chain .
the branches at a permutation @xmath23 consists of all possible covers @xmath193 , where @xmath8 is the minimal position @xmath34 such that @xmath194 and @xmath195 and @xmath9 is any position greater than @xmath8 such that @xmath196 and if @xmath49 , then @xmath197 is not between @xmath54 and @xmath55 .
we may order the branches at @xmath23 by the position @xmath9 .
this leads to an efficient lexicographic search of this tree for generating all such increasing chains , and thus all rc - graphs in @xmath138 as well as the multiset of monomials in the schubert polynomial @xmath198 .
the computational complexity and memory usage of this algorithm depend only upon @xmath42 ( the number of entries in @xmath19 ) , the length @xmath199 of @xmath19 , and the number of increasing chains , @xmath200 .
this is true , in fact , for any algorithm generating @xmath201 combinatorial objects corresponding to the monomials in the schubert polynomial @xmath198 .
furthermore , it is not hard to see that the above algorithm is of order @xmath202 , where @xmath203 .
indeed , note that the number of internal nodes of a rooted tree with a fixed number of leaves ( @xmath201 in our case ) and a fixed distance from its root to all its leaves ( @xmath9 in our case ) is maximized by the tree with all internal nodes having only one descendant ; thus , there are @xmath204 nodes . on the other hand ,
finding the descendants of a given node is of order @xmath205 .
let us also note that a closely related algorithm , with the same complexity as the one above , appears in @xcite ; it is described in terms of removing crossings from the rc - graph corresponding to @xmath71 .
more precisely , the authors of the mentioned papers use combinatorial versions of the divided difference operators in order to recursively generate the rc - graphs corresponding to a permutation @xmath19 .
one starts from @xmath71 and applies these operators using any chain in the weak order on @xmath15 from @xmath71 to @xmath19 . a certain choice of chain guarantees that each rc - graph in the obtained tree has at least one descendant upon applying the corresponding operator .
it turns out that the rc - graphs on a given level @xmath34 in this tree correspond ( via the bijection described above ) to the subchains starting at level @xmath206 in the tree constructed by the previous algorithm .
while this algorithm for generating the multiset of monomials in the schubert polynomial @xmath207 also computes this polynomial , a. buch has pointed out to us that the transition formula of lascoux and schtzenberger @xcite provides a more efficient algorithm to _ compute _ @xmath207 .
we conclude by mentioning the connection with the insertion algorithm for rc - graphs in @xcite , which bijectively proves the pieri formula for multiplying a schubert polynomial by a complete homogeneous symmetric polynomial @xmath208 . given an rc - graph @xmath136 with @xmath209 ,
the chain @xmath162 can be obtained by successive insertions into @xmath136 corresponding to multiplications of @xmath210 by the monomials @xmath211 in @xmath212 , for @xmath213 . in this this special case
, the insertion algorithm reduces to the older one in @xcite .
sergey fomin and anatol n. kirillov , _ the yang - baxter equation , symmetric functions , and schubert polynomials _ , proceedings of the 5th conference on formal power series and algebraic combinatorics ( florence , 1993 ) , vol . 153 , 1996 , pp .
123143 .
anatol n. kirillov and toshiaki maeno , _ quantum double schubert polynomials , quantum schubert polynomials and vafa - intriligator formula _ , discrete math . * 217 * ( 2000 ) , no . 1 - 3 , 191223
, formal power series and algebraic combinatorics ( vienna , 1997 ) .
richard p. stanley , _ some combinatorial aspects of the schubert calculus _ , combinatoire et reprsentation du groupe symtrique ( actes table ronde cnrs , univ .
louis - pasteur strasbourg , strasbourg , 1976 ) , springer , berlin , 1977 , pp .
lecture notes in math .
, vol . 579 . | we define skew schubert polynomials to be normal form ( polynomial ) representatives of certain classes in the cohomology of a flag manifold .
we show that this definition extends a recent construction of schubert polynomials due to bergeron and sottile in terms of certain increasing labeled chains in bruhat order of the symmetric group .
these skew schubert polynomials expand in the basis of schubert polynomials with nonnegative integer coefficients that are precisely the structure constants of the cohomology of the complex flag variety with respect to its basis of schubert classes .
we rederive the construction of bergeron and sottile in a purely combinatorial way , relating it to the construction of schubert polynomials in terms of rc - graphs . |
we thank prof . l. huang for helpful discussions . this work was supported by aro under grant w911nf-14 - 1 - 0504 , and the nsf of china under grants nos .
11575072 , 11135001 and 11275003 .
zgh and ycl devised the research project .
jqz , zgh , and rs performed numerical simulations .
jqz , zgh , ycl , and zxw analyzed the results and wrote the paper .
* competing financial interests * : the authors declare no competing financial interests . as a function of the pinning fraction @xmath46 for scale - free networks of different connection densities .
* the average degree of the networks for simulation are @xmath214 , @xmath215 , @xmath216 , to @xmath62 in ( a - d ) , respectively , and the value of the pinning pattern indicator @xmath50 ranges from @xmath217 to @xmath64 for each panel . the results are averaged over @xmath218 realizations for scale - free networks of size @xmath58 and degree exponent @xmath59 . ] as a function of the pinning fraction @xmath46 for scale - free networks of varying degrees of heterogeneity .
* the scaling exponents of the networks are @xmath124 , @xmath219 , @xmath220 , and @xmath221 in ( a - d ) , respectively , and the value of the pinning pattern indicator @xmath50 ranges from @xmath217 to @xmath64 for each panel .
the results are averaged over @xmath218 realizations for scale - free networks of size @xmath222 and average degree @xmath223 . ]
is obtained from the transition matrix eq .
( [ eq : t20pin2 ] ) for @xmath58 .
the value of the pinning pattern indicator @xmath50 is set as @xmath63 , @xmath224 , @xmath163 and @xmath64 in ( a - d ) , respectively , and the pinning fraction @xmath46 ranges from @xmath225 to @xmath163 . ] in comparison with the simulation results . *
the system has size @xmath58 and power - law degree distribution @xmath120 with scaling exponent @xmath111 .
the theoretical prediction does not depend on the value of the average degree . in direct simulations ,
the values of the average degree are @xmath214 , @xmath215 , @xmath216 , and @xmath62 . in each case
, the simulation result is averaged over @xmath218 network realizations .
the value of the pinning pattern indicator is set to @xmath226 for systems with different degree scaling exponents . *
the system has size @xmath58 and power - law degree distribution @xmath120 with different values of the degree exponent : ( a - d ) @xmath124 , @xmath219 , @xmath220 , @xmath221 , respectively . in each case
, the value of the pinning pattern indicator @xmath50 ranges from @xmath217 to @xmath64 . ] and @xmath150 denote nonzero optimal pinning fraction @xmath140 given by eq .
( [ eq : optimal_rho ] ) .
the scale - free networks have the degree exponents @xmath124 , @xmath219 , @xmath220 , and @xmath221 , respectively . the response function is for @xmath125 ( corresponding to @xmath20 ) .
( b ) contour map of @xmath71 in the parameter space of @xmath159 and @xmath160 for scale - free networks with @xmath111 . in the lower - left region below the boundary @xmath161 ( white dashed line )
, nonzero solution of @xmath71 can not be obtained .
( c ) optimal pinning fraction @xmath140 as a function of @xmath162 for scale - free networks .
the analytical results from eq .
( [ eq : optimal_rho ] ) ( red solid curve ) and the simulation results ( black open squares ) agree well with each other .
the red arrow marks the theoretical prediction of the boundary , where nonzero @xmath71 solutions exist on the left side .
( d ) for er random networks , @xmath140 as a function of @xmath162 . theoretical results from eq . (
[ eq : optimal_rho ] ) ( red open circle ) and simulation results ( black open squares ) are shown .
the boundaries @xmath227 and @xmath228 obtained theoretically ( pointed to by solid arrows ) , respectively , stand for the constraint in eqs .
( [ eq : re_func_condition ] ) and ( [ eq : disc_net_condition2 ] ) . in ( c ) and ( d ) , the value of @xmath160 varies but @xmath159 is set to @xmath163 .
the scale - free and random networks used in the simulations have @xmath229 and @xmath222 . ] for different pinning patterns . *
( a - c ) simulation results of @xmath191 from scale - free networks for @xmath124 , @xmath220 , and @xmath221 , which correspond to the results of @xmath65 in figs .
[ fig:2](a ) , [ fig:2](c ) , and [ fig:1](d ) , respectively .
( d - f ) theoretical results of @xmath191 from eq .
( [ eq : kappa ] ) for the cases shown in figs .
[ fig:5](a ) , [ fig:5](c ) , and [ fig:5](d ) , respectively .
the reference pinning pattern indicator is @xmath230 . ] and @xmath192 in eq .
( [ eq : pi ] ) . * ( a , b ) collapse of @xmath199 for various @xmath50 values , where the reference value is @xmath231 in ( a ) and @xmath205 in ( b ) .
the values of @xmath191 are predicted from eq .
( [ eq : kappa ] ) for a scale - free network with @xmath111 and @xmath229 .
( c ) the function @xmath200 for @xmath232 , @xmath224 , @xmath205 , @xmath163 , @xmath64 , and @xmath125 . ] | * resource allocation takes place in various types of real - world complex systems such as urban traffic , social services institutions , economical and ecosystems .
mathematically , the dynamical process of complex resource allocation can be modeled as _ minority games _ in which the number of resources is limited and agents tend to choose the less used resource based on available information .
spontaneous evolution of the resource allocation dynamics , however , often leads to a harmful herding behavior accompanied by strong fluctuations in which a large majority of agents crowd temporarily for a few resources , leaving many others unused .
developing effective control strategies to suppress and eliminate herding is an important but open problem .
here we develop a pinning control method . that the fluctuations of the system consist of intrinsic and systematic components allows us to design a control scheme with separated control variables .
a striking finding is the universal existence of an optimal pinning fraction to minimize the variance of the system , regardless of the pinning patterns and the network topology . we carry out a detailed theoretical analysis to understand the emergence of optimal pinning and to predict the dependence of the optimal pinning fraction on the network topology .
our theory is generally applicable to systems with heterogeneous resource capacities as well as varying control and network topological parameters such as the average degree and the degree distribution exponent .
our work represents a general framework to deal with the broader problem of controlling collective dynamics in complex systems with potential applications in social , economical and political systems . *
resource allocation is an essential process in many real - world systems such as ecosystems of various sizes , transportation systems ( e.g. , internet , urban traffic grids , rail and flight networks ) , public service providers ( e.g. , marts , hospitals , and schools ) , and social and economic organizations ( e.g. , banks and financial markets ) . the underlying system that supports resource allocation often contains a large number of interacting components or agents on a hierarchy of scales , and there are multiple resources available for each agent . as a result , complex behaviors are expected to emerge ubiquitously in the dynamical evolution of resource allocation . in particular , in a typical situation , agents or individuals possess similar capabilities in information processing and decision making , and they share the common goal of pursuing as high payoffs as possible .
the interactions among the agents and their desire to maximize payoffs in competing for limited resources can lead to vast complexity in the system dynamics .
given resource - allocation system that exhibits complex dynamics , a defining virtue of optimal performance is that the available resources are exploited evenly or uniformly by all agents in the system .
in contrast , an undesired or even catastrophic behavior is the emergence of herding , in which a vast majority of agents concentrate on a few resources , leaving many other resources idle or unused @xcite .
if this behavior is not controlled , the few focused resources would be depleted , possibly directing agents to a different but still small set of resources . from a systems point of view
, this can lead to a cascading type of failures as resources are being depleted one after another , eventually resulting in a catastrophic breakdown of the system on a global scale . in this paper , we analyze and test an effective strategy to control herding dynamics in complex resource - allocation systems . a universal paradigm to model and understand the interactions and dynamical evolutions in many real world systems is complex adaptive systems @xcite , among which minority game ( mg ) @xcite stands out as a particularly pertinent framework for resource allocation .
mg dynamics was introduced by challet and zhang to address the classic el farol bar - attendance problem conceived by arthur @xcite . in an mg system , each agent makes choice ( e.g. , @xmath0 or @xmath1 , to attend a bar or to stay at home ) based on available global information from the previous round of interaction .
the agents who pick the minority strategy are rewarded , and those belonging to the majority group are punished due to limited resources .
the mg dynamics has been studied extensively in the past @xcite .
to analyze , understand , and exploit the mg dynamics , there are two theoretical approaches : mean field approximation and boolean dynamics .
the mean field approach was mainly developed by researchers from the statistical - physics community to cast the mg problem in the general framework of non - equilibrium phase transitions @xcite . in the boolean dynamics , for any agent
, detailed information about the other agents that it interacts with is assumed to be available , and the agent responds accordingly @xcite .
both approaches can lead to `` better than random '' performance in resource utilization .
however , herding behavior in which many agents take identical action @xcite can also take place , which has been extensively studied and recognized as one important factor contributing to the origin of complexity that leads to enhanced fluctuations and , consequently , to significant degradation in efficiency @xcite .
the control strategy we analyze in this paper is the pinning method that has been studied in controlling the collective dynamics , such as synchronization , in complex networks @xcite . for the general
setting of pinning control , the two key parameters are the `` pinning fraction , '' the fraction of agents chosen to hold a fixed strategy , and the `` pinning pattern , '' the configuration of plus or minus strategies assigned to the pinned agents .
our previous work @xcite treated the special case of two resources of identical capacities , where the pinning pattern was such that the probabilities of agents pinned to positive or negative strategies ( to be defined later ) are equal . here
, we investigate a more realistic model setting and articulate a general mathematic control framework .
a striking finding is that biased pinning control pattern can lead to an optimal pinning fraction for a variety of network topologies , so that the system efficiency can be improved remarkably .
we develop a theoretical analysis based on the mean - field approximation to understand the non - monotonic behavior of the system efficiency about the optimal pinning fraction .
we also study the dependence of the optimal fraction on the topological features of the system , such as the average degree and heterogeneity , and obtain a theoretical upper bound of the system efficiency .
the theoretical predictions are validated with extensive numerical simulations .
our work represents a general framework to optimally control the collective dynamics in complex mg systems with potential applications in social , economical and political systems .
+ + * results * [ [ boolean - dynamics . ] ] * boolean dynamics .
* + + + + + + + + + + + + + + + + + + + in the original boolean system , a population of @xmath2 agents compete for two alternative resources , denoted as @xmath3 and @xmath4 , which have the same accommodating capacity @xmath5 .
similar to the mg dynamics , only the agents belonging to the _ global minority _ group are rewarded by one unit of payoff . as a result , the profit of the system is equal to the number of agents selecting the resource with attendance less than the accommodating capacity , which constitute the global - minority group .
the dynamical variable of the boolean system is denoted as @xmath6 , the number of @xmath0 agents in the system at time step @xmath7 .
the variance of @xmath6 about the capacity @xmath8 characterizes the efficiency of the system .
the densities of the @xmath0 and @xmath1 agents in the whole system are @xmath9 and @xmath10 , respectively .
the state of the system can be conveniently specified by the column vector @xmath11 .
a boolean system has two states ( a binary state system ) , in which agents make decision according to the local information from immediate neighbors .
the neighborhood of an agent is determined by the connecting structure of the underlying network .
each agent receives inputs from its neighboring agents and updates its state according to the boolean function , a function that generates either @xmath0 and @xmath1 from the inputs @xcite .
realistically , for any agent , global information about the minority choice from all other agents at the preceding time step may not be available . under this circumstance , the agent attempts to decide the global minority choice based on neighbors previous strategies . to be concrete , we assume @xcite that agent @xmath12 with @xmath13 neighbors chooses @xmath0 at time step @xmath14 with the probability @xmath15,\ ] ] and chooses @xmath1 with the probability @xmath16 , where @xmath17 and @xmath18 , respectively , are the numbers of @xmath0 and @xmath1 neighbors of @xmath12 at time step @xmath7 , with @xmath19 . the expressions of probabilities , however , are valid only under the assumption that the two resources have the _ same _ accommodating capacity , i.e. , @xmath20 . in real - world resource allocation systems , typically we have @xmath21 .
consider , for example , the extreme case of @xmath22 .
suppose we have @xmath23 for agent @xmath12 .
in this case , rationality demands a stronger preference to the resource @xmath0 ( i.e. , with a higher probability ) . to investigate the issues associated with the control of realistic boolean dynamics , we define @xmath24 where @xmath25 is the response function of each agent to its local environment @xmath26 , i.e. , the local neighbor s configuration with @xmath27 and @xmath28 .
the quantity @xmath29 ( or @xmath30 ) characterizes the contribution of the @xmath31-neighbors ( or @xmath32-neighbors ) to the probability for @xmath12 to adopt @xmath32 .
the quantity @xmath30 represents the strength of _ assimilation _ effect among the neighbors , while @xmath29 quantifies the _ dissimilation _ effect .
intuitively , the resource with a larger accommodating capacity would have a stronger assimilation effect among agents . by definition ,
the elements in each column in the matrix @xmath25 satisfy @xmath33 , i.e. , the total probability for an agent to choose @xmath0 and @xmath1 is unity .
using the mean - field assumption that the configuration of neighbors is uniform over the whole system , i.e. , @xmath34 , we have that the stable solution for eq .
( [ eq : interaction ] ) satisfies @xmath35 , which leads to the eigenstate of @xmath25 as @xmath36 the rational response ( @xmath25 ) of agents to nonidentical accommodation capacities of resources will lead to the equality @xmath37 , i.e. , the stable fraction of the agent densities in @xmath0 and @xmath1 is simply the ratio of the capacities .
the elements of @xmath25 can then be defined accordingly using this ratio and the condition @xmath38 , which characterizes a stronger preference to the resource with a larger capacity . for the specific case of identical - capacity resources
, we have @xmath39 , and the solution reduces to the result @xmath40 of the original boolean dynamics @xcite . the optimal solution for the resource allocation
is @xmath41 . a general measure of boolean system s performance is the variance of @xmath6 with respect to the _ capacity _ @xmath8 : @xmath42 which characterizes , over a time interval @xmath43 , the statistical deviations from the optimal resource utilization @xcite .
a smaller value of @xmath44 indicates that the resource allocation is more optimal .
a general phenomenon associated with boolean dynamics is that , as agents strive to join the minority group , an undesired herding behavior can emerge , as characterized by large oscillations in @xmath45 .
our goal is to understand , for the general setting of nonidentical resource capacities , the effect of pinning control on suppressing / eliminating the herding behavior . [
[ pinning - control - scheme . ] ] * pinning control scheme .
* + + + + + + + + + + + + + + + + + + + + + + + + + our basic idea to control the herding behavior is to `` pin '' certain agents to freeze their states so as to realize optimal resource allocation , following the general principle of pinning control of complex dynamical networks @xcite .
let @xmath46 be the fraction of agents to be pinned , so the fraction of unpinned ( or free ) nodes is @xmath47 .
the numbers of the two different types of agents , respectively , are @xmath48 and @xmath49 .
the free agents make choices according to local time - dependent information , for whom the inputs from the pinned agents are fixed .
the two basic quantities characterizing a pinning control scheme are the order of pinning ( the way how certain agents are chosen to be pinned ) and the pinning pattern @xcite .
we adopt the degree - preferential pinning ( dpp ) strategy in which the agents are selected to be pinned according to their connectivity or degrees in the underlying network .
in particular , agents of higher degrees are more likely to be pinned .
this pinning strategy originated from the classic control method to mitigate the effects of intentional attacks in complex networks @xcite .
the selection of the pinning pattern can be characterized by the fractions @xmath50 and @xmath51 of the pinned agents that select @xmath3 and @xmath4 , respectively , where @xmath52 .
the quantities @xmath50 and @xmath51 are thus the _ pinning pattern indicators_. different from the previous work @xcite that investigated the specific case of @xmath53 ( half - half pinning pattern ) , here we consider the more general case where @xmath50 is treated as a variable .
the pinning schemes are implemented on random networks and scale - free networks with different values of the scaling exponent @xmath54 in the power - law degree distribution @xcite @xmath55 .
as we will see below , one uniform optimal pinning fraction @xmath46 exists for various values of the pinning pattern indicator @xmath50 .
[ [ simulation - results . ] ] * simulation results .
* + + + + + + + + + + + + + + + + + + + + + to gain insight , we first study the original boolean dynamics with @xmath56 and @xmath57 for different values of the pinning pattern indicator @xmath50 .
the game dynamics are implemented on scale - free networks of size @xmath58 and of the scaling exponent @xmath59 with the average degree @xmath60 ranging from @xmath61 to @xmath62 .
the dpp scheme is performed with pinning fraction @xmath46 and @xmath50 values ranging from @xmath63 ( i.e. , half - half pinning ) to @xmath64 ( i.e. , all to @xmath0 pinning ) . the variance @xmath65 versus @xmath46 for different values of @xmath50 and different degree @xmath60 are shown in fig .
[ fig:1 ] .
we see that , in general , systems with larger values of @xmath50 exhibit larger variance , implying that a larger deviation of @xmath50 from the ratio of the capacity @xmath66 can lead to lower efficiency in resource allocation .
surprisingly , there exists a universal optimal pinning fraction ( denoted by @xmath67 ) about @xmath68 , where the variance @xmath65 is minimized and exhibits an opposite trend for @xmath69 , i.e. , larger values of @xmath50 result in smaller values of @xmath65 .
the implication is that , deviations of @xmath50 from @xmath66 provide an opportunity to achieve better performance ( with smaller variances @xmath65 ) , due to the non - monotonic behavior of @xmath65 with @xmath46 . to understand the emergence of the optimal pinning fraction @xmath67
, we see from fig .
[ fig:1 ] that the values of @xmath67 are approximately identical for different values of @xmath50 , which decrease with the average degree @xmath60 . as we will see below , in the large degree limit @xmath70
, the value of @xmath65 can be predicted theoretically ( c.f . , fig .
[ fig:4 ] ) .
simulations using scale - free networks of different degrees of heterogeneity also indicate the existence of the universal optimal pinning control strategy , as can be seen from the behaviors of the variance calculated from scale - free networks of different degree exponents ( fig .
[ fig:2 ] ) , where smaller values of @xmath54 point to a stronger degree of heterogeneity of the system .
we see that an optimal value of @xmath71 exists for all cases , which decreases only slightly with @xmath54 , i.e. , more heterogeneous networks exhibit larger values of the optimal pinning fraction @xmath71 , a phenomenon that can also be predicated theoretically ( c.f . , fig . [ fig:5 ] ) .
+ + * theoretical analysis * + + the phenomenon of the existence of a universal optimal pinning fraction @xmath72 , independent of the specific values of pinning pattern indicator @xmath50 , is remarkable .
here we develop a quantitative theory to explain this phenomenon . to begin , we note that mg is effectively a stochastic dynamical process due to the randomness in the selection of strategies by the agents .
the variance of the system , a measure of the efficiency of the system , is determined by two separated factors .
the first , denoted as @xmath73 , is the intrinsic fluctuations of @xmath74 about its expected value @xmath75 , defined as @xmath76 , which can be calculated once the stable distribution of attendance @xmath77 is known , where @xmath77 can be obtained either analytically ( c.f . , fig . [ fig:3 ] ) or numerically . the second factor , denoted as @xmath78 , is the difference of the expected value @xmath75 from the capacity @xmath8 of the system : @xmath79 , which leads to a constant contribution to the variance of the system . taking into account the two factors , we can write the system variance @xmath65 [ defined in eq .
( [ eq : sigma ] ) ] as @xmath80 a direct summation of the two factors @xmath81 and @xmath82 .
in contrast to the simplified case discussed in the previous works @xcite , the expected value @xmath75 in the dynamical process is not necessarily equal to the capacity @xmath8 .
nonzero values of @xmath83 are a result of biased pinning pattern ( @xmath84 ) or improper response to the capacities of resources .
in addition , recent studies of flux - fluctuation law in complex systems also found that , the variance of the system is determined by the two factors : the intrinsic fluctuation and the systematic external drives @xcite .
[ [ stable - distribution - of - attendance . ] ] * stable distribution of attendance . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to quantify the process of biased pinning control , we derive a discrete - time master equation and then discuss the effect of network topology on control . [ [ discrete - time - master - equation - for - biased - pinning - control . ] ] discrete - time master equation for biased pinning control .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to understand the response of the boolean dynamics to pinning control with varied values of the pinning pattern indicator @xmath50 , we generalize our previously developed analysis @xcite .
let @xmath85 be the probability for a _ neighbor _ of one given _
free _ agent to be pinned so that the probability of encountering a free agent is @xmath86 .
the transition probability of the system from @xmath45 to @xmath87 can be expressed in terms of @xmath50 .
in particular , note that the state transition is due to updating of the @xmath88 free agents , as the remaining @xmath89 agents are fixed . to simplify notations , we set @xmath90 , @xmath91 , and @xmath92 , for @xmath93 $ ] .
the conditional transition probability from @xmath12 at @xmath7 to @xmath94 at @xmath14 is @xmath95 where @xmath96/n_f$ ] is the probability for a free agent to choose @xmath0 with the first and second terms representing the contributions of the pinned @xmath1 and free @xmath1 neighbors , respectively . in the boolean system ,
the values of attendance @xmath74 oscillate about its equilibrium value @xcite .
the transition probability between the state at @xmath7 and @xmath97 can be expressed as a function of @xmath50 : @xmath98 \nonumber\\ & & \times [ { n_f\choose j - n_{p}\eta_{+ } } \times ( p_{p}\eta_{-}+p_{f}\frac{n_f - k+n_p\eta_{+}}{n_f})^{j - n_{p}\eta_{+ } } \nonumber\\ & & \times ( p_{p}\eta_{+}+p_{f}\frac{k - n_p\eta_{+ } } { n_f})^{n_{f}-(j - n_{p}\eta_{+ } ) } ] \}.\end{aligned}\ ] ] equation ( [ eq : t20pin2 ] ) takes into account the effect of pinning patterns , which was ignored previously @xcite .
the resulting balance equation governing the dynamics of the markov chains becomes @xmath99 which is the discrete - time master equation .
the stable state that the system evolves into can be defined in the matrix form as @xmath100 where @xmath101 is an @xmath102 matrix with elements @xmath103 , and @xmath104 is the corresponding vector of @xmath77 with @xmath74 ranging from @xmath105 to @xmath2 .
the probability distribution @xmath77 is a binomial function with various expectation values , as shown in fig .
[ fig:3 ] .
in addition , the probability @xmath77 is zero for @xmath106 \cup [ n_p(1-\eta_{+}),n]$ ] , which defines the boundary condition in the sense that there are @xmath107 pinned agents . once the stable distribution @xmath77 is obtained from eq .
( [ eq : distribution2 ] ) , the cumulative variance of the system can be calculated from @xmath108 the theoretical prediction of @xmath44 as a function of @xmath46 can thus be made through ( _ a _ ) identifying the function @xmath109 , ( _ b _ ) defining the matrix @xmath101 that depends on @xmath110 and @xmath109 , and ( _ c _ ) calculating the stable state @xmath77 .
[ [ effect - of - network - topology - on - pinning - control . ] ] effect of network topology on pinning control .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the topology of the network system has an effect on the probability @xmath85 . for the particular case of scale - free networks with degree exponent @xmath111
, our previous work @xcite demonstrated that preferential pinning of the large - degree agents leads to @xmath112 . here , we consider systems with degree distribution @xmath113 , where @xmath114 is the minimum degree of the network . for the dpp scheme where pinning occurs in the order from large to small degree agents ,
the relation between the minimum degree of _ pinned _ agents ( denoted by @xmath115 ) and the pinning fraction @xmath46 is @xmath116 for a given pinning fraction @xmath117 in which all the agents with @xmath118 are pinned , the probability @xmath85 for one neighbor of a given _ free _ agent to be a pinned agent is given by @xmath119 equations ( [ eq : rhopk ] ) and ( [ eq : rho_k_pk ] ) are applicable to dpp scheme on networks of any degree distribution @xmath120 . for a scale - free network , @xmath85 as a function of @xmath46 can be expressed as @xmath121 for the special case of @xmath111 , eq .
( [ eq : pfp_rho ] ) reduces to the specific relationship obtained earlier @xcite : @xmath122 . as indicated by eqs .
( [ eq : t20pin2]-[eq : sigmaa ] ) , the specific form of matrix @xmath101 with respect to @xmath46 can be obtained by substituting eq .
( [ eq : pfp_rho ] ) into eq .
( [ eq : t20pin2 ] ) , leading to the distribution @xmath77 and finally the variance of the system @xmath44 as a function of @xmath46 .
figure [ fig:4 ] displays the theoretical predicted @xmath44 ( dashed curves ) for various values of the pinning fraction @xmath46 and of the pinning pattern indicator @xmath50 .
the trend and , more importantly , the existence of the optimal pinning fraction @xmath71 , agree well with the simulation results ( marked with different symbols ) . in the limit @xmath123 ,
the system approaches a well - mixed state that can be fully characterized by eq .
( [ eq : pfp_rho ] ) , indicating that the simulation results approach the curve predicted by the mean - field theory as the average degree @xmath60 is increased .
figure [ fig:5 ] shows the theoretical prediction of @xmath44 for scale - free networks with different values of the degree exponent @xmath54 , which agrees well with the results from direct simulation as in fig .
[ fig:2 ] . for the case of highly heterogeneous networks ( @xmath124 ) ,
the theoretical prediction deviates slightly from the numerical results for the reason that the networks in simulation inevitably exhibit certain topological features that are not taken into account in the theoretical analysis of @xmath109 , such as the degree associativity .
[ [ optimal - pinning . ] ] * optimal pinning . *
+ + + + + + + + + + + + + + + + + + our analysis based on the master equation ( [ eq : distribution ] ) applies to systems with @xmath125 and identical resource capacity .
we now consider the more general case of varying @xmath25 values to further understand the optimal pinning control scheme .
[ [ deviation - of - expected - attendance - from - resource - capacity . ] ] deviation of expected attendance from resource capacity .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the dependence of @xmath75 on @xmath50 can be obtained through the general form of the response matrix @xmath25 . for convenience
, we use the column vector @xmath126 to denote the fraction of the agents pinned at @xmath0 and @xmath1 , where @xmath52 , @xmath127 is the fraction of free agents adopting strategies @xmath0 and @xmath1 , respectively , with @xmath128 .
the state of the system can be expressed as @xmath129 , from which we have @xmath130/\rho_{f},\end{aligned}\ ] ] at the next time step , the expected value of the state based on @xmath131 through the response matrix @xmath25 can be written as @xmath132.\end{aligned}\ ] ] substituting eq .
( [ eq : etarhop ] ) into eq .
( [ eq : rhot+1 ] ) , we get the relationship between @xmath133 and @xmath131 .
a self - consistency process stipulated by eqs .
( [ eq : etarhop ] ) and ( [ eq : rhot+1 ] ) can yield the stable state of the system with the expected number of agents choosing @xmath0 given by @xmath134 in a free system without pinning , the rational response @xmath25 of agents to nonidential capacities of resources leads to eq .
( [ eq : optimal ] ) , implying the relationship @xmath135 . from eq .
( [ eq : pstabel ] ) , we can obtain @xmath83 as a function of the value of the pinning pattern indicator @xmath50 , the elements of the matrix @xmath25 , the pinning fraction @xmath46 , and the parameter @xmath85 associated with network topology .
we have @xmath136\cdot \nonumber\\ & & \frac{[\rho_{p}(f_{+|- } + f_{-|+})-p_{p}(f_{+|-}+f_{-|+}-1)]}{(f_{+|- } + f_{-|+})\cdot[f_{+|-}+f_{-|+}-(f_{-|+}+f_{-|+}-1)p_{p}]},\end{aligned}\ ] ] which has the form of separated variables associated with @xmath50 and @xmath46 . [
[ optimal - pinning - pattern - and - fraction . ] ] optimal pinning pattern and fraction .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + optimizing the system requires minimum @xmath78 , i.e. , @xmath137 in eq .
( [ eq : deltap ] ) , leading to two independent solutions : @xmath138 which respectively correspond to the optimal value of the pinning pattern indicator @xmath139 and the optimal pinning fraction @xmath140 . here , for convenience , we define a parameter : @xmath141 so that eq .
( [ eq : optimal_rho ] ) can be expressed concisely as @xmath142 . once the values of @xmath50 and
@xmath46 satisfy either eq .
( [ opt_pin_pattern ] ) or eq .
( [ eq : optimal_rho ] ) , we can obtain @xmath143 .
the variance @xmath44 depends on the fluctuation factor @xmath144 only .
equation ( [ opt_pin_pattern ] ) specifies the pinning pattern with the same ratio as that of the resource capacity .
the boolean dynamics studied previously @xcite is a special case where the optimal pinning pattern indicator is @xmath145 for systems with @xmath146 , and the variance @xmath65 is simply determined by the factor @xmath144 . from eq .
( [ eq : optimal_rho ] ) , we see that the optimal pinning fraction @xmath71 is independent of @xmath50 but depends on both the network structure through @xmath109 and on the response function @xmath25 . additionally , the condition @xmath147 $ ] and nonzero denominator require @xmath148 the function @xmath109 for scale - free networks , as in eq .
( [ eq : pfp_rho ] ) , increases monotonically with @xmath46 .
figure [ fig:6](a ) displays the curves @xmath149 and @xmath150 , i.e. , both sides of eq .
( [ eq : optimal_rho ] ) .
the existence of nonzero @xmath71 for @xmath151 demands @xmath152 for scale - free networks , @xmath109 diverges at @xmath153 .
equation ( [ eq : cont_net_conditon ] ) thus holds , implying that the dpp pinning scheme has a nonzero optimal pinning fraction @xmath71 , leading to @xmath137 .
however , for homogeneous networks , eq .
( [ eq : cont_net_conditon ] ) may not hold . in this case
, a more specific implicit condition can be obtained from eq .
( [ eq : cont_net_conditon ] ) through the following analysis . in particular , without an analytical expression of @xmath109 , the derivative of @xmath109 with respect to @xmath46 can be obtained from eqs .
( [ eq : rhopk ] ) and ( [ eq : rho_k_pk ] ) : @xmath154 for degree preferential pinning , in the limit @xmath155 , the maximum degree for _ free _ agents is @xmath156 .
we thus have @xmath157 which requires that the network be heterogeneous . for @xmath125
, we have @xmath158 , ensuring the existence of a nonzero @xmath71 value for @xmath137 .
the contour map of the optimal pinning fraction @xmath71 in the parameter space of @xmath159 and @xmath160 for scale - free networks with @xmath111 is shown in fig .
[ fig:6](b ) . the boundary @xmath161 associated with condition eq .
( [ eq : re_func_condition ] ) is represented by the white dashed line , where nonzero solutions of @xmath71 do not exist below the lower - left region .
figures [ fig:6](c ) and [ fig:6](d ) show @xmath71 for @xmath137 as a function of @xmath162 for scale - free and random networks , respectively , where @xmath160 is varied and @xmath159 is fixed to @xmath163 .
the theoretical prediction of @xmath71 [ red solid curve in ( c ) and red open circle in ( d ) ] is given by the intersections of the curves @xmath164 and @xmath165 in fig .
[ fig:6](a ) . for scale - free networks , since eq .
( [ eq : cont_net_conditon ] ) holds , eq .
( [ eq : re_func_condition ] ) is the only constraint on the value of @xmath162 ( red dashed arrow ) , with the region at the right - hand side yielding nonzero @xmath71 solutions . the red solid curve in fig .
[ fig:6](c ) represents the theoretical prediction , and the open squares denote the simulation results from scale - free networks of size @xmath166 , power - law exponent @xmath111 , and average degree @xmath167 . for random networks , the existence of nonzero @xmath71 solutions requires that eqs .
( [ eq : re_func_condition ] ) and ( [ eq : disc_net_condition ] ) or ( [ eq : disc_net_condition2 ] ) hold . for the poisson degree distribution ,
the maximum degree of the network can be calculated from @xmath168 we can obtain an estimate of the value of @xmath162 that satisfies eq .
( [ eq : disc_net_condition2 ] ) , as indicated by the blue arrow ( labeled as boundary 2 ) in fig .
[ fig:6](d ) .
the right - hand side of this point satisfies both eqs .
( [ eq : re_func_condition ] ) and ( [ eq : disc_net_condition2 ] ) , implying the existence of nonzero @xmath71 . comparison of the results from random and scale - free networks with different scaling exponents ( figs .
[ fig:2 ] , [ fig:5 ] and [ fig:6 ] ) shows that , stronger heterogeneity tends to enhance the values of @xmath71 , which can also be seen from eq .
( [ eq : cont_net_conditon ] ) . to better understand the non - monotonic behavior of @xmath65 with @xmath46
, we provide a physical picture of the behavioral change for @xmath46 greater or less than @xmath71 .
the effect of pinning control is determined by the number of edges between pinned and free agents , which are _ pinning - free edges_. for a small pinning fraction @xmath46 , the average effect per pinned agent on the system ( represented by the number of pinned - free edges per pinned agent ) is relatively large .
however , as @xmath46 is increased , the average impact is reduced for two reasons : ( _ a _ ) an increase in the edges within the pinned agents community itself ( i.e. , two connected pinned agents ) , which has no effect on control , and ( _ b _ ) a decrease in the number of free agents , which directly reduces the number of pinned - free edges .
consider the special case of @xmath169 and @xmath20 . for small @xmath46 ,
the pinned @xmath0 agents have a significant impact so that the free agents tend to overestimate the probability of winning by adopting @xmath1 . in this case , the expected value @xmath170 is smaller than @xmath171 , corresponding to @xmath172 . for highly heterogeneous systems ,
the average impact per pinned agent is larger for a given small value of @xmath173 .
as @xmath46 is increased , the average influence per pinned agent reduces and , consequently , @xmath170 restores towards @xmath174 . for @xmath175 and @xmath137 , the system variance [ eq . ( [ eq : sigmaf1f2 ] ) ]
is minimized due to @xmath143 , and the corresponding pinning fraction achieves the optimal value @xmath71 . for strongly heterogeneous systems , due to the large initial average impact caused by pinning the hub agents
, the optimal pinning fraction @xmath71 appears in the larger @xmath173 region .
further increase in @xmath46 with @xmath169 will lead to @xmath176 and @xmath177 , thereby introducing nonzero @xmath78 again and , consequently , generating an increasing trend in @xmath65 . [ [ collapse - of - variance . ] ] collapse of variance .
+ + + + + + + + + + + + + + + + + + + + + for certain networks , the variance @xmath178 is determined by the values of the pinning pattern indicator @xmath50 and the pinning fraction @xmath46 .
our analysis so far focuses on the contribution of @xmath78 to the variance @xmath65 as the pinning fraction @xmath46 is increased but for fixed @xmath50 .
it is thus useful to define a quantity related to the variance @xmath65 , which can be expressed in the form of separated variables . for two different values of the pinning pattern indicator , @xmath50 and @xmath179 , for a given value of @xmath46
, the relative weight of @xmath78 can be obtained from eq .
( [ eq : deltap ] ) as @xmath180 ^ 2 = [ \frac{f_{+|-}(\eta_{+}-1)+f_{-|+}\eta_{+ } } { f_{+|-}(\eta_{+}^{\prime}-1)+f_{-|+}\eta_{+}^{\prime}}]^2,\end{aligned}\ ] ] where @xmath181 is a function of both @xmath50 and @xmath46 . remarkably , the ratio @xmath182 depends on @xmath50 and @xmath179 but it is independent of @xmath46 , due to the form of separated variables in eq .
( [ eq : deltap ] ) . from the simple relationship eq .
( [ eq : lambda ] ) , we can define the relative changes in these quantities due to an increase in the value of @xmath50 from a _ reference value @xmath179 _ as @xmath183 and then obtain the change rate associated with @xmath65 and @xmath78 as , @xmath184 where @xmath185 is independent of @xmath46 . in the limit @xmath186 , the rate of change @xmath187 becomes @xmath188 }
{ \partial\eta_{+}^{\prime}}/\frac{d \ln [ x_{2 } ( \varepsilon(\eta_{+}^{\prime}))]}{d \eta_{+}^{\prime}}.\end{aligned}\ ] ] figure [ fig:7 ] shows @xmath189 as a function of @xmath46 for scale - free networks , where the value of the reference pinning pattern indicator is @xmath190 . to obtain the values of @xmath191 , we first calculate @xmath192 by substituting the values of @xmath50 , @xmath193 and the elements of @xmath25 into eqs .
( [ eq : lambda ] ) and ( [ eq : omega ] ) .
we then obtain @xmath194 by substituting the values of @xmath44 into eq .
( [ eq : pi1 ] ) , with @xmath44 either from simulation as in figs
. [ fig:1 ] and [ fig:2 ] or from theoretical analysis as in fig .
[ fig:5 ] .
we see that the @xmath191 values from simulation results of @xmath44 [ figs .
[ fig:7](a - c ) marked by `` simulation results '' ] and theoretical prediction of @xmath44 [ figs .
[ fig:7](d - f ) marked by `` theoretical results '' ] show the behavior in which the curves of @xmath191 for different values of @xmath50 collapse into a single one .
this indicates that @xmath191 depends solely on the pinning fraction @xmath46 ; it is independent of the value of the pinning pattern indicator @xmath50 .
extensive simulations and analysis of scale - free networks with different average degree @xmath60 or different degree exponent @xmath54 verify the generality of the collapsing behavior . from eq .
( [ eq : kappa2 ] ) , we see that the variance @xmath195 and the quantity @xmath191 are closely related . for example
, a smaller value of @xmath191 indicates that @xmath78 contributes more to the variance of @xmath195 as @xmath50 is changed , and vice versa . in fig .
[ fig:7 ] , @xmath196 corresponds to the intersecting points of the curves of @xmath65 with different values of @xmath50 shown in figs .
[ fig:1 ] , [ fig:2 ] , and [ fig:5 ] .
it can also be verified analytically that , the minimal point with @xmath197 coincides with the optimal pinning fraction @xmath71 at which @xmath65 is minimized , which is supported by simulation results in figs .
[ fig:7 ] , [ fig:1 ] , [ fig:2 ] , and [ fig:5 ] . [ [ variance - in - the - form - of - separated - variables . ] ] variance in the form of separated variables .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + from eq .
( [ eq : kappa ] ) , for a given value of the reference pinning pattern indicator @xmath179 , we can obtain an expression of @xmath194 in the form of separated variables as @xmath198 where @xmath199 is independent of the change in @xmath50 , and @xmath200 is independent of @xmath46 . the consequence of eq .
( [ eq : pi ] ) is remarkable , since it defines in the parameter space @xmath201 a function @xmath194 in the form of separated variables which , as compared with the original quantity @xmath65 , not only simplifies the description but also gives a more intuitive picture of the system behavior .
specifically , for the mg dynamics , the influences of various factors on the variance @xmath65 or @xmath202 can be classified into two parts : ( i ) the function @xmath191 that reflects the effects of the pinning fraction @xmath173 and the network structure among agents ( in terms of the degree distribution @xmath120 , the average degree @xmath60 , and the scaling exponent @xmath54 ) , and ( ii ) the function @xmath192 that characterizes the impact of the pinning pattern indicator @xmath50 and the response of agents to resource capacities @xmath8 and @xmath203 through @xmath25 .
figures [ fig:8](a ) and [ fig:8](b ) show the values of @xmath191 as a function of @xmath46 for @xmath204 and @xmath205 , respectively , whereas fig .
[ fig:8](c ) shows @xmath200 for several values of @xmath179 . from eqs .
( [ eq : lambda ] ) and ( [ eq : omega ] ) , we see that @xmath192 is a quadratic function of @xmath50 with the symmetry axis at @xmath206 , which depends on the setting of response function @xmath25 .
the second derivative of the function depends on @xmath193 . from the definition in eq .
( [ eq : pi1 ] ) , the variance of the system for arbitrary values of @xmath46 and @xmath50 can be obtained as @xmath207 \cdot\sigma^{2}(\eta_{+}^{\prime},\rho_{p}),\end{aligned}\ ] ] where @xmath179 specifies the reference pinning pattern .
once we have the two respective @xmath195 curves for the two specific pinning patterns as specified by @xmath208 and @xmath209 , @xmath65 in the whole parameter space @xmath201 can be calculated accordingly .
in particular , the quantities @xmath210 and @xmath211 serve as a _
holographic _ representation of the dynamical behavior of the system in the whole parameter space .
in particular , one can first obtain @xmath200 from eqs .
( [ eq : deltap ] ) and ( [ eq : omega ] ) , and then calculate @xmath212 , and finally obtain the value of @xmath195 by substituting @xmath200 and @xmath213 into eq .
( [ eq : sigma2 ] ) .
+ + * discussions * + + the phenomenon of herding is ubiquitous in social and economical systems .
especially , in systems that involve and/or rely on fair resource allocation , the emergence of herding behavior is undesirable , as in such a state a vast majority of the individuals in the system share only a few resources , a precursor of system collapse at a global scale .
a generic manifestation of herding behavior is relatively large fluctuations in the dynamical variables of the system such as the numbers of individuals sharing certain resources .
it is thus desirable to develop effective control strategies to suppress herding .
an existing and powerful mathematical framework to model and understand the herding behavior is minority games . investigating
control of herding in the mg framework may provide useful insights into developing more realistic control strategies for real - world systems .
built upon our previous works in mg systems @xcite , in this paper we articulate , test , and analyze a general pinning strategy to control herding behavior in mg systems .
a striking finding is the universal existence of an optimal pinning fraction that minimizes the fluctuations in the system , regardless of system details such as the degree of homogeneity of the resource capacities , topology and structures of the underlying network , and different patterns of pinning .
this means that , generally , the efficiency of the system can be optimized for some relatively small pinning fraction . employing the mean - field approach
, we develop a detailed theory to understand and predict the dynamics of the mg system subject to pinning control , for various network topologies and pinning schemes . the key observation underlying our theory
is the two factors contributing to the system fluctuations : intrinsic dynamical fluctuations and systematic deviation of agents expected attendance from resource capacity .
the theoretically predicted fluctuations ( quantified by the system variance ) agree with those from direct simulation .
in particular , in the large degree limit , for a variety of combinations of the network and pinning parameters , the numerical results approach those predicted from our mean field theory . our theory also correctly predicts the optimal pinning fraction for various system and control settings
. in real world systems in which resource allocation is an important component , resource capacities and agent interactions can be diverse and time dependent . to develop mg model to understand the effects of diversity and time dependence on herding dynamics , and to exploit the understanding to develop optimal control strategies to suppress or eliminate herding are open issues at the present
furthermore , implementation of pinning control in real systems may be associated with incentive policies that provide compensations or rewards to the pinned agents .
how to minimize the optimal pinning fraction then becomes an interesting issue .
our results provide insights into these issues , and represent a step toward the goal of designing highly stable and efficient resource allocation systems in modern society and economy .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ _ ( , ) . .
_ _ * * , ( ) . , & _
_ , vol .
( , ) . & .
_ _ * * , ( ) . , , _
et al . _ . _
_ ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ ( ) . , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . _
_ , chap .
( , ) . & . in ( ed
. )
_ _ , ( , ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , & .
_ _ * * , ( ) . |
formulation of chiral fermions on the lattice has been one of long - standing problems in lattice field theories .
several years ago , the domain - wall fermion ( dwf ) formalism @xcite , which is a wilson fermion in @xmath1 dimensions with dirichlet boundary condition in the extra dimension , has been proposed as a new formulation of lattice chiral fermion . in the limit of large extra dimension size , @xmath2 , the spectrum of free domain - wall fermion contains massless modes at the edges in the extra dimension .
while the massless modes are shown to be stable in perturbation theory@xcite , their existence may be spoiled non - perturbatively in the presence of dynamical gauge fields .
we studied this issue through an anomalous quark mass @xmath0 in ref .
this quantity measures the magnitude of chiral symmetry breaking with the domain - wall qcd ( dwqcd ) . in this article
we make a status report of our attempt to understand the results on the @xmath3-dependence of @xmath0 obtained in ref .
@xcite through measurements of the eigenvalue distribution of the 4-dimensional wilson - dirac operator .
we define the anomalous quark mass by@xcite @xmath4 this quantity measures the chiral symmetry breaking effect in the axial ward - takahashi identity : @xmath5 where @xmath6 is the axial - vector current , @xmath7 is the pseudoscalar density , and @xmath8 represents the explicit breaking of chiral symmetry . on a smooth gauge field background ,
the anomalous contribution @xmath9 vanishes as @xmath10 at large @xmath3 @xcite .
therefore @xmath0 also vanishes exponentially in this case .
= 7.5 cm in ref .
@xcite , we carried out quenched simulations to study the @xmath3 dependence in detail .
two values of lattice spacing , @xmath11 and 2 gev , are explored , using both the plaquette and an rg - improved gauge actions .
our main results for @xmath0 are summarized in fig .
[ fig : m5q ] .
the results for the plaquette action are obtained at @xmath12 ( @xmath13 gev ) and 6.0 ( 2 gev ) , and those for the rg - improved action at @xmath14 ( 1 gev ) and 2.6 ( 2 gev ) .
solid lines are fits to @xmath15 , and dashed lines to @xmath16 important points to note in fig .
[ fig : m5q ] are : ( i ) comparing results for the spatial sizes @xmath17 and @xmath18 at @xmath13 gev , we find the finite volume effects in @xmath0 to be small , ( ii ) @xmath0 decreases with the lattice spacing , ( iii ) the magnitude of @xmath0 is smaller for the rg - improved action than for the plaquette action , and ( iv ) from data in the range of @xmath3 we explore , @xmath0 seems to remain non - zero in the limit @xmath19 , in all cases except at @xmath20 for the rg - improved action .
if confirmed with studies at larger values of @xmath3 , the last point means that dwqcd realizes chiral symmetry at @xmath21 gev only for the case of the rg - improved action .
chiral symmetry of dwqcd can be studied also through the transfer matrix in the direction of the extra dimension@xcite .
when the transfer matrix has a unit eigenvalue , chiral symmetry is not realized in dwqcd because the left and right chiral modes on the two edges in the extra dimension couple with each other .
a unit eigenvalue of the transfer matrix is in one - to - one correspondence with a zero eigenvalue of the hermitian wilson - dirac operator defined by @xmath22 which is much easier to calculate . here
, @xmath23 is the four dimensional wilson - dirac kernel with a bare mass @xmath24 .
therefore , a failure of exponential decay of @xmath0 would result if @xmath25 develops a zero eigenvalue .
we calculate eigenvalues of @xmath26 by the lanczos method using 50100 configurations at several values of coupling in the range @xmath272 gev using both plaquette and rg - improved actions .
the results from the lanczos method are checked by the ritz functional method for @xmath25 .
we also study the dependence on the lattice size . the maximum lattice at @xmath27 gev is @xmath28 for both actions , while the one at @xmath21 gev is @xmath29 for the rg - improved action and @xmath30 for the plaquette action .
= 7.5 cm = 7.5 cm in figs .
[ fig : evdistp ] and [ fig : evdistr ] we plot monte carlo time histories for the six lowest eigenvalues of @xmath26 for the plaquette and rg - improved actions . in each figure
the left panel shows results for @xmath13 gev and the right panel for @xmath21 gev .
the lattice size at @xmath31 gev is the same as in the previous work of @xmath0 shown in fig .
[ fig : m5q ] .
open squares plot the minimum eigenvalue @xmath32 and filled diamonds are the five higher eigenvalues .
there is a clear trend that the minimum eigenvalues become larger for smaller lattice spacings .
another interesting point is that the rg - improved action gives larger values of @xmath32 than the plaquette action , which indicates that the rg - improved action has a better chiral behavior .
these trends are parallel to the features we noted for @xmath0 in sec .
[ sec : m5q ] .
the spectral density of @xmath25 is defined by @xmath33 where the summation is over the eigenvalues of @xmath25 .
we are interested in the density of zero eigenvalues , @xmath34 , since we expect this quantity to be related to the existence of unit eigenvalue of the transfer matrix . to calculate this quantity
, we adopt the accumulation method proposed in @xcite , which is based on the relation @xmath35 where @xmath36 is the spectral density function for @xmath26 .
we note that , for the small-@xmath37 expansion of @xmath38 in ( [ eq : ansatz ] ) , analyticity of @xmath39 at the origin is assumed .
= 7.5 cm = 7.5 cm = 7.5 cm in fig .
[ fig : acc ] , we show typical examples of the accumulation @xmath38 from the eigenvalue distribution of @xmath26 for the case of the rg - improved action .
results for @xmath34 obtained by a linear fitting following ( [ eq : ansatz ] ) , normalized by the string tension , are summarized in fig .
[ fig : rhob ] .
our results for @xmath34 for the plaquette action are consistent with the previous data by edwards _
results for the rg - improved action show a similar @xmath40 dependence .
a significant difference is that the rg - improved action leads to much smaller values of @xmath34 than the plaquette action , roughly by an order of magnitude .
we have applied the accumulation method to estimate the spectral density at zero eigenvalue of the hermitian wilson - dirac operator , @xmath34 .
we found that this method leads to non - zero values of @xmath34 at @xmath412 gev for both the plaquette and rg - improved actions . at @xmath41 gev ,
the non - zero result for @xmath34 is consistent with the finite @xmath0 in the large @xmath3 limit observed in @xcite with both the plaquette and rg - improved actions . at @xmath42 gev , while a consistency also holds with the plaquette action , there is an apparent contradiction for the case of the rg - improved action since @xmath0 seems to decay exponentially with @xmath3 for this case . in the accumulation data shown in fig .
[ fig : acc ] we observe that results are very noisy at @xmath20 ( 2 gev ) . since the fit results for @xmath34 fluctuates with volume , it is difficult to determine the size dependence .
therefore simulations with larger lattices are needed to check if the slope remains non - vanishing toward infinite volume .
another point to examine is if the analyticity assumption for @xmath39 at the origin is justified if there is a spectral gap .
further studies are required to clarify these points .
this work is supported in part by grants - in - aid of the ministry of education ( nos.10640246 , 10640248 , 10740107 , 11640250 , 11640294 , 11740162 , 12014202 , 12304011 , 12640253 , 12740133 ) .
aak and tm are supported by jsps research for the future program ( no .
jsps - rftf 97p01102 ) .
se , tk , kn , jn and hps are jsps research fellows .
l d. kaplan , phys .
b288 * ( 1992 ) 342 .
y. shamir , nucl .
* b406 * ( 1993 ) 90 .
s. aoki and y. taniguchi , phys . rev .
* d59 * ( 1999 ) 054510 .
y. kikukawa , h. neuberger and a. yamada , nucl .
b526 * ( 1998 ) 572 .
cp - pacs collaboration : a. ali khan _ et al .
_ , hep - lat/0007014 . v. furman and y. shamir , nucl . phys .
* b439 * ( 1995 ) 54 .
h. neuberger , phys . rev .
* d57 * ( 1998 ) 5417 .
y.kikukawa and t.noguchi,hep-lat/9902022 .
r. edwards , u. heller and r. narayanan , phys . rev .
* d60 * ( 1999 ) 034502 . | chiral properties of qcd formulated with the domain - wall fermion ( dwqcd ) are studied using the anomalous quark mass @xmath0 and the spectrum of the 4-dimensional wilson - dirac operator .
numerical simulations are made with the standard plaquette gauge action and a renormalization - group improved gauge action .
results are reported on the density of zero eigenvalue obtained with the accumulation method , and a comparison is made with the results for @xmath0 . |
in thermal field theory , in order to deal with the infrared singularities which occur at finite temperature , it is necessary to put thermal masses into the zeroth order of a resummed perturbation theory . to this end
, one must first calculate the hard thermal loops , where all the external energies and momenta are much smaller than the temperature @xmath0 .
these loops yield gauge - invariant contributions , which are in general non - local functionals of the external fields @xcite . however , there are two special cases : the static and the long wave - length limits , when the hard thermal amplitudes become local functions , which are independent of the external energies and momenta @xcite .
nevertheless , these two limits give different functions @xcite . of special interest
is the evaluation in the above limits of causal thermal self - energy functions , which determine the high - temperature behavior of screening lengths and plasma frequencies @xcite .
the purpose of this work is to derive a simple method for calculating the leading contributions of retarded thermal loops in static external bosonic fields .
we will show that these contributions can be directly obtained by evaluating the hard thermal loops at zero external energies and momenta .
this result has been previously derived in gauge theories at one - loop level @xcite and verified by explicit calculations at two - loops @xcite . here , we present an argument which is valid to all orders in thermal perturbation theory . the above result may be more readily understood in the analytically continued imaginary - time formalism @xcite which is well suited for the study of retarded green s functions @xcite .
this formalism of thermal field theory defines the bosonic green s functions at integral values of @xmath1 , where @xmath2 is the energy of the @xmath3-th external particle . after performing the sums over the integral values ( half - integral for fermions ) of @xmath4 , where @xmath5 is the energy of the @xmath6-th internal particle , one arrives at bosonic ( fermionic ) thermal distribution functions of the form @xmath7 where @xmath8 is some linear combination of external energies and
@xmath9 is some combination of external and internal momenta .
it is worthwhile to note that if one would now analytically continue the external energies , one would get an analytic behavior when all @xmath10 and @xmath11 become small , leading to a well defined result in the limit @xmath12 .
however , the analytic continuation constructed in this manner would yield , after performing the integrations over internal momenta , factors like @xmath13 which are exponentially increasing for large values of @xmath14 .
the proper procedure which avoids the appearance of such factors makes use of the relation @xmath15 which is valid before analytic continuation , since then , @xmath16 is an integer . in this way , the green s functions will be well behaved when @xmath2 are analytically continued to complex values , and various limits approaching the real axis from different directions may be taken .
but this procedure , in contrast to the previous one , introduces non - analyticities in the thermal loops when @xmath17 and @xmath18 .
nevertheless , the leading contributions of hard thermal loops in the static case @xmath19 , @xmath20 , still agree with those obtained by setting directly in the loops all external energies and momenta @xmath21 .
this agreement occurs only in the static limit which entails that @xmath22 , since then the condition reduces to an identity .
thus , in the static case , analytic continuation preserves the form of the original thermal distribution functions , which lead to an analytic behavior when all @xmath18 . we will first exemplify the above argument in the case of a two - loop thermal amplitude .
next , we shall present a more general approach , based on a spectral representation of the thermal green s functions @xcite , which allows to verify this argument to all orders .
let us consider , for example , a two - loop diagram in the scalar @xmath23 theory , as shown in figure [ fig1 ] .
the result for the one - loop sub - diagram in the renormalizable six - dimensional theory may be written in the @xmath24 where @xmath25 is a polynomial of degree @xmath26 in @xmath27 and @xmath28 ( when @xmath29 , there appear additional logarithmic terms ) and @xmath30 is given by @xmath31-\frac{1}{4\pi^2 } \log \frac{t}{\mu } \right\},\ ] ] where @xmath32 is a renormalization mass scale .
the above one - loop result for @xmath30 has branch points at @xmath33 . when this term is inserted in the two - loop graph of fig .
[ fig1 ] , these branch points coincide with the poles at @xmath33 which arise from the other terms in this diagram .
therefore , in order to perform the @xmath27 integration in a well defined manner , it is necessary to regularize @xmath34 , which we do using dimensional regularization in @xmath35 dimensions .
this leads to the regularized form of @xmath30 given by @xmath36-\frac{1}{4\pi^2 } \log \frac{t}{\mu}\right\},\ ] ] which reduces to eq . in the limit @xmath37 . in eq .
, @xmath38 denotes the angle between the momenta @xmath39 and @xmath40 shown in fig .
[ fig1 ] and @xmath41 denote terms which are well behaved at @xmath33 . using this result in the two - loop graph of fig .
[ fig1 ] , and performing the sum over the integral values of @xmath42 by contour integration , we obtain a leading thermal contribution of the form @xmath43\ ] ] where the anticlockwise contour @xmath44 , along the imaginary @xmath27-axis , is closed in the right half @xmath27 plane .
evaluating the @xmath27-integral in terms of the poles inside @xmath44 , and using the fact that the leading contribution from the pole at @xmath45 in @xmath46 vanishes , leads to the result ( @xmath47 ) @xmath48.\end{aligned}\ ] ] at this point , we note that if one would analytically continue to complex values of @xmath8 , one would obtain an expression which is analytic as @xmath49 , with the limit @xmath50^\prime_{p_0=p } - \frac{n(p ) \pi_t(p , p,\epsilon)}{p}\right\ } , \ ] ] where the prime denotes a derivative with respect to @xmath27 .
however , for the reasons mentioned previously , one must first use in the relation @xmath51 and then make the analytic continuation of @xmath8 .
this procedure will introduce a non - analyticity when @xmath49 . to study this
, we remark that the leading contribution in @xmath0 of the @xmath28-integral in comes from the region where @xmath52 .
thus , in order to find the high - temperature behavior of the hard thermal loop , one may assume that @xmath53 , @xmath54 .
then , it is easy to see that the leading contribution obtained in the static limit agrees with the result given in @xmath55 on the other hand , the leading contributions in the long wave - length limit would lead to a different result , namely @xmath56.\ ] ] let us now evaluate the leading static thermal contribution which arises from . in this case
, there appear individual terms proportional to @xmath57 which exhibit a collinear singularity in the region where @xmath58 , with @xmath39 and @xmath40 being nearly parallel ( see fig . [ fig1 ] ) .
however , such collinear singularities turn out to cancel so that in the present case , it is not necessary to resort to the kln mechanism at finite temperature @xcite . using eqs . and , together with the relations @xmath59 one then gets a leading @xmath60 contribution of the form @xmath61
in order to verify to all orders that the leading static contributions agree with the result obtained by setting directly , in the self - energy , the external energy - momentum equal to zero , it is convenient to use a spectral representation of the analytically continued bosonic self - energy ( see , for example , chapter 3 in @xcite and references therein ) .
one then obtains for the analytic retarded self - energy the spectral form @xmath62 where @xmath8 is a real energy and , for simplicity , only the energy - momentum dependence has been written explicitly . here
, the spectral density @xmath63 is related to the discontinuity of @xmath64 across the real axis and the second term approaches zero when @xmath65 .
these features can be easily seen at one - loop in the scalar @xmath66 model , where @xmath67 and @xmath68.\ ] ] here , @xmath69 and the free spectral density @xmath70 is given by @xmath71 where @xmath72 is the ordinary mass of the scalar particles .
one can check that , by substituting into and performing the @xmath73 integration , one gets the one - loop thermal contribution @xmath74 , \ ] ] where @xmath75 is to be understood and we have neglected @xmath72 with respect to @xmath28 .
this agrees with the result obtained from by setting @xmath76 and employing the relation .
evaluating in the static and long wavelength limits , leads to distinct terms of order @xmath77 . using the spectral representation , where we take for definiteness @xmath78
, we now consider the leading thermal contributions which arise in the long wave - length and static limits of the retarded self - energy @xmath79 @xmath80 next , let us compare these contributions with the result obtained by setting directly @xmath22 , @xmath81 in the retarded thermal self - energy function @xmath82 since the integrand in @xmath83 is not a uniformly continous function of @xmath8 , we can not take the limit @xmath84 inside the integral .
thus , we infer that the leading thermal contribution got in the long wave - length limit , would generally differ from the result given in .
on the other hand , we will argue that in the static case , the limit @xmath85 can be taken inside the integral . to this end , using the fact that @xmath86 is an odd function of @xmath73 , it is convenient to write in the alternative form ( where @xmath87 denotes the principal value ) @xmath88 consider now the integral @xmath89 it is well known that if @xmath90 converges uniformly , then the limit @xmath85 can be taken inside the integral .
such a convergence may be shown by considering the physical meaning of @xmath86 , which gives the imaginary part of the retarded self - energy .
it yields the rates of processes occurring in a thermal plasma , such as particle creation / annihilation or scattering , in the presence of an external field @xcite . at high temperatures , such that
@xmath91 , the leading contributions to these rates have a smooth behavior when @xmath85 , in which case @xmath86 would be a well behaved function in this limit .
furthermore , in consequence of unitarity ( conservation of probability ) such rates should decrease at large values of the energy @xmath73 .
assuming , for example , that @xmath86 behaves for large @xmath73 like @xmath92 , where @xmath44 is a positive constant ( which is consistent with ) , one gets @xmath93 for every @xmath94 , provided @xmath95 is sufficiently large .
a similar condition is obtained also for more general forms of the spectral density at large @xmath73 , which lead to a @xmath77 behaviour of the self - energy at high temperatures .
thus , @xmath90 will be uniformly convergent so that the limit @xmath85 can be taken inside and , therefore , inside the integral @xmath96 .
consequently , it follows that the leading static contribution will agree to all orders with the result , got by calculating @xmath97 at vanishing energy and momentum .
this behavior is in agreement with the two - loop results given in eqs . and .
according to the simple arguments given following eqs . and , the above result should hold as well in the case of higher point functions .
we will now derive this property for the @xmath98-point green s functions , using a treatment which generalizes the previous method .
the spectral representation of the retarded @xmath98-point functions calculated in the imaginary time formalism may be written in the form @xcite where @xmath2 are real energy variables and @xmath100 . setting in , for example ,
@xmath101 positive and all other epsilons negative such that @xmath102 , defines the analytic @xmath6-th retarded function .
the spectral densities are the difference of two thermal wightman functions which , in the case of pure bosonic fields , are given by one may now consider the leading thermal contributions which arise from in the static limit ( when all @xmath106 ) which is well defined due to the analyticity properties of @xmath107 . in this case , to leading order in @xmath0 , one may next take the limits @xmath108 and proceed similarly to the previous analysis .
we then find a result which agrees with that obtained by setting in all external energies and momenta equal to zero : we have explicitly verified this relation at two - loops order , by calculating the leading static thermal contributions of the three - point functions in gauge theories .
thus , we conclude that to all orders in the static limit , the leading thermal contributions of retarded bosonic green s functions may be directly obtained by evaluating them at zero external energies and momenta . | we discuss the high temperature behavior of retarded thermal loops in static external fields .
we employ an analytic continuation of the imaginary time formalism and use a spectral representation of the thermal amplitudes .
we show that , to all orders , the leading contributions of static hard thermal loops can be directly obtained by evaluating them at zero external energies and momenta . |
knowledge on the density dependence of the symmetry energy is crucial to understand the structure of exotic nuclei , dynamics of heavy - ion collisions , and many important issues in nuclear astrophysics such as neutron star cooling and supernova explosive @xcite .
while great progress has been made to constrain the symmetry energy at low densities @xcite , the high - density behavior of the symmetry energy is divergent widely from interpreting fopi data @xcite .
@xmath4 ratio was found to be a sensitive probe to the high - density behavior of the symmetry energy by several transport models @xcite .
in fact for pion production , at lower beam energies effects of symmetry energy and in - medium effect may both become larger @xcite .
it is thus necessary to do a comparative study of effects of the in - medium @xmath0 cross section and the effects of the symmetry energy on pion production in heavy - ion collisions at lower beam energies . and these experimental studies will become possible at facilities that offer fast radioactive beams , such as nscl and frib in the us , fair in germany , or ribf in japan
in this study , we adopt the semi - classical transport model ibuu04 , in which the isospin - dependent initial neutron and proton density distributions of the projectile and target are obtained by using the skyrme - hartree - fock with skyrme @xmath5 ( sm ) force parameters @xcite . and an isospin- and momentum - dependent mean - field single nucleon potential is also used , i.e. , @xmath6 where @xmath7 for neutrons(protons ) , @xmath8 is the isospin asymmetry , and @xmath9 , @xmath10 denote neutron and proton densities , respectively .
the parameters @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 @xmath16 , and @xmath17 are all given in ref .
@xmath18 is the phase - space distribution function at coordinate * r * and momentum * p*. different @xmath19 parameters can be used to mimic different forms of the symmetry energy . in this model , reaction channels on pion production and absorption are @xmath20 where @xmath21 represents @xmath22 or @xmath23 resonances .
the experimental free - space nucleon - nucleon ( @xmath0 ) scattering cross section and the in - medium @xmath0 cross section can be used optionally .
for the later , we use the isospin - dependent in - medium @xmath0 elastic cross section , which is from the scaling model according to nucleon effective masses @xcite @xmath24 where @xmath25 and @xmath26 are the reduced masses of the colliding nucleon pair in free space and medium , respectively . and the effective mass of nucleon in isospin asymmetric nuclear matter is given by @xmath27 from the definition and eq .
( 1 ) , it is seen that the effective mass depends not only on density and asymmetry of nuclear matter , but also the momentum of nucleon @xcite . for the inelastic cross section we use the experimental data from free space @xmath0 collision since
the in - medium inelastic @xmath0 cross section is still very controversial .
the total and differential cross sections for all other particles are taken either from experimental data or obtained by using the detailed balance formula .
the isospin dependent phase - space distribution functions of the particles involved are solved by using the test - particle method numerically .
the isospin - dependence of pauli blockings for fermions is also considered .
[ t ] as shown in fig .
[ density ] , the maximal baryon densities reached in @xmath1 collisions are about 1.5@xmath282 times saturation density at @xmath29=100 , 200 and 300 mev / nucleon .
we can also see that the maximal baryon density reached increase with incident beam energy .
however , existing time of supradensity matter becomes shorter with increase of beam energy .
[ t ] figure [ ratiobeam ] shows effects of in - medium @xmath0 elastic scattering cross section on the @xmath4 ratio in central collision @xmath1 at @xmath29=100 , 150 , 200 , 250 , and 300 mev / nucleon , respectively . from fig .
[ ratiobeam ] we can see that the value of @xmath4 ratio decreases with increase of beam energy , which is consistent with that in ref .
this is partially because the production of pions is from repeated nucleon - nucleon collisions at higher beam energies , i.e. , a neutron converts a proton by producing @xmath30 and subsequent collisions of that proton can convert again to neutron by producing @xmath31 .
more interestingly , one can see that the effects of in - medium @xmath0 elastic cross section on the value of @xmath4 ratio become larger and larger with decrease of beam energy . at lower beam energy
100 mev / nucleon , effects of in - medium @xmath0 elastic cross section on the value of @xmath4 ratio can reach about 40% .
when one changes @xmath0 elastic cross section , total @xmath0 collision number would also changes accordingly .
as there is certain probability of inelastic process in total @xmath0 collision , @xmath0 inelastic collision is thus affected by @xmath0 elastic cross section . at high beam energy ,
@xmath0 inelastic processes may be more than elastic processes .
so elastic @xmath0 cross section should have small effects on pion production .
but at low beam energy , pion production is via many @xmath0 scatterings , large number of @xmath0 elastic scatterings increases the whole @xmath0 scatterings , since there is certain probability of inelastic process in @xmath0 collision , pion production should be affected by @xmath0 elastic scatterings .
[ t ] [ htb ] to see more clearly how the in - medium @xmath0 cross section affect @xmath4 ratio and effect of the symmetry energy on pion production , we plot evolution of pion meson production with different @xmath0 cross sections and symmetry energies .
shown in figure [ pion - t1 ] and figure [ pion - t2 ] are time evolutions of @xmath30 and @xmath31 mesons with different symmetry energies ( x=0 , x=1 ) and different @xmath0 elastic scattering cross sections ( in - medium , free ) in the central collision @xmath1 at beam energies of 100 and 200 mev / nucleon , respectively . in both fig .
[ pion - t1 ] and fig .
[ pion - t2 ] , we can see that the effects of in - medium @xmath0 elastic scattering cross section on the @xmath31 production is larger than that of @xmath30 production . however , the effects of symmetry energy on the @xmath31 production is smaller than that of @xmath30 production .
the reason of in - medium effects on the @xmath31 production is larger than that of @xmath30 production is that the reduction factor @xmath32 of @xmath33 ( proton - proton ) pair is smaller than that of @xmath34 ( neutron - neutron ) pair @xcite as well as @xmath33 @xmath35 collision mainly produce @xmath31 ( @xmath30 ) . and due to coulomb actions among protons , @xmath31 production is also less sensitive to the symmetry energy . in figure [ pion - t1 ] and
figure [ pion - t2 ] , we can also see that for @xmath30 production , effects of in - medium @xmath0 cross section are smaller than that of symmetry energy whereas for @xmath31 production , effects of in - medium @xmath0 cross section are obviously larger than that of symmetry energy . and with increase of beam energy , effects of in - medium @xmath0 cross section become smaller .
[ t ] to see more clearly effects of in - medium @xmath0 elastic scattering cross section and symmetry energy on @xmath4 ratio , in fig .
[ pionratio - t ] , we show time evolution of @xmath4 ratio with different symmetry energies and @xmath0 elastic scattering cross sections at beam energies of 100 and 200 mev / nucleon , respectively .
we can clearly see that effects of in - medium @xmath0 cross section on the value of @xmath4 ratio are about 2 times larger than that of the symmetry energy at incident beam energy of 100 mev / nucleon .
however , at incident beam energy of 200 mev / nucleon , effects of in - medium @xmath0 cross section on the value of @xmath4 ratio are almost equal to that of the symmetry energy .
[ t ] [ t ] in order to reduce the effects of in - medium and retain effects of the symmetry symmetry on @xmath4 ratio , we calculate double ratio of @xmath4 using two reaction systems of neutron rich and neutron poor of same isotopes @xcite . shown in fig .
[ doubleratio - t ] is time evolution of the double @xmath4 ratio from @xmath1 and @xmath36 .
it is seen that effects of the in - medium @xmath0 cross section on the double @xmath4 ratio is about 30% , but effects of the symmetry energy is about 42% .
however , from the top panel of fig .
[ pionratio - t ] , we see that effects of the in - medium @xmath0 cross section on the @xmath4 ratio is about 50% , but effects of the symmetry energy is about 25% .
thus the double ratio of @xmath4 from two reaction systems of neutron rich and neutron poor of same isotopes can indeed reduce uncertainties from in - medium properties of hadrons and keep the effects of symmetry energy . while from fig .
[ doubleratio - t ] , we can see that even we use the double ratio of observable , effects of in - medium @xmath0 cross section are still very obvious .
besides theoretical efforts @xcite and advanced method such as photon emission in heavy - ion collisions @xcite , we also simulated the double ratio of @xmath2 from two reaction systems ( with the same mass number of system but different isotopes ) @xmath1 and @xmath3 .
interestingly , from fig .
[ doubleratio - t1 ] one sees that effects of symmetry energy are kept but the effects of in - medium @xmath0 elastic scattering cross section are almost fully cancelled out .
this is because the latter two neutron - rich and neutron - poor reaction systems have the same baryon number , evolutions and distributions of baryon density in the two reactions are almost the same .
in summary , we studied effects of in - medium @xmath0 cross section and effects of symmetry energy on @xmath4 ratio at lower incident beam energies .
we find that at lower incident beam energy , for @xmath4 ratio , effects of in - medium @xmath0 cross section are larger than that of symmetry energy .
the double ratio of @xmath4 from reaction systems of neutron rich and neutron poor of same isotopes can not fully cancel out the effects of in - medium @xmath0 cross section .
however , the double ratio of @xmath4 from reaction systems of neutron rich and neutron poor of different isotopes but same mass number of reaction system almost fully cancels out the effects of in - medium @xmath0 cross section .
the authors are grateful to the c3s2 computing center in huzhou teachers college for calculation support .
this work is supported by the national natural science foundation of china ( 11375239 , 11175219 , 11435014 ) , the 973 program of china ( no .
2007cb815004 ) , the knowledge innovation project(kjcx2-ew - n01 ) of chinese academy of sciences .
chen , and c.m .
ko , phys .
rep . * 464 * , 113(2008 ) .
v. baran , m. colonna , v. greco , m. di toro , phys .
rep . * 410 * , 335(2005 ) .
lattimer , m. prakash , science * 304 * , 536(2004 ) .
a.w . steiner , m. prakash , j.m .
lattimer , et al .
, phys . rep . *
411 * , 325(2005 ) .
tsang , yingxun zhang , p. danielewicz , m. famiano , zhuxia li , w.g .
lynch , a.w .
steiner , phys .
rev . lett . * 102 * , 122701 ( 2009 ) .
tsang , z. chajecki , d. coupland , p. danielewicz , f. famiano , r. hodges , m.kilburn , f. lu , w.g .
lynch , j. winkelbauer , m. youngs , yingxun zhang , prog .
phys . * 66 * , 400 ( 2011 ) .
m. b. tsang , j. r. stone , f. camera , p. danielewicz , s. gandolfi , k. hebeler , c. j. horowitz , jenny lee , w. g. lynch , z. kohley , r. lemmon , p. moller , t. murakami , s. riordan , x. roca - maza , f. sammarruca , a. w. steiner , i. vida@xmath37a , s. j. yennello , phys .
c * 86 * , 015803 ( 2012 ) .
zhang , z. li , k. zhao , h. liu , m. tsang , nuclear science and techniques * 24 * , 050503 ( 2013 ) .
jeong , s.h .
lee , nuclear science and techniques * 24 * , 050506 ( 2013 ) .
guo , g.c .
yong , y.j .
wang , q. li , h.f .
zhang , w. zuo , arxiv:1404.7217 ( 2014 ) .
xiao , b.a .
chen , g.c .
yong , m. zhang , phys . rev
. lett . * 102 * , 062502 ( 2009 )
. z.q . feng , g.m .
jin , phys .
b * 683 * , 140 ( 2010 ) .
p. russotto , p.z .
wu , m. zoric , m. chartier , y. leifels , r.c .
lemmon , q. li , j. ukasik , a. pagano , p. pawowski , w. trautmann , phys .
b * 697 * , 471 ( 2011 ) .
xie , f.s .
zhang , nuclear science and techniques * 24 * , 050502 ( 2013 ) .
cozma , y. leifels and w. trautmann , q. li , p. russotto , phys .
c * 88 * , 044912 ( 2013 )
yong , b.a .
li , l.w . chen and w. zuo ,
c * 73 * , 034603 ( 2006 ) .
li , phys .
* 88 * , 5296 ( 2002 ) .
t. gaitanos , m. di toro , s. typel , et al .
a * 732 * , 24 ( 2004 ) .
li , s. soff , m. bleicher , and h. stcker , phys . rev .
c * 72 * , 034613 ( 2005 ) .
y. gao , g.c .
yong , y.j .
wang , q.f .
li , w. zuo , phys .
c * 88 * , 057601 ( 2013 ) .
f. zhang , y. liu , g.c .
yong , and w. zuo , chin .
phys . lett .
* 29 * , 052502 ( 2012 ) .
j. xu , l.w .
chen , c.m .
ma , phys .
c * 87 * , 067601 ( 2013 ) .
j. friedrich and p.g .
reinhard , phys .
c * 33 * , 335 ( 1986 ) .
das et al . , phys .
c * 67 * , 034611 ( 2003 ) ; b.a .
li et al . , nucl .
a * 735 * , 563 ( 2004 ) .
j. w. negele and k. yazaki , phys .
47 * , 71 ( 1981 ) .
v.r . pandharipande and s. c. pieper , phys .
c * 45 * , 791 ( 1992 ) .
f. sammarruca , eur .
j. * a50 * , 22 ( 2014 ) .
li and l. w. chen , phys .
c * 72 * , 064611 ( 2005 ) .
yong , eur .
j. * a46 * , 399 ( 2010 ) .
yong , w. zuo , x.c .
zhang , phys .
b * 705 * , 240 ( 2011 ) . | based on the isospin - dependent boltzmann - uehling - uhlenbeck(ibuu04 ) transport model , we explored effects of in - medium @xmath0 elastic scattering cross section and nuclear symmetry energy on the sub - threshold pion production in @xmath1 reaction .
we find that with decrease of incident beam energy , effects of in - medium @xmath0 elastic scattering cross section on the @xmath2 ratio are larger than that of the symmetry energy although the latter may be also larger .
while keeping the effect of symmetry energy , the double ratio of @xmath2 from neutron - rich and neutron - poor reaction systems ( with the same mass number of system ) @xmath1 and @xmath3 almost fully cancels out the effects of in - medium @xmath0 elastic scattering cross section . |
the @xmath0-nucleus interaction near threshold is strongly attractive and absorptive as suggested by fits to the strong - interaction shifts and widths of @xmath2-atom levels @xcite .
global fits yield ` deep ' optical potentials with re @xmath3 mev @xcite , whereas more theoretically inclined studies that fit the low - energy @xmath4 reaction data , including the @xmath5 quasi - bound state @xmath6 as input for constructing density dependent optical potentials , suggest relatively ` shallow ' potentials with re @xmath7 mev @xcite .
the issue of depth of the attractive @xmath0-nucleus potential is discussed in section [ sec : pot ] .
paradoxically , due to the strong ( absorptive ) i m @xmath8 , relatively narrow @xmath2 deeply bound _
atomic _ states are expected to exist @xcite , independently of the size of re @xmath8 .
figure [ fig : kpb ] from ref .
@xcite shows on the left - hand side a calculated spectrum of @xmath2 atomic states in @xmath9pb where , in particular , all the circular states below the @xmath10 state are not populated by x - ray transitions due to the strong @xmath2-nuclear absorption , and on the right - hand side it demonstrates saturation of the @xmath11 atomic - state width as a function of the absorptivity parameter i m @xmath12 of @xmath8 .
the physics behind is that a strong i m @xmath8 acts repulsively , suppressing the _ atomic _ wavefunction in the region of overlap with i m @xmath8 .
the calculated width of the ` deeply bound ' atomic @xmath13 and @xmath11 is less than 2 mev , also confirmed by the calculation of ref .
@xcite , calling for experimental ingenuity how to form these levels selectively by a non - radiative process @xcite .
this saturation mechanism does not hold for @xmath0-nuclear states which retain very good overlap with the potential .
hence , the questions to ask are ( i ) whether it is possible at all to bind _ strongly _ @xmath0 mesons in nuclei , and ( ii ) are such quasi - bound states sufficiently narrow to allow observation and identification ?
the first question was answered favorably by nogami @xcite as early as 1963 arguing that the @xmath14 system could acquire about 10 mev binding in its @xmath15 state .
yamazaki and akaishi , within a single - channel @xmath14 calculation @xcite , reported a binding energy @xmath16 mev and width @xmath17 mev ; however , the recent coupled - channel @xmath18 faddeev calculation for @xmath14 by shevchenko et al .
@xcite finds a substantially broader @xmath15 state ( @xmath19 mev , @xmath20 mev ) .
the current experimental and theoretical interest in @xmath0-nuclear bound states was triggered back in 1999 by the suggestion of kishimoto @xcite to look for such states in the nuclear reaction @xmath21 in flight , and by akaishi and yamazaki @xcite who suggested to look for a @xmath22 @xmath5 state bound by over 100 mev for which the main @xmath23 decay channel would be kinematically closed . some controversial evidence for relatively narrow states was presented initially in @xmath24 and @xmath25 reactions on @xmath26he ( kek - ps e471 ) @xcite but has just been withdrawn ( kek - ps e549/570 ) @xcite .
@xmath0-nuclear states were also invoked to explain few weak irregularities in the neutron spectrum of the @xmath27 in - flight reaction on @xmath28o ( bnl - ags , parasite e930 @xcite ) , but subsequent @xmath27 and @xmath21 reactions on @xmath29c at @xmath30 gev / c ( kek - ps e548 @xcite ) have not disclosed any peaks beyond the appreciable strength observed below the @xmath0-nucleus threshold .
ongoing experiments by the finuda spectrometer collaboration at da@xmath31ne , frascati , already claimed evidence for a relatively broad @xmath32 deeply bound state ( @xmath33 mev ) in @xmath34 reactions on li and @xmath29c , by observing back - to - back @xmath35 pairs from the decay @xmath36
@xcite , but these pairs could more naturally arise from conventional absorption processes at rest when final - state interaction is taken into account @xcite .
indeed , the @xmath37 reaction on @xmath38li has also been recently observed @xcite .
another recent claim for a very narrow and deep @xmath32 state ( @xmath39 mev , @xmath40 mev ) is also based on observing decay @xmath35 pairs , using @xmath41 annihilation data on @xmath26he from the obelix spectrometer at lear , cern @xcite .
one can not rule out that the @xmath35 pairs assigned in the above analyses to @xmath14 decay in fact result from nonmesonic decays of different clusters , say the @xmath22 @xmath5 quasi - bound state .
a definitive identification may only be reached through a fully exclusive formation analysis , such as the one scheduled for j - parc @xcite : @xmath42 finally , preliminary evidence for the @xmath22 @xmath5 state ( @xmath43 mev , @xmath44 mev ) has been recently presented by the finuda collaboration on @xmath38li by observing back - to - back @xmath45 pairs @xcite .
it is clear that the issue of @xmath0 nuclear states is far yet from being experimentally resolved and more dedicated , systematic searches are necessary .
it is interesting then to study theoretically @xmath0 nuclear quasi - bound states in the range of binding energy @xmath46 mev , in particular the width anticipated for such deeply bound states .
this is described in section [ sec : rmf ] within the relativistic mean field ( rmf ) model for a system of nucleons and one @xmath0 meson interacting through the exchange of scalar ( @xmath47 ) and vector ( @xmath48 ) boson fields which are treated in the mean - field approximation @xcite .
the born approximation for @xmath8 due to the leading - order tomozawa - weinberg ( tw ) vector term of the chiral effective lagrangian @xcite yields a sizable attraction : @xmath49 for @xmath50 @xmath51 , where @xmath52 mev is the pseudoscalar meson decay constant . iterating the tw term plus next - to - leading - order terms , within an _ in - medium _ coupled - channel approach constrained by the @xmath53 data near the @xmath54 threshold , roughly doubles this @xmath0-nucleus attraction .
it is found ( e.g. ref .
@xcite ) that the @xmath6 quickly dissolves in the nuclear medium at low density , so that the repulsive free - space scattering length @xmath55 , as function of @xmath56 , becomes _ attractive _ well below @xmath57 .
since the attractive @xmath58 @xmath59 is only weakly density dependent , the resulting in - medium @xmath54 isoscalar scattering length @xmath60 ) translates into a strongly attractive @xmath8 : @xmath61 however , when @xmath8 is calculated _ self consistently _ , namely by including @xmath8 in the propagator @xmath62 used in the lippmann - schwinger equation determining @xmath63 , one obtains re @xmath64 mev @xcite .
the main reason for this weakening of @xmath8 , approximately back to that of eq .
( [ eq : chiral ] ) , is the strong absorptive effect which @xmath8 exerts within @xmath62 to suppress the higher born terms of the @xmath54 tw potential .
the @xmath2-atom data used in global fits @xcite span a range of nuclei from @xmath65li to @xmath66u , with 65 level - shift and -width data points . figure [ fig : v ] shows fitted @xmath0-nucleus potentials for @xmath67ni and @xmath9pb , for a @xmath68 potential with a complex strength @xmath69 , and for two density - dependent potentials marked by dd and f , also fitted to the same data . for the real part ,
the depth of the @xmath68 potential is about @xmath70 mev , whereas the density - dependent potentials are considerably deeper , @xmath71 mev .
these latter potentials yield substantially lower @xmath72 values of 103 and 84 , respectively , than the value 129 for the @xmath68 potential .
we note that , although the two density - dependent potentials have very different parameterizations , the resulting potentials are quite similar . in particular , the shape of potential f departs appreciably from @xmath73 for @xmath74 , where the physics of the @xmath6 still plays some role .
the density dependence of the potential f is qualitatively similar to that of eq .
( [ eq : trho ] ) , by far providing the best fit ever reported for a global @xmath2-atom data fit .
additional considerations for estimating @xmath8 are listed below .
\(i ) qcd sum - rule estimates @xcite for vector ( v ) and scalar ( s ) self - energies : @xmath75 where @xmath76 is the strange - quark ( current ) mass .
the factor 1/2 in eq .
( [ eq : qcdv ] ) is due to the one nonstrange antiquark in the @xmath0 out of two possible , and the minus sign is due to @xmath77-parity going from quarks to antiquarks .
this rough estimate gives then @xmath78 mev .
the qcd sum - rule approach essentially refines the mean - field argument @xcite @xmath79 where the 1/3 factor is again due to the one nonstrange antiquark in the @xmath0 , but here with respect to the three nonstrange quarks of the nucleon .
\(ii ) the ratio of @xmath80 production cross sections in nucleus - nucleus and proton - nucleus collisions near threshold , measured by the kaon spectrometer ( kaos ) at sis , gsi , gives clear evidence for attractive @xmath8 , estimated @xcite as @xmath81 mev by relying on buu transport calculations normalized to the value @xmath82 mev .
since the buu calculations apparently disregard @xmath83 absorption , a deeper @xmath8 may follow when nonmesonic absorption is included .
in this model , expanded in ref .
@xcite , the ( anti)kaon interaction with the nuclear medium is incorporated by adding to @xmath84 the lagrangian density @xmath85 : @xmath86 the covariant derivative @xmath87 describes the coupling of the ( anti)kaon to the vector meson @xmath88 .
the coupling of the ( anti)kaon to the isovector @xmath56 meson was neglected , a good approximation for the light @xmath89 nuclei .
the @xmath0 meson induces additional source terms in the equations of motion for the meson fields @xmath47 and @xmath90 .
it thus affects the scalar @xmath91 and the vector @xmath92 potentials which enter the dirac equation for nucleons , and this leads to rearrangement or polarization of the nuclear core .
furthermore , in the klein - gordon equation satisfied by the @xmath0 , the scalar @xmath93 and the vector @xmath94 potentials become _ state dependent _ through the _ dynamical _ density dependence of the mean - field potentials @xmath95 and @xmath96 , as expected from a rmf calculation .
the @xmath0 potential was made complex by adding @xmath97 in a @xmath68 form fitted to the @xmath2 atomic data @xcite . from phase - space considerations , @xmath97
was then multiplied by a suppression factor , taking into account the binding energy of the antikaon for the initial decaying state , and assuming two - body final - state kinematics for the decay products in the @xmath98 mesonic modes and in the @xmath99 nonmesonic modes .
the coupled system of equations was solved self - consistently .
to give rough idea , whereas the static calculation gave @xmath100 mev for the @xmath2 @xmath13 state in @xmath29c , using the values @xmath101 corresponding to the @xmath2-atom fit , the dynamical calculation gave @xmath102 mev for this same state . in order to produce different values of binding energies ,
@xmath103 and @xmath104 were varied in given intervals of physical interest .
beginning approximately with @xmath29c , the following conclusions may be drawn : \(i ) for given values of @xmath105 , the @xmath0 binding energy @xmath106 saturates , except for a small increase due to the coulomb energy ( for @xmath2 ) .
\(ii ) the difference between the binding energies calculated dynamically and statically , @xmath107 , is substantial in light nuclei , increasing with @xmath106 for a given value of @xmath108 , and decreasing monotonically with @xmath108 for a given value of @xmath106 .
it may be neglected only for very heavy nuclei .
the same holds for the nuclear rearrangement energy @xmath109 which is a fraction of @xmath107 .
\(iii ) the width @xmath110 decreases monotonically with @xmath108 , according to fig .
[ fig : gamma ] which shows calculated widths @xmath111 as function of the binding energy @xmath112 for @xmath13 states in @xmath113c and @xmath114ca , using the nonlinear nl - sh version @xcite of the rmf model , and in @xmath115pb using the nl - tm1 version @xcite .
the dotted line shows the static ` nuclear - matter ' limit corresponding to the @xmath2-atom fitted value @xmath116 fm and for @xmath117 @xmath51 , using the same phase - space suppression factor as in the ` dynamical ' calculations .
it is clearly seen that the functional dependence @xmath118 follows the shape of the dotted line .
this dependence is due primarily to the binding - energy dependence of the suppression factor @xmath119 which falls off rapidly until @xmath120 mev , where the dominant @xmath121 gets switched off , and then stays rather flat in the range @xmath122 mev where the width is dominated by the @xmath83 absorption modes .
the widths calculated in this range are considerably larger than given by the dotted line ( except for pb in the range @xmath123 mev ) due to the dynamical nature of the rmf calculation , whereby the nuclear density is increased by the polarization effect of the @xmath2 .
adding perturbatively the residual width neglected in this calculation , partly due to the @xmath124 secondary mesonic decay channel , a representative value for a lower limit @xmath1 mev holds in the binding energy range @xmath122 mev .
figure [ fig : rho ] shows on the left - hand side the calculated average nuclear density @xmath125 as a function of @xmath112 for @xmath2 nuclear @xmath13 states across the periodic table . in the light @xmath2 nuclei , @xmath126 increases substantially with @xmath112 to values about 50% higher than without the @xmath0 .
the increase of the central nuclear densities is even bigger , as demonstrated on the right - hand side for a family of @xmath13 @xmath2 states in @xmath127pb , but is confined to a small region subsiding almost completely by @xmath128 fm , well within the nuclear volume . as a result ,
@xmath129 on the left - hand side of fig .
[ fig : rho ] is only weakly enhanced as function of @xmath112 .
in this talk i reviewed the phenomenological and theoretical evidence for a substantially attractive @xmath0-nucleus interaction potential , from a ` shallow ' potential of depth @xmath130 mev to a ` deep ' potential of depth @xmath71 mev at nuclear - matter density .
i then reported on recent _ dynamical _ calculations @xcite for deeply quasi - bound @xmath2 nuclear states across the periodic table .
substantial polarization of the core nucleus was found in light nuclei , but the ` high ' densities reached are considerably lower than those found in the few - body calculations due to akaishi , yamazaki and collaborators @xcite .
an almost universal dependence of @xmath0 widths on the binding energy was found , for a given nucleus , reflecting the phase - space suppression factor on top of the increase provided by the density of the compressed nuclear core .
the present results place a lower limit @xmath1 mev which is particularly useful for @xmath0 binding energies exceeding 100 mev .
i wish to thank my collaborators eli friedman , ji mare and nina shevchenko , as well as tullio bressani , evegny drukarev , tomofumi nagae and wolfram weise for stimulating discussions , and lauro tomio and the co - organizers of fb18 for their kind hospitality and support .
special thanks go to wolfram weise for his kind hospitality at the tu munich where this report was prepared under the support of the alexander von humboldt foundation .
this work is supported in part by the israel science foundation , jerusalem , grant 757/05 . | experimental evidence for @xmath0-nuclear quasi - bound states is briefly reviewed .
theoretical and phenomenological arguments for and against the existence of such states are considered , based on constructing @xmath0-nuclear optical potentials from various sources .
results of rmf calculations that provide a lower limit of @xmath1 mev on the width of @xmath0-nuclear quasi - bound states are discussed . |
learning representations that are invariant to irrelevant transformations of the input is an important step towards building recognition systems automatically .
invariance is a key property of some cells in the mammalian visual cortex .
cells in high - level areas of the visual cortex respond to objects categories , and are invariant to a wide range of variations on the object ( pose , illumination , confirmation , instance , etc ) .
the simplest known example of invariant representations in visual cortex are the complex cells of v1 that respond to edges of a given orientation but are activated by a wide range of positions of the edge .
many artificial object recognition systems have built - in invariances , such as the translational invariance of convolutional network @xcite , or sift descriptors @xcite .
an important question is how can useful invariant representations of the visual world be learned from unlabeled samples . in this paper
we introduce an algorithm for learning features that are invariant ( or robust ) to common image transformations that typically occur between successive frames of a video or statistically within a single frame . while the method is quite simple , it is also computationally efficient , and possesses provable bounds on the speed of inference .
the first component of the model is a layer of sparse coding .
sparse coding @xcite constructs a dictionary matrix @xmath0 so that input vectors can be represented by a linear combination of a small number of columns of the dictionary matrix .
inference of the feature vector @xmath1 representing an input vector @xmath2 is performed by finding the @xmath1 that minimizes the following energy function @xmath3 where @xmath4 is a positive constant .
the dictionary matrix @xmath0 is learned by minimizing @xmath5 averaged over a set of training samples @xmath6 , while constraining the columns of @xmath0 to have norm 1 .
the first idea of the proposed method is to accumulate sparse feature vectors representing successive frames in a video , or versions of an image that are distorted by transformations that do not affect the nature of its content .
@xmath7 where the sum runs over the distorted images @xmath8 .
the second idea is to connect a second sparse coding layer on top of the first one that will capture dependencies between components of the accumulated sparse code vector .
this second layer models vector @xmath9 using an _ invariant code _
@xmath10 , which is the minimum of the following energy function @xmath11 where @xmath12 denotes the @xmath13 norm of @xmath10 , @xmath14 is a matrix , and @xmath15 is a positive constant controlling the sparsity of @xmath10 . unlike with traditional sparse coding , in this method
the dictionary matrix interacts _ multiplicatively _ with the input @xmath9 . as in traditional sparse coding , the matrix @xmath14 is trained by gradient descent to minimize the average energy for the optimal @xmath10 over a training set of vectors @xmath9 obtained as stated above .
the columns of @xmath14 are constrained to be normalized to 1 .
essentially , the matrix @xmath14 will connect a component of @xmath10 to a set of components of @xmath9 if these components of @xmath9 co - occur frequently .
when a component of @xmath10 turns on , it has the effect of lowering the coefficients of the components of @xmath16 to which it is strongly connected through the @xmath14 matrix . to put it another way ,
if a set of components of @xmath9 often turn on together , the matrix @xmath14 will connect them to a component of @xmath10 . turning on
this component of @xmath10 will lower the overall energy ( if @xmath15 is small enough ) because the whole set of components of @xmath9 will see their coefficient being lowered ( the exponential terms ) .
hence , each unit of @xmath10 will connect units of that often turn on together within a sequence of images .
these units will typically represent distorted version of a feature . the energies ( [ eq_sparse_coding ] ) and ( [ eq_fulle ] ) can be naturally combined into a single combined model of @xmath1 and @xmath10 as explained in section [ sec_model ] . there
the second layer @xmath10 is essentially modulating sparsity of the first layer @xmath1
. single model of the image is more natural . for the invariance properties we did nt find much qualitative difference and since the former has provable inference bounds we presented the results for separate training .
however the a two layer model should capture the statistics of an image . to demonstrate this we compared the in - paining capability of one and two layer models and found that two layer model does better job .
for these experiments , the combined two layer model is necessary .
we also found that despite the assumptions of the fast inference are not satisfied for the two layer model , empirically the inference is fast in this case as well .
the first way to implement invariance is to take a known invariance , such as translational invariance in images , in put it directly into the architecture .
this has been highly successful in convolutional neural networks @xcite and sift descriptors @xcite and its derivatives .
the major drawback of this approach is that it works for known invariances , but not unknown invariances such as invariance to instance of an object .
a system that would discover invariance on its own would be desired .
second type of invariance implementation is considered in the framework of sparse coding or independent component analysis .
the idea is to change a cost function on hidden units in a way that would prefer co - occurring units to be close together in some space @xcite .
this is achieved by pooling units close in space together .
this groups different inputs together producing a form of invariance .
the drawback of this approach is that it requires some sort of imbedding in space and that the filters have to arrange themselves . in the third approach ,
rather then forcing units to arrange themselves , we let them learn whatever representations they want to learn and instead figure out which to pool together . in @xcite ,
this was achieved by modulating covariance of the simple units with complex units .
the fourth approach to invariance uses the following idea : if the inputs follow one another in time they are likely a consequence of the same cause .
we would like to discover that cause and therefore look for representations that are common for all frames .
this was achieved in several ways .
in slow feature analysis @xcite one forces the representation to change slowly . in temporal product network
@xcite one breaks the input into two representations - one that is common to all frames and one that is complementary . in @xcite
the idea is similar but in addition the complementary representation specifies movement . in the simplest instance of hierarchical temporal memory @xcite one forms groups based on transition matrix between states .
the @xcite is a structured model of video .
a lot of the approaches for learning invariance are inspired by the fact that the neo - cortex learns to create invariant representations .
consequently these approaches are not focused on creating efficient algorithms . in this paper , we given an efficient learning algorithm that falls into the framework of third and fourth approaches .
the basic idea is to modulate the sparsity of sparse coding units using higher level units that are also sparse .
the fourth approach is implemented by using the same higher level representation for several consecutive time frames . in the form our model is similar to that of @xcite but a little simpler . in a sense comparing our model to @xcite
is similar to comparing sparse coding to independent component analysis .
independent component analysis is a probabilistic model , whereas sparse coding attempts to reconstruct input in terms of few active hidden units .
the advantage of sparse coding is that it is simpler and easier to optimize .
there exist several very efficient inference and learning algorithms @xcite and sparse coding has been applied to a large number problems .
it is this simplicity that allows efficient training of our model .
the inference algorithm is closely derived from the fast iterative shrinkage - thresholding algorithm ( fista ) @xcite and has a convergence rate of @xmath17 where @xmath18 is the number of iterations .
the model described above comprises two separately trained modules , whose inference is performed separately .
however , one can devise a unified model with a single energy function that is conceptually simpler : @xmath19 given a set of inputs @xmath20 , the goal of training is to minimize @xmath21 .
we do this by choosing one input @xmath2 at a time , minimizing ( [ eq_fulle ] ) over @xmath1 and @xmath10 with @xmath0 and @xmath14 fixed , then fixing the resulting @xmath22,@xmath23 and taking step in a negative gradient direction of @xmath0 , @xmath14 ( stochastic gradient descent ) .
an algorithm for finding @xmath22,@xmath23 is given in section [ sec_theoretical ] .
it consists of taking step in @xmath1 and @xmath10 separately , each of which lowers the energy .
note : the @xmath24 functions in ( [ eq_fulle ] ) is different from that of the simple ( split ) model .
the reason is that , in our experiments , either @xmath10 units lower the sparsity of @xmath1 too much , not resulting in a sparse @xmath1 code or the units @xmath10 do not turn on at all .
we now describe a toy example that illustrates the main idea of the model @xcite .
the input , with @xmath25 , is an image patch consisting of a subset of the set of parallel lines of four different orientations and ten different positions per orientation .
however for any given input , only lines with the same orientation can be present , figure [ fig_lines]a ( different orientations have equal probability and for a given orientation a line of this orientation is present with probability 0.2 independently of others ) .
this is a toy example of a texture .
training sparse coding on this input results in filters similar to one in figure [ fig_lines]b .
we see that a given simple unit responds to one particular line .
the noisy filters correspond to simple units that are inactive - this happens because there are only 40 discrete inputs . in realistic data such as natural images
, we have a continuum and typically all units are used .
clearly , sparse coding can not capture all the statistics present in the data .
the simple units are not independent .
we would like to learn that that units corresponding to lines of a given orientation usually turn on simultaneously .
we trained ( [ eq_fulle ] ) on this data resulting in the filters in the figure [ fig_lines]b , c .
the filters of the simple units of this full model are similar to those obtained by training just the sparse coding .
the invariant units pool together simple units with filters corresponding to lines of the same orientation .
this makes the invariant units invariant to the pattern of lines and dependent only on the orientation .
only four invariant units were active corresponding to the four groups .
as in sparse coding , on a realistic data such as natural images , all invariant units become active and distribute themselves with overlapping filters as we will se below . ) .
a ) randomly selected input image patches .
the input patches are generated as follows .
pick one of the four orientations at random .
consider all lines of this orientation .
put any such line into the image independently with probability 0.2 .
b ) learned sparse coding filters .
a given active unit responds to a particular line .
c ) learned filters of the invariant units .
each row corresponds to an invariant unit .
the sparse coding filters are ordered according to the strength of their connection to the invariant unit .
there are only four active units ( arrows ) and each responds to a given orientation , invariant to which lines of a given orientation are present . ]
let us now discuss the motivation behind introducing a sequence of inputs ( @xmath26 ) in ( [ eq_fulle ] ) .
inputs that follow one another in time are usually a consequence of the same cause .
we would like to discover that cause .
this cause is something that is present at all frames and therefore we are looking for a single representation @xmath10 in ( [ eq_fulle ] ) that is common to all the frames .
another interesting point about the model ( [ eq_fulle ] ) is a that nonzero @xmath10 lowers the sparsity coefficient of units of @xmath1 that belong to a group making them more likely to become activated .
this means that the model can utilize higher level information ( which group is present ) to modulate the activity of the lower layer .
this is a desirable property for multi - layer systems because different parts of the system should propagate their belief to other parts . in our
invariance experiments the results for the unified model were very similar to the results of the simple ( split ) model .
below we show the results of this simple model because it is simple and because we provably know an efficient inference algorithm .
however in the section [ sec_theoretical ] we will revisit the full system , generalize it to an @xmath27-layer system , give an algorithm for training it , and prove that under some assumptions of convexity , the algorithm again has a provably efficient inference . in the final section we use the full system for in - paining and show that it generalizes better then a single layer system .
here we discuss how to find @xmath10 efficiently and give the numerical results of the paper .
the results for the full model ( [ eq_fulle ] ) were similar .
the advantage of ( [ eq_ei_gh ] ) compared to ( [ eq_fulle ] ) is that the fast iterative shrinkage - thresholding algorithm ( fista ) @xcite applies to it directly .
fista applies to problems of the form @xmath28 where : * @xmath29 is continuously differentiable , convex and lipschitz , that is @xmath30 .
the @xmath31 is the lipschitz constant of @xmath32 .
* @xmath33 is continuous , convex and possibly non - smooth the problem is assumed to have solution @xmath34 . in our case @xmath35 and @xmath36 which satisfies these assumptions ( @xmath14 is initialized with nonnegative entries which stay nonnegative during the algorithm without a need to force it ) .
this solution converges with bound @xmath37 where @xmath38 is the value of @xmath10 at the @xmath18-@xmath39 iteration and @xmath4 is a constant .
the cost of each iteration is @xmath40 where @xmath27 is the input size and @xmath41 is the output size .
more precisely the cost is one matrix multiplications by @xmath14 and by @xmath42 plus @xmath43 cost .
we used the back - tracking version of the algorithm to find @xmath44 which contains a fixed number of @xmath40 operations ( independent of desired error ) .
it is a standard knowledge and easy to see that the algorithm applies to the sparse coding ( [ eq_sparse_coding ] ) as well .
the input to the network was prepared as follows .
we converted all the images of the berkeley data - set into gray - scale images .
we locally removed the mean for each pixel by subtracting a gaussian - weighted average of the nearby pixels .
the width of the gaussian was @xmath45 pixels .
then , we locally normalized the contrast by dividing each pixel by gaussian - weighted standard deviation of the nearby pixels ( with a small cutoff to prevent blow - ups ) .
the width of the gaussian was also @xmath45 pixels .
then , we picked a @xmath46 window in the image and , for a randomly chosen direction and magnitude , we moved it for @xmath47 frames and extracted them .
the magnitude of the displacement was random in the range of @xmath48 pixels .
a very large collection of such triplets of frames was extracted .
we trained the sparse coding algorithm with @xmath49 code units in @xmath1 on each individual frame ( not on the @xmath50 concatenated frames ) .
after training we found the sparse codes for each frame .
there were @xmath51 units in the layer of invariant units @xmath10 . for larger a system with @xmath52 simple units and @xmath49 invariant units ,
see the supplementary material .
are equivalent ) .
we see that invariant units typically learn to group together units with similar orientation and frequency .
there are few other types of filters as well .
the units in the middle and right panel correspond to each other and correspond to the units in the left panel reading panels left to right and then down . see the supplementary material for all the filters the system : @xmath53 input patches , @xmath52 simple units , @xmath49 invariant units . ]
the results are shown in the figure [ fig_rfu ] , see caption for description .
we see that many invariant cells learn to group together filters of similar orientation and frequency but at several positions and thus learn invariance with respect to translations .
however there are other types of filters as well .
remember that the algorithm learns statistical co - occurrence between features , whether in time or in space . ) .
left panel are responses of simple units trained with sparsity @xmath54 in ( [ eq_sparse_coding ] ) .
the right four panels are responses of invariant units trained with sparsities @xmath55 in ( 3 ) on the values of simple units .
the x - axis of each panel is the distance of the edge from the center of the image - the @xmath56 in ( [ eq_edge ] ) .
the y - axis is the orientation of each edge - the @xmath57 in ( [ eq_edge ] ) .
30 cells were chosen at random in each panel .
different colors correspond to different cells .
the color intensity is proportional to the response of the unit .
we see that sparse coding inputs respond to a small range of frequencies and positions ( the elongated shape is due to the fact that an edge of orientation somewhat different from the edge detector orientation sweeps the detector at different positions @xmath56 ) . on the other hand invariant cells
respond to edges of at similar range of frequencies but larger range of positions . at high sparsities
the response boundaries are sharp and response regions do nt overlap .
as we lower the sparsity the boundaries become more blurry and regions start to overlap .
@xmath58 was used in ( [ eq_edge ] ) .
other frequencies produced similar effect . ]
the values of the weights give us important information about the properties of the system .
however ultimately we are interested in how the system responds to an input .
we study the response of these units to a commonly occurring input - an edge .
specifically the inputs are given by the following function .
@xmath59 where @xmath60 is the position of a pixel from the center of a patch , @xmath56 a real number specifying distance of the edge from the center and @xmath57 is the orientation of the edge from the @xmath2 axis .
this is not an edge function , but a function obtained on an edge after local mean subtraction .
the responses of the simple units and the invariant units are shown in the figure .
[ fig_responses ] , see caption for description . as expected the sparse coding units respond to edges in a narrow range of positions and relatively narrow range of orientations .
invariant cells on the other hand are able to pool different sparse coding units together and become invariant to a larger range of positions .
thus the invariant units do indeed have the desired invariance properties .
note that for large sparsities the regions have clear boundaries and are quite sharp .
this is similar to the standard implementation of convolutional net , where the pooling regions are squares ( with clear boundaries ) .
it is probably more preferable to have regions that overlap as happens at lower sparsities since one would prefer smoother responses rather then jumps across boundaries .
in this section we return to the full model ( [ eq_fulle ] ) .
we generalize it to an @xmath27 layer system , give an inference algorithm and outline the proof that under certain assumptions of convexity the algorithm has the fast @xmath17 convergence of fista , there @xmath18 is the iteration number .
the basic idea of minimizing over @xmath61 in ( [ eq_fulle ] ) is to alternate between taking energy - lowering step in @xmath1 while fixing @xmath10 and taking energy - lowering step in @xmath10 while fixing @xmath1 .
note that both of the restricted problems ( problem in @xmath1 fixing @xmath10 and problem in @xmath10 fixing @xmath1 ) satisfy conditions of the fista algorithm .
this will allow us to take steps of appropriate size that are guaranteed to lower the total energy .
before that however , we generalize the problem , which will reveal its structure and which does not introduce any additional complexity . consider system consisting of @xmath27 layers with units @xmath62 in the @xmath63-@xmath39 layer with @xmath64
we define @xmath65 that is all the vectors @xmath62 concatenated .
we define two sets of functions .
let @xmath66 be continuously differentiable , convex and lipschitz functions .
there can be several such functions per layer , which is denoted by index @xmath15 .
let @xmath67 be continuous and convex functions , not necessarily smooth . for convenience
we define @xmath68 , @xmath69 , @xmath70 and @xmath71 .
we define the energy of the system to be @xmath72 where in the second equality we drop the @xmath15 from the notation for simplicity .
we will omit writing the @xmath15 for the rest of the paper .
the equation ( [ eq_fulle ] ) is a special case of ( [ eq_hierarchy ] ) with @xmath73 , @xmath70,@xmath74 , @xmath75 , @xmath76 , @xmath77 the problem in @xmath62 keeping other variables fixed satisfies the conditions of the fista algorithm .
we can define a step in the variable @xmath62 to be ( in analogy to @xcite eq .
( 2.6 ) ) : @xmath78 where the later equality holds if @xmath79 . here
sh is the shrinkage function @xmath80 . in the case where @xmath62 is restricted to be
nonnegative we need to use @xmath81 instead of the shrinkage function .
let us describe the algorithm for minimizing ( [ eq_hierarchy ] ) with respect to @xmath1 ( we will write it explicitly below ) .
in the order from @xmath82 to @xmath83 , take the step @xmath84 in ( @xmath85 ) .
repeat until desired accuracy .
the @xmath86 s have to be chosen so that @xmath87 this can be assured by taking @xmath88 where the later @xmath44 denotes the lipschitz constant of its argument .
otherwise , as used in our simulations , it can be found by backtracking , see below .
this will assure that each steps lowers the total energy ( [ eq_hierarchy ] ) and hence the overall procedure will keep lowering it . in fact
the step @xmath89 with such chosen @xmath44 is in some sense a step with ideal step size .
let us now write the algorithm explicitly : * hierarchical ( f)ista . *
take @xmath90 , some @xmath91 and @xmath92 .
set @xmath93 , @xmath94 , @xmath95 .
+ step k. ( @xmath96 ) .
+ loop a=1:n \ { backtracking \ { + .
find smallest nonnegative integer @xmath97 such that with @xmath98 @xmath99 set @xmath100 } + .
compute @xmath101 @xmath102 @xmath103 the algorithm described above is this algorithm with the choice @xmath104 in the second last line .
let us discuss the @xmath105 s . for single layer system ( @xmath106 )
@xmath104 choice is called ista and has convergence bound of @xmath107 .
the other choice of @xmath105 is the fista algorithm .
the convergence rate of fista is much better than that of ista , having @xmath108 in the denominator . for hierarchical system , the choice @xmath104 guarantees that each step lowers the energy .
the question is whether introducing the other choice of @xmath105 would speed up convergence to the fista convergence rate .
the trouble is that the in general the product @xmath109 is non - convex , which is the case for ( [ eq_fulle ] ) .
for example we can readily see that if the function has more then one local minima , this convergence would certainly not be guaranteed ( imagine starting at a non - minimal point with zero derivative ) .
the effect of @xmath105 is that of momentum and this momentum increases towards one as @xmath18 increases .
with such a large momentum the system is in a danger of `` running around '' .
it might be effective to introduce this momentum but to regularize it ( say bound it by a number smaller then one ) . in any case one can always use the algorithm with @xmath104 . in the special cases when all @xmath109 s are convex however , we give an outline of the proof that the algorithm converges with the fista rate of convergence . for this purpose
we define the full step in @xmath1 , @xmath110 , to be the result of the sequence of steps @xmath84 eq .
( [ eq_stepa ] ) from @xmath82 to @xmath83 .
that is we have @xmath111 : @xmath112 .
we assume that all the @xmath86 s are the same ( this is always possible by making all @xmath86 s equal the largest value ) .
the core of the proof is to show the lemma 2.3 of @xcite : * lemma 2.3 : * assume that @xmath113 , @xmath114 is continuously differentiable , lipschitz , @xmath115 is continuous , @xmath116 is convex and @xmath111 is defined by the sequence of @xmath89 s of ( [ eq_stepa ] ) as described above .
then for any @xmath117 @xmath118 the proof in @xcite shows that if the algorithm consists of applying the sequence of @xmath111 s and these @xmath111 s satisfy lemma 2.3 , then the algorithm converges with rate @xmath119 .
thus we need to prove lemma 2.3 .
we start with the analog of lemma 2.2 of ( @xcite ) .
* lemma 2.2 : * for any @xmath1 , one has @xmath120 if and only if there exist @xmath121 , the subdifferential of @xmath122 , such that @xmath123 this lemma follows trivially from the definition of @xmath124 as in @xcite .
* proof of lemma 2.3 : * define @xmath125 . from convexity
we have @xmath126 next we have the property ( [ eq_pproperty ] )
. however the @xmath127 should be primed ( @xmath128 ) because the @xmath127 has already been updated .
due to space limitations we wo nt write out all the calculations but specify the sequence of operations . the details are written out in the supplementary material .
we take the first term on the left side of ( [ eq_lemma23 ] ) , @xmath129 and express it in its terms ( [ eq_hierarchy ] ) .
then , replace the terms using the convexity equations and substitute @xmath130 s using the lemma 2.2 .
then we take the second term of the left side of ( [ eq_lemma23 ] ) , @xmath131 , again express it using ( [ eq_hierarchy ] ) , and use the inequalities ( [ eq_pproperty ] ) . putting it all together ,
all the gradient terms cancel and the other terms combine to give lemma 2.3 .
this completes the proof .
we introduced simple and efficient algorithm from learning invariant representation from unlabelled data .
the method takes advantage of temporal consistency in sequential image data . in the future we plan to use the invariant features discovered by the method to hierarchical vision architectures , and apply it to recognition problems
\1 ) we give details of the proof of lemma 2.3 .
+ 2 ) we show all of the invariant filters for system of : 20x20 patches input patches , 1000 simple units , 400 invariant units .
* lemma 2.3 : * assume that @xmath113 , @xmath114 is continuously differentiable , lipschitz , @xmath115 is continuous , @xmath132 is convex and @xmath111 is defined by the sequence of @xmath89 s in the paper . then for any @xmath117 @xmath118 * proof of lemma 2.3 : * define @xmath125 .
we first collect the inequalities that we will need .
+ from convexity we have @xmath133 next we have the property for step @xmath134 that guarantees that the energy is lowered in each step .
@xmath135 finally we have the lemma 2.2 @xmath136 now we can put these equations together .
the steps are : write out the left side of ( [ eq_lemma23 ] ) in terms of the definition of e. use inequalities ( [ eq_convex ] ) and ( [ eq_pproperty ] ) . eliminate @xmath137 s using ( [ eq_gamma ] ) . simplify . here
are the details : @xmath138 which is the formula ( [ eq_pproperty ] ) .
note that in the line 5 and in the first term of lines 6 we shifted @xmath63 by one .
this completes the proof .
[ [ simple - unit - and - invariant - unit - filters .- alpha0.5-beta0.3 ] ] simple unit and invariant unit filters .
@xmath54 , @xmath139 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | we propose a simple and efficient algorithm for learning sparse invariant representations from unlabeled data with fast inference . when trained on short movies sequences , the learned features are selective to a range of orientations and spatial frequencies , but robust to a wide range of positions , similar to complex cells in the primary visual cortex .
we give a hierarchical version of the algorithm , and give guarantees of fast convergence under certain conditions . |
protoplanetary disks begin as massive , gas - rich remnants of their parent interstellar molecular clouds , but rapidly evolve into tenuous , dusty disks within about 10 myr . these disks , called debris disks , differ from protoplanetary disks in that they are composed of secondary material recently generated by collisions between and evaporation of asteroids and comets .
planetesimal collisions presumably give rise to rocky planets , and therefore the younger debris disks ( @xmath2 myr ) provide valuable insight into the later stages of terrestrial planet formation .
debris disks are gas - poor , as evidenced by a general lack of sub - mm carbon monoxide emission ( e.g. * ? ? ?
however , small amounts of gas have been found in several debris disks , the most famous of which is @xmath3 pictoris ( e.g. * ? ? ?
* ; * ? ? ?
this primarily atomic gas must also be recently produced secondary material , as many of the species seen have short lifetimes in optically - thin environments ( for example , the short photoionization lifetime of ) .
as for the dust , the ultimate source of the gas is the destruction of planetesimals ( see @xcite for a fuller discussion ) .
the well - known 49 ceti system consists of a bright debris disk surrounding a single a1v star at a _ hipparcos _ distance of @xmath4 pc @xcite .
the stellar age estimated by comparison of the star s position on the h - r diagram to theoretical stellar evolutionary tracks is either @xmath5 myr from pre - main sequence tracks or @xmath6 myr from post - main sequence tracks @xcite .
recently , @xcite identified 49 cet as a co - moving member of the @xmath7 myr - old argus association @xcite .
spectral energy distribution ( sed ) fitting suggested that the 49 cet disk has two distinct components , a cold outer disk and a warmer inner one ( e.g. * ? ? ? * ) .
the inner dust has been spatially resolved in mid - ir thermal emission but no emission features associated with @xmath0m - sized grains are observed in the _ spitzer _ irs mid - ir spectrum of the system @xcite .
49 cet is one of two debris disks that show sub - mm co emission @xcite ; the other is hd21997 , whose co emission was recently discovered @xcite .
the co emission from 49 cet led to questions about its evolutionary status , possibly marking it as a rare late - stage protoplanetary disk that has dissipated most but not all of its primordial gas ( e.g. * ? ? ? * ) .
moderate gas drag at such a time may be important for establishing the final orbital characteristics of any terrestrial planets ( e.g. * ? ? ?
however , the relatively low fractional infrared luminosity of 49 cet ( @xmath8 ; * ? ? ?
* ) shows that the dust is optically thin , which is characteristic of a debris disk . on the strength of the system s relatively advanced @xmath7 myr age
, @xcite proposed that the observed co gas is not primordial but is rather coming from a massive belt of colliding comet - like planetesimals . at this time
, however , the evolutionary status of 49 cet s gas component is uncertain . in this paper , we present far - ir / sub - mm imaging and spectroscopy of 49 cet obtained with the _ herschel space observatory _ @xcite . in section 2 , the observations and data reduction are described .
analysis and results appear in section 3 : analysis of the spatially resolved 70 @xmath0 m image in section [ sub : image ] , all photometry in section [ sub : phot ] , spectroscopic analysis in section [ sub : spec ] , and simple gas mass estimates in section [ sub : gas_mass ] . in section 4
, we show the new 49 cet sed , present an improved model for the stellar spectrum , and describe our simple dust model fitting .
preliminary modeling of the gas component is discussed in section 5 .
an examination of the evolutionary status of the 49 cet disk material appears in section 6 .
finally , section 7 summarizes our primary findings and some planned future work .
49 ceti was observed with the far - ir pacs instrument @xcite , as part of the `` gas in protoplanetary systems '' ( gasps ) open time key programme ( e.g. * ? ? ?
* ; * ? ? ?
single pacs scan maps at 70 @xmath0 m and 160 @xmath0 m ( obsid 1342188485 ) were obtained on 2009 - 12 - 23 using the medium scan speed ( @xmath9 ) .
the maps consisted of 8 scan legs with @xmath10 lengths and @xmath11 cross - scan steps . the on - source exposure time for each map was 72 sec .
we also obtained three pacs spectroscopy observations in chop - nod mode .
the first was a rangespec observation that included 6 small wavelength ranges centered on the following transitions : co 73 @xmath0 m , h@xmath12o 79 @xmath0 m , co 90 @xmath0 m , [ ] 145 @xmath0 m , [ ] 158 @xmath0 m , h@xmath12o 180 @xmath0 m ( taken 2009 - 12 - 22 , obsid 1342188423 , 251 sec on - source time per range ) . the second was a linespec observation that covered two small wavelength ranges centered on the [ ] 63 @xmath0 m and dco@xmath13 190 @xmath0 m transitions ( taken 2009 - 12 - 22 , obsid 1342188424 , 370 sec on - source time per range ) .
the third was a deeper rangespec observation focused on the [ ] 158 @xmath0 m line ( taken 2011 - 07 - 08 , obsid 1342223790 , 670 sec on - source time ) .
all pacs data were calibrated with hipe v8.2 , using pipeline calibration scripts @xcite .
final scan maps for each wavelength were generated with two different pixel scales : one with @xmath14 pixels for use in resolved imaging analysis ( section [ sub : image ] ) and one with the native pixel scale of the pacs detectors ( @xmath15 at 70 @xmath0 m and @xmath16 at 160 @xmath0 m ) for use in photometry analysis ( section [ sub : phot ] ) . since 49
cet is bright at far - ir wavelengths , standard calibration of scan maps will lead to over - subtraction of the background during high - pass filtering .
therefore , a region around the source was masked before filtering .
the absolute calibration uncertainties for the 70 and 160 @xmath0 m scan maps are 2.64% and 4.15% , respectively . ]
pacs is an integral field spectrometer ( ifs ) with a @xmath17 array of spectral pixels ( spaxels ) , each @xmath18 in size .
all spectroscopy was obtained in chopnod mode to remove telescope and sky background emission . for each spectroscopic observation ,
25 spectra were extracted , one from each spaxel .
the pipeline calibration script applied a flux correction to each spectrum , accounting for the expected fraction of the point - spread function ( psf ) falling outside the spaxel .
spectra with two different pixel scales were produced : one corresponding to the native resolution of the instrument ( 2 pixels per resolution element , i.e. nyquist sampling ) and one with smaller pixels ( oversampled with 3 pixels per resolution element ) .
all analysis was performed on the nyquist - sampled data , which gave results with the highest signal - to - noise ( @xmath19 ) , as expected .
the oversampled spectra were used only to check results and for display purposes .
in addition to the pacs data , we made use of sub - mm 49 cet observations taken with the spire instrument @xcite .
the spire small scan maps at 250 @xmath0 m , 350 @xmath0 m , and 500 @xmath0 m ( obsid 1342236226 ) were acquired in january 2012 as part of a _ herschel _ open time program ( ot1_pabraham_2 , pi : p. brahm ) .
the data were calibrated using hipe v8.2 and the standard pipeline calibration script , producing maps with units of jy beam@xmath20 , which we converted to jy pixel@xmath20 before analysis .
the map pixel scales were @xmath21 , @xmath22 , and @xmath23 at 250 @xmath0 m , 350 @xmath0 m , and 500 @xmath0 m , respectively .
the estimated calibration uncertainty for the spire images is 7% . ] finally , we re - calibrated the low - resolution _
spitzer _
irs spectrum of 49 cet published in @xcite using the feps _ spitzer _ legacy project pipeline . as is typical
, there was a flux disjoint between the short - low and long - low portions of the spectrum .
we shifted the long - low portion to match the _
spitzer _
mips 24 @xmath0 m flux ( discussed in section [ sub : phot ] ) , which removed the disjoint between the two halves of the spectrum .
49 cet is significantly extended in the 70 @xmath0 m image compared to a similar image of the diskless psf reference star @xmath24 boo obtained using the same medium scan speed , as shown in the left - hand panel of figure [ fig : cet_70 ] .
this is the first time thermal emission from dust in the outer disk of 49 cet has been spatially resolved .
we began our analysis by fitting gaussians to the 49 cet and @xmath24 boo images ; the results appear in table [ tab : resolve ] .
we estimated the uncertainty in our full - width at half - maximum ( fwhm ) measurements by performing the same analysis on a different 70 @xmath0 m scan map of @xmath24 boo .
the fwhm of the best - fitting gaussian varied by @xmath25 between the two @xmath24 boo images . adopting this value as our fwhm uncertainty ,
49 cet is resolved along the major axis at the @xmath26 level .
the measured position angle ( pa ) of the disk major axis ( @xmath27 e of n ) is in reasonable agreement with the pa determined from spatially resolved sub - mm co contours ( @xmath28 e of n ; * ? ? ?
* ) . to estimate the true extent of the dust disk , we deconvolved the 49 cet image using the @xmath24 boo image as the reference psf , rotated to match the roll angle of the telescope at the time of the 49 cet observation .
the stellar contribution was removed from the 49 cet image before deconvolution , by subtracting the psf reference image scaled so that the total flux from @xmath24 boo matched the expected stellar flux from 49 cet .
since the expected 49 cet stellar flux is only @xmath29 of the actual measured flux ( see section [ sub : star ] ) ,
this had a negligible effect on the deconvolution .
we employed the richardson - lucy deconvolution method , conservatively limiting the number of iterations to five .
the reason for choosing this number will become apparent later .
the resulting deconvolved image appears in the right - hand panel of figure [ fig : cet_70 ] .
the deconvolved disk image shows no sign of a central clearing or any asymmetric structure .
the position angle of the major axis changed only slightly .
we re - fit a gaussian to the deconvolved 49 cet image ; the results are included in table [ tab : resolve ] . the half - width at half maximum ( hwhm ) along the major axis is @xmath30 au . in their modeling of the 49 cet gas and dust disk , @xcite found that they had to set the outer radius of the disk to 200 au .
while the agreement between that value and our hwhm is partly coincidental , it does indicate that the cold co gas observed in the sub - mm and the cold dust we observed here may be co - spatial . during the deconvolution procedure
, we found that we could make 49 cet significantly smaller than @xmath24 boo along the minor axis by iterating many times .
but no matter how many iterations were used ( up to @xmath31 ) , 49 cet was always larger than @xmath24 boo along the major axis .
we took this as an indication that the disk is significantly resolved along the major axis but not the minor axis .
therefore , we chose the number of deconvolution iterations such that the fwhm of the disk along the minor axis nearly matched that of @xmath24 boo .
this was accomplished in 5 iterations .
since the minor axis is not resolved , we only obtain a lower limit on the outer disk inclination of @xmath32 , assuming the disk is inherently circular .
this is broadly consistent with the high inclination derived from the sub - mm co contours ( @xmath33 ; * ? ? ?
49 cet is not significantly extended in the 160 @xmath0 m image or the spire images .
lccc major axis pa ( e of n ) & @xmath27 & & @xmath34 + major axis ( fwhm ) & @xmath35 & @xmath36 & @xmath37 + semi - major axis ( hwhm ) & 250 au & & 200 au + minor axis ( fwhm ) & @xmath38 & @xmath39 & @xmath40 + semi - minor axis ( hwhm ) & @xmath41 au & & @xmath42 au + inclination & @xmath43 & & @xmath43 to measure the total flux from 49 cet at 70 and 160 @xmath0 m , aperture photometry was performed on the scan maps with the native pixel scale .
the radii of the circular object apertures were @xmath23 and @xmath44 at 70 and 160 @xmath0 m , respectively . these are larger than recommended by the _ herschel _
pacs instrument control centre ( icc ) for aperture photometry of unresolved faint sources , since 49 cet is bright and the disk is spatially resolved at 70 @xmath0 m .
the apertures were chosen to be roughly 1.5 times the fwhm of the disk at that wavelength ( section [ sub : image ] ) , so that they encompass nearly all of the disk flux .
background subtraction was performed using the mean brightness in sky annuli @xmath45 from the star center for the 70 @xmath0 m image and @xmath46 for the 160 @xmath0 m image .
aperture corrections provided by the _
herschel _
pacs iccwere applied to the total fluxes , but color corrections were not applied . assuming background - limited imaging , the statistical uncertainties in the final pacs fluxes are given by @xmath47 where @xmath48 is the standard deviation of the pixels in the sky annulus , @xmath49 is the aperture correction , @xmath50 is the correlated noise correction(0.95 at 70 @xmath0 m , 0.88 at 160 @xmath0 m ) , @xmath51 is the number of pixels in the object aperture , and @xmath52 is the number of pixels in the sky annulus .
the absolute calibration uncertainties ( 2.64% at 70 @xmath0 m , 4.15% at 160 @xmath0 m ) were added in quadrature to the statistical uncertainties to give the total flux uncertainties .
the total 49 cet fluxes at 70 and 160 @xmath0 m are @xmath53 and @xmath54 .
we performed aperture photometry on the spire maps using circular apertures with radii of @xmath44 , @xmath55 , and @xmath56 , respectively . for each wavelength , the background was estimated in a sky annulus @xmath57 from the star center .
aperture corrections were applied and the fluxes color - corrected for a @xmath58 point - source spectrum . ]
the total 49 cet fluxes at 250 , 350 , and 500 @xmath0 m are @xmath59 , @xmath60 , and @xmath61 .
the flux uncertainties include statistical and calibration uncertainties ; since 49 cet is relatively bright , the latter dominate . for use in the 49 cet sed and to correct the _
spitzer _
irs long - low spectrum , we also calculated the continuum flux at 24 @xmath0 m from archival _ spitzer _
mips imaging taken as part of the mips - gto program disklegacy ( pi : g. rieke , aor # 21942016 ) .
the post - bcd image was downloaded from the spitzer heritage archive and aperture photometry performed using the aperture size and correction from @xcite .
we found @xmath62 , where the error is given by the @xmath63 absolute calibration uncertainty ] ( the statistical uncertainty is negligible ) .
in all the 49 cet pacs spectra , significant continuum or line emission appears only in the central spaxel .
we verified that the star was well - centered on the array during the spectroscopic observations ( shifts @xmath64 spaxel ) .
this was done by taking theoretical psfs , offsetting them on a virtual pacs ifs , and comparing the fraction of flux in each spaxel as a function of shift to the observed values . all further analysis discussed here
was performed on the spectra from the central spaxel , using the nyquist - sampled data .
only one emission line is detected : [ ] 158 @xmath0 m ( discussed below ) .
the spectra without significant emission lines were analyzed in the following manner .
for each spectrum , we least - squares fit a 1st - degree polynomial to the continuum .
the statistical flux uncertainties were then estimated by taking the standard deviation of the fluxes minus the continuum fit , in a range centered on the expected line position .
the width of the range was chosen so that roughly @xmath65 of the pixels were within @xmath66 of the continuum .
the continuum flux values at the expected line center wavelengths appear in table [ tab : spec ] .
the flux errors given are the statistical flux uncertainties and the absolute flux calibration uncertainties ( 11% at 63 @xmath0 m , 12% at other wavelengths ) added in quadrature .
upper limits on the total emission line fluxes were calculated by integrating the continuum - subtracted spectrum over a small wavelength range centered on the expected line wavelength and propagating the final statistical flux errors ; the wavelength range was @xmath67 the expected width of an unresolved emission line .
the results appear in table [ tab : spec ] .
a plot of the region around the undetected [ ] 63 @xmath0 m line appears in the left - hand panel of figure [ fig : spec ] ; this is the brightest line observed from protoplanetary disks with _
herschel _
( e.g. * ? ? ?
* ; * ? ? ?
the only line detected , [ ] 158 @xmath0 m , was seen in the original shallow spectrum , though it was not a significant detection in the data calibrated with earlier versions of hipe ( e.g. v4.2 ) .
therefore , we obtained a deeper follow - up spectrum to confirm the detection , shown in the right - hand panel of figure [ fig : spec ] .
for both the shallow and deep spectra , we analyzed the [ ] 158 @xmath0 m line in the following way .
we modeled the continuum plus line by least - squares fitting the sum of a 1st - degree polynomial and a gaussian .
then the statistical flux uncertainties were estimated by taking the standard deviation of the fluxes minus the best - fitting total model , in a range around the line .
this time , the width of the range was chosen to maximize the @xmath19 of the emission line ( @xmath68 the instrumental fwhm ) .
as before , the total emission line flux was integrated from the continuum - subtracted data ; the results appear in table [ tab : spec ] .
the integrated flux from the deep spectrum is within @xmath69 of the flux from the noisier shallow spectrum . to date
, 49 cet is one of only two disks observed with _
herschel _
that show emission but no .
the other is hd32297 , a bright @xmath70 myr - old debris disk system @xcite .
hd32297 was previously known to contain atomic gas , but no molecular gas has yet been detected @xcite .
lrcccc & 63.185 & @xmath71 & @xmath72 & 3.68 & + co j=36 - 35 & 72.843 & @xmath73 & @xmath74 & 4.96 & + o - h@xmath12o & 78.741 & @xmath75 & @xmath76 & 3.95 & + co j=29 - 28 & 90.163 & @xmath77 & @xmath78 & 3.08 & + & 145.535 & @xmath79 & @xmath80 & 2.06 & + shallow & 157.741 & @xmath81 & 7.23 & 2.22 & @xmath82
+ deep & & @xmath83 & 3.690 & 0.798 & @xmath84 + o - h@xmath12o & 179.527 & @xmath85 & @xmath86 & 4.06 & + dco@xmath13 & 189.570 & @xmath87 & @xmath88 & 4.42 & with only one carbon emission line detected , we are not able to measure the excitation temperature of the gas from our data , which is needed to calculate a model - independent total mass of ions from the integrated [ ] 158 @xmath0 m line flux .
however , we may determine a temperature - insensitive lower limit on the total mass , assuming the emission is optically thin . in this case , the mass is given by @xmath89 where @xmath90 is the wavelength of the line , @xmath91 is the observed integrated emission line flux , @xmath92 is the mass of an atom , @xmath93 is the distance from the emitting region to the observer , @xmath94 and @xmath95 designate the upper and lower energy levels involved in the transition , @xmath96 is the spontaneous transition probability , and @xmath97 is the fraction of atoms in the upper energy level . assuming local thermal equilibrium ( lte ) , @xmath97 is given by @xmath98 where @xmath99 is the angular momentum quantum number of the upper level , @xmath100 is the energy of the upper level , @xmath101 is the excitation temperature , and @xmath102 is the partition function for the given excitation temperature .
figure [ fig : gas_mass ] shows a plot of the mass as a function of the assumed excitation temperature
. the lower limit on the mass is @xmath103 , valid for @xmath101 between 1 k and 2000 k. this value is close to the total co mass calculated from the observed sub - mm co emission ( @xmath104 ; * ? ? ?
* ) . our upper limits on the co 73 and 90 @xmath0
m line fluxes do not provide useful checks on the co abundance or excitation temperature .
these lines arise from very high energy levels ( @xmath105 k , respectively ) and therefore are insensitive tracers of the bulk of the co gas . to determine the total mass of carbon atoms in the 49 cet disk , we need to know the ionization balance in the gas .
this may be calculated with complex thermochemical disk models , further discussed in section 5 .
here we make a simple estimate of the mass lower limit by assuming a plausible ionization fraction , measured in an analogous environment .
the radiation and density characteristics of the 49 cet circumstellar environment are similar to those of @xmath3 pic ; both a stars are surrounded by optically thin disks containing a similar amount of dust , judging from the systems fractional infrared luminosities .
both disks contain circumstellar gas , including co ( although the @xmath3 pic co is not yet detected in sub - mm emission , only uv absorption ; * ? ? ?
the measured column densities of neutral and first - ionized carbon in the @xmath3 pic gas are roughly equal , giving a carbon ionization fraction of @xmath106 @xcite , in agreement with the value predicted by photoionization calculations @xcite .
therefore , the rough lower limit on the total mass of carbon atoms in the 49 cet gas is @xmath107 .
if we further assume that the atomic gas has solar elemental abundances , we may estimate a lower limit on the total atomic gas mass by dividing the carbon mass limit by the solar carbon mass fraction ( 0.288% ; * ? ? ?
with all these assumptions ( optically thin [ ] emission , lte , carbon @xmath108 , 50% ionization fraction , and solar elemental abundances ) , the rough lower limit on the total atomic gas mass is @xmath109 . turning to the other undetected emission lines ,
we can not at present use them to calculate model - independent upper limits on the masses of other gases , due to a lack of precise information on the gas excitation temperatures .
examination of figure [ fig : gas_mass ] shows that the mass vs. excitation temperature curve changes slowly at high temperatures but shoots up at low temperatures ; this is a generic characteristic of such curves .
it is easy to set a mass lower limit that is valid for a wide range of temperatures , since it s controlled by the highest temperature considered , where the curve is changing slowly . setting a temperature - insensitive upper limit
is difficult , since it s controlled by the lowest temperature considered , where the curve is changing rapidly .
the situation is worse for the [ ] 63 @xmath0 m line , since the energy of the upper level ( @xmath110 k ) is much higher than it is for the [ ] 158 @xmath0 m line ( @xmath111 k ) .
therefore , the mass upper limit is extremely sensitive to the range of temperatures considered .
for example , assuming lte and changing the lowest excitation temperature considered from 10 k to 50 k decreases the mass upper limit by 6 orders of magnitude .
furthermore , the large energy separation for the levels involved in the [ ] 63 @xmath0 m line means that the levels are unlikely to be thermalized and assuming lte is not appropriate .
we have collected a wide range of unresolved photometry and spectroscopy of 49 cet ; our compilation of continuum fluxes at wavelengths from the optical to the millimeter appears in table [ tab : all_phot ] . in figure
[ fig : sed ] , we show the most complete sed to date .
it includes our new far - ir / sub - mm continuum fluxes from _ herschel _ ; the _ spitzer _ irs spectrum ; the _ spitzer _ mips 24 @xmath0 m flux ; and new near - ir photometry from _
wise_. some redundant fluxes given in table [ tab : all_phot ] were not used in the sed .
table [ tab : all_phot ] includes two unrefereed detections of 49 cet at sub - mm / mm wavelengths : one at 850 @xmath0 m from jcmt / scuba @xcite and one at 1.2 mm from iram @xcite . as noted by @xcite ,
the two fluxes can not be simultaneously fitted with any dust excess model that decreases with increasing wavelength in the sub - mm / mm .
it is not obvious which flux value is more accurate . in the absence of other sub - mm fluxes
, @xcite conservatively adopted the lower flux value from scuba @xcite .
the new spire fluxes allow us to reasonably constrain the sed fitting without relying on the ambiguous scuba and iram fluxes .
therefore , we avoided making an arbitrary choice between them by excluding both points from our new sed .
we eagerly await sensitive new sub - mm / mm observations of 49 cet with alma .
lccr@@xmath112lp2.2in[ht ] tycho-2 & johnson @xmath113 & 0.44 & 22.26 & 0.35 & @xcite + tycho-2 & johnson @xmath114 & 0.55 & 21.60 & 0.20 & @xcite + 2mass & @xmath115 & 1.24 & 10.18 & 0.19 & 2mass all - sky catalog of point sources + 2mass & @xmath116 & 1.65 & 6.30 & 0.13 & 2mass all - sky catalog of point sources + 2mass & @xmath117 & 2.16 & 4.373 & 0.081 & 2mass all - sky catalog of point sources + wise & w1 & 3.35 & 1.99 & 0.12 & @xcite + wise & w2 & 4.60 & 1.294 & 0.049 & @xcite + wise & w3 & 11.56 & 0.211 & 0.021 & @xcite + iras & & 12.00 & 0.228 & 0.034 & iras faint source catalog , v2.0 + keck@xmath118mirlin & & 12.50 & 0.200 & 0.026 & @xcite + keck@xmath118mirlin & & 17.90 & 0.186 & 0.025 & @xcite + wise & w4 & 22.09 & 0.238 & 0.024 & @xcite + spitzer@xmath118mips & & 24.00 & 0.259 & 0.010 & this work + iras & & 25.00 & 0.312 & 0.042 & iras faint source catalog , v2.0 + iras & & 60.00 & 2.17 & 0.12 & iras faint source catalog , v2.0 + iras & & 100.00 & 1.88 & 0.21 & iras faint source catalog , v2.0 + herschel@xmath118pacs & spec .
& 63.19 & 2.09 & 0.35 & this work + herschel@xmath118pacs & phot .
& 70.00 & 2.142 & 0.058 & this work + herschel@xmath118pacs & spec .
& 72.84 & 1.95 & 0.32 & this work + herschel@xmath118pacs & spec .
& 78.74 & 1.90 & 0.31 & this work + herschel@xmath118pacs & spec .
& 90.16 & 1.88 & 0.32 & this work + herschel@xmath118pacs & spec .
& 145.54 & 1.16 & 0.18 & this work + herschel@xmath118pacs & spec .
& 157.68 & 0.98 & 0.13 & this work + herschel@xmath118pacs & phot .
& 160.00 & 1.004 & 0.053 & this work + herschel@xmath118pacs & spec . &
179.53 & & this work + herschel@xmath118pacs & spec .
& 189.57 & & this work + herschel@xmath118spire & phot .
& 250.00 & 0.372 & 0.027 & this work + herschel@xmath118spire & phot .
& 350.00 & 0.180 & 0.014 & this work + herschel@xmath118spire & phot .
& 500.00 & 0.086 & 0.009 & this work + jcmt & & 800.00 & & @xcite + jcmt@xmath118scuba & & 850.00 & 0.0082 & 0.0019 & @xcite + iram & & 1200.00 & 0.0127 & 0.0028 & @xcite + sma & & 1300.00 & & @xcite to determine the expected stellar flux at all wavelengths , we compared photospheric models to the 49 cet sed shortward of about 8 @xmath0 m , where the dust excess starts to become noticeable .
a careful analysis of the 49 cet stellar parameters appears in @xcite .
we were motivated to revisit that work by the existence of important information not available or utilized at the time : specifically , the independent stellar age estimate ( @xmath7 myr ; * ? ? ?
* ) and the _ spitzer _ irs spectrum .
we found that while the photospheric model computed using the best stellar parameters in @xcite fit the optical and near - ir fluxes well , it appeared slightly too bright when compared with the short wavelength portion of the irs spectrum .
therefore , we fit atlas9 solar - metallicity photospheric models to the optical and near - ir data @xcite .
the data used were the @xmath113- and @xmath114-band fluxes from the tycho-2 catalog , the @xmath115 , @xmath116 , and @xmath117 fluxes from 2mass , and the color - corrected wise band 1 flux .
no interstellar reddening was applied to the models .
the best - fitting model had an effective temperature @xmath119 k and also matched the short wavelength portion of the irs spectrum well .
this temperature is typical of an a1v star but slightly cooler than the 9500 k effective temperature in @xcite .
however , this fitting procedure could not provide a useful constraint on the stellar gravity .
using the new effective temperature and age estimate , we placed the star on a @xmath120 h - r diagram , forcing it to lie precisely on the 40-myr isochrone .
we used the evolutionary tracks and isochrones with solar metallicity from the yonsei - yale group @xcite . the stellar gravity indicated by the age constraint
is @xmath121 , the corresponding stellar mass is 2 @xmath122 , and the stellar luminosity is 15.5 @xmath123 .
a photospheric model with @xmath124 k and @xmath121 was built by interpolating two atlas9 solar - metallicity photospheric models with @xmath1254.0 and 4.5 .
no extinction was applied to the model , which was then normalized to the irs spectrum at 6 @xmath0 m .
the final stellar model is shown with a black solid line in figure [ fig : sed ] . a strong check of the internal consistency of the set of stellar parameters consists of using the stellar luminosity derived from the age constraint ( @xmath126 ) and the total photospheric flux ( @xmath127 , computed by integrating the final stellar model ) to recover the observed distance .
using the simple expression @xmath128 , the derived distance is 56 pc , in excellent agreement with the _ hipparcos _ distance of 59 pc .
a lower gravity would imply a larger luminosity and hence a much greater distance ; as an example , @xmath124 k and @xmath129 in the @xmath130 h - r diagram corresponds to a luminosity of 34.67 @xmath123 ( using pms tracks ) or 36.31 @xmath123 ( using post - ms tracks ) .
these luminosities give distances of @xmath131 pc , inconsistent with the _ hipparcos _ distance .
we began our analysis of the 49 cet dust excess emission with a very simple model , i.e. the sum of the best stellar model and a single - temperature modified blackbody .
starting from equation 6 in @xcite , an optically thin modified blackbody has the form @xmath132 where @xmath133 is the planck function for temperature @xmath134 , @xmath135 is the optical depth , and @xmath136 is the solid angle of the emitting area ( effectively a constant of proportionality ) . the optical depth is proportional to the dust mass opacity . at mid - ir to mm wavelengths , mass opacities for expected debris disk grains
have a power - law dependence on wavelength ( see figure 7 in * ? ? ?
* ) . at shorter wavelengths ,
the mass opacity for simple astronomical silicate grains abruptly becomes nearly independent of wavelength .
therefore , our formulation for the optical depth is @xmath137 where @xmath138 is the optical depth at reference wavelength @xmath90 and @xmath3 is the emissivity power - law index .
this simple model was least - squares fit to the 49 cet sed , ignoring flux upper limits .
we found that a single - temperature blackbody model could not simultaneously fit the mid - ir emission seen with _ spitzer _ and the far - ir emission peak .
the parameters for the best dust model that fits the far - ir data appear in table [ tab : sed ] .
the fractional infrared luminosity ( @xmath139 ) , which is the ratio of the blackbody model integrated over all frequencies to the similarly integrated stellar model , is proportional to @xmath138 . the reduced @xmath140 value is extremely large due to the model s failure to also match the numerous points in the irs spectrum .
that failure indicates the disk has dust grains at a wide range of temperatures , which can be interpreted as a warm inner dust component and cold outer material .
@xcite came to the same conclusion and fit the 49 cet sed with a two - component dust disk .
lccc @xmath134 ( k ) & @xmath141 & @xmath142 & @xmath143 + @xmath90 ( @xmath0 m ) & @xmath144 & @xmath145 & + @xmath3 & @xmath146 & @xmath147 & + @xmath139 @xmath148 & @xmath149 & @xmath150 & @xmath151 + reduced @xmath140 & @xmath152 & we therefore made a total sed model consisting of the best stellar model , a warm simple blackbody , and a cold modified blackbody .
this model was also least - squares fit to the 49 cet sed , ignoring flux upper limits as before .
the best - fitting two - temperature model is over - plotted in figure [ fig : sed ] , along with the two individual blackbody components ; the model parameters appear in table [ tab : sed ] .
the reduced @xmath140 value is greatly improved compared to the single - temperature model .
the temperature of the warm component ( @xmath143 k ) is higher than the temperature found by fitting a single - temperature simple blackbody to the irs spectrum alone ( @xmath153 k ; * ? ? ?
this is presumably because we are now accounting for the flux in the irs spectrum coming from the cold component .
the emissivity power - law index found for the cold component ( @xmath154 ) is smaller that the values typically seen in the interstellar medium ( @xmath155 ; e.g. * ? ? ?
this is often seen in protoplanetary and debris disk seds and it indicates that the emitting grains are larger than the typical sub - micron interstellar grains ( e.g. * ? ? ?
assuming the dust grains are in radiative equilibrium with the central star , we estimated the minimum characteristic radius of each dust component .
starting from equation 1 in @xcite , we adopted simple blackbody grains and used the bolometric stellar flux from our best stellar model .
more realistic grains will be further from the star than simple blackbody grains at the same temperature , so the distances calculated are lower limits on the characteristic dust radii .
the characteristic radius of the warm component is @xmath156 au , while that of the cold component is @xmath157 au .
there are two possible explanations for the gas content of the 49 cet disk .
as mentioned in the introduction , the most common previous scenario interprets the co as the last remnant of the primordial gas left over from star formation ( e.g. * ? ? ?
in the first subsection , we describe our attempts to fit the 49 cet sed and gas observations using a primordial - like disk model .
the next subsection considers the second explanation , that the 49 cet gas is secondary material coming from the destruction of planetary material .
we began a detailed investigation of the 49 cet gas emission using an advanced disk modeling code , prodimo , which self - consistently determines the temperature , density , and chemical structure of combined gas and dust disks @xcite .
prodimo calculates the heating and cooling of both gas and dust , and takes into account 960 chemical reactions involving 71 species .
photochemistry , h@xmath12 formation on grain surfaces , and cosmic - ray heating are all included . as a first step ,
we considered the 49 cet disk model from @xcite that successfully reproduced the basic features of the spatially and spectrally resolved sub - mm co emission , as well as the dust emission from the outer disk ( the inner disk was ignored ) . in that work ,
steady - state chemistry was adopted and solar elemental abundances assumed .
some key features of the model were 1 ) a lack of gas within 40 au of the star , 2 ) an outer disk radius of 200 au , and 3 ) a total gas mass of @xmath158 . in the original @xcite model , the gas - to - dust ratio had to be set to the unusually high value of 500 .
this was likely caused by necessary simplifying assumptions about the dust properties : well - mixed gas and dust , a uniform dust composition ( astronomical silicates ) , and a single large grain size .
here we retained the first two assumptions but used a dohnanyi power - law grain size distribution ( @xmath159 ; * ? ? ? * ) .
we then calculated several models that reproduced the observed sub - mm co emission while varying other disk parameters ( e.g. gas - to - dust ratios from 300 to 65 ) . while it was possible to find models that matched the sub - mm co emission and the upper limit on [ ] 63 @xmath0 m emission , they under - predicted the [ ] 158 @xmath0 m emission by factors of @xmath160 . in sum
, simple adjustments to the @xcite model were unsuccessful at simultaneously reproducing the sub - mm co emission , the emission , and the lack of emission .
for example , reducing the oxygen abundance in an attempt to decrease the emission led to less co formation and reduced co cooling , which resulted in a warmer gas disk .
paradoxically , this had the effect of increasing the emission rather than decreasing it .
the only adjustment that showed some promise involved increasing the carbon abundance over the solar value . further work on detailed modeling of the 49 cet disk with prodimo is underway . for
now , it appears that the observations can not be reproduced with a `` primordial - like '' protoplanetary disk model .
we therefore consider whether the gas could also be consistent with a secondary source from planetesimals .
if so , its composition could shed light on the make - up of the young parent planetesimals outcomes of the planet formation process and the building blocks of full - sized planets .
questions that can be addressed include whether the parent planetesimals are rocky or icy and whether there are any abundance anomalies ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in this scenario ,
the presence of relatively abundant co ( for a debris disk ) suggests that this gas comes from icy material , more comet - like than asteroid - like ( as proposed in * ? ? ?
there are several possible mechanisms for its production , including outgassing of comet - like bodies , photodesorption of ice - coated grains , and grain - grain collisions @xcite .
how large a mass of comets would be needed to supply the observed mass of co ?
in equilibrium , the co production rate will match the loss rate .
assuming the co is in a low - density environment , chemical reactions can be ignored and the primary loss mechanism is photodissociation .
therefore , the co loss rate is @xmath161 where @xmath162 is the co number volume density , @xmath163 is the photodissociation rate , @xmath164 is the total number of co molecules , @xmath114 is the total volume of the gas .
given the low @xmath139 value , shielding of the co from dissociating uv radiation by dust grains should be negligible .
shielding by h@xmath12 is hard to assess , although most hydrogen in comets is locked in h@xmath12o rather than h@xmath12 ( e.g. * ? ? ?
for now , we ignore the competing effects of shielding and the stellar uv field , and consider only the interstellar field .
the unshielded photodissociation rate for co in a draine interstellar uv field is @xmath165 , giving an unshielded co lifetime @xmath166 yrs @xcite .
the co production rate is @xmath167 where @xmath168 is the fractional abundance of co relative to water in comets , @xmath169 is the water production rate , and @xmath170 is the total number of @xmath171 molecules .
observations of solar system comets show @xmath168 values ranging from about @xmath172 to @xmath173 @xcite . equating the production and loss rates
shows that equilibrium is achieved for a water mass loss rate of @xmath174 where @xmath175 is the total mass of h@xmath12o , @xmath176 is the mass of a h@xmath12o molecule , @xmath177 is the mass of a co molecule , and @xmath178 is the total mass of co ( @xmath179 ; * ? ? ?
therefore , the water mass loss rate needed to produce the observed co is @xmath180 since comets are mostly water , we can adopt these values as approximate comet mass loss rates .
collisional modeling of the solar system s kuiper belt predicts an initial mass of @xmath181 @xcite .
scaling this value up by the larger stellar mass of 49 cet ( @xmath182 ) gives an initial planetesimal mass of @xmath183 . with the loss rates given above
, it would take between roughly 0.4 myr and 32 myr to exhaust the total initial planetesimal mass .
since the star is likely to be @xmath7 myr old @xcite , this analysis would suggest that if the co is being produced from comets , it is a relatively short - lived phenomenon .
the dissipation of abundant primordial material left over from star formation sets crucial constraints on the formation of planetary systems . on one hand ,
the removal of the gas limits the time available for formation of gas giant planets . on the other hand ,
the presence of a modest amount of gas during the later stages of formation can help damp the inclinations and eccentricities of terrestrial planets ( e.g. * ? ? ?
therefore , understanding the speed of gas dissipation and the mechanisms by which it occurs is vitally important for informing planet formation theories .
primordial dust lifetimes are fairly well - constrained by observations to be @xmath184 myr , although there is a large dispersion in values for individual stars of nearly the same age and uncertainties remain about the effects of stellar mass , binarity , and star - forming environment ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
determining gas lifetimes , one of the primary goals of the _ herschel _ gasps project , is far harder .
while arguments can be made that gas and dust dissipate on roughly similar timescales @xcite , there is significant uncertainty about the co - evolution of gas and dust ( e.g. * ? ? ?
from this perspective , 49 cet might be a nearly unique disk system that is just finishing dissipation of its primordial gas . with gas emission reminiscent of a low - mass protoplanetary disk , 49 cet
does look to be in some sort of intermediate state .
as pointed out in @xcite , the presence of an inner dust disk that lacks significant molecular gas could be consistent with disk dissipation from the inside out through photoevaporation , if the inner disk grains are large enough not to be entrained in a photoevaporative flow ( e.g. * ? ? ?
* ) . with the previous age estimate of @xmath185
myr @xcite , 49 cet would have a relatively long but not implausible primordial gas lifetime .
however , the new 40 myr age @xcite requires an anomalously long lifetime for the observed co to be primordial gas .
our preliminary disk modeling with prodimo casts further doubt on the 49 cet gas being primordial , since we have great difficulty fitting all the gas observations with a low - mass but otherwise normal protoplanetary disk model .
looking at 49 cet s dust properties , it greatly resembles a young debris disk .
the low fractional dust luminosity ( @xmath186 of a typical herbig ae disk value ; * ? ? ? * ) and the lack of 10 @xmath0 m silicate emission indicating few small grains in the warm inner disk are both highly characteristic of debris disks ( e.g. * ? ? ?
all of the estimated ages for 49 cet could place it in a late stage of terrestrial planet formation ( e.g. * ? ? ?
furthermore , recent work has shown that many debris disk seds are best fit by two - temperature models , like 49 cet , suggesting radially separated inner and outer dust belts @xcite .
the 49 cet gas may also be secondary debris material coming from destruction of planetesimals .
our simple calculations of gas production from comet - like ices indicate that the amount of material required to produce the observed co is not implausibly large .
unfortunately , we are not able to say at this time if the lower limit on the mass and non - detection of emission are also consistent with the comet - evaporation scenario .
first , if the gas is coming from planetesimals , then the parent species for the bulk of the carbon gas is not necessarily co. in the better - studied case of @xmath3 pic , balancing production by photodissociation of co with loss by photoionization suggests that only about 2% of the total carbon gas comes from dissociation of co @xcite .
since the primary atomic gas production mechanism is likely to be photodesorption of grain surfaces or grain - grain collisions , many other materials could supply carbon gas to the disk ( e.g. amorphous carbon ) .
this makes interpreting the lower limit on the 49 cet mass difficult .
second , while we expect oxygen gas as an end - product of water ice evaporation , we are not able to turn the observed flux upper limit into a useful oxygen mass upper limit ( see section [ sub : gas_mass ] ) . however , recent monitoring of lines in optical spectra of 49 cet show absorption features that are variable in both strength and velocity shift @xcite .
similar features are also seen in spectra of @xmath3 pic and are attributed to so - called `` falling evaporating bodies '' ( febs ) , which are star - grazing planetesimals passing through the line of sight to the central star ( e.g. * ? ? ?
therefore , each absorption feature is effectively a transit of an exo - comet or exo - asteroid . taken together , the difficulty fitting the 49 cet observations with a primordial disk model , the possibility of producing the observed co from comet - like material , and the detection of gas - producing planetesimals all indicate that 49 cet is a close analog of the better - understood @xmath3 pic debris disk .
tighter constraints on the gas - to - dust ratios in both systems , through more accurate determination of dust masses and measurements of additional gas species , would help confirm their similarity and allow studies of the effect of the gas on grain dynamics .
the _ herschel _ far - ir imaging and spectroscopy of 49 cet has provided an extremely rich dataset for better characterizing both the dust and gas .
our primary findings are as follows . 1 . in our 70 @xmath0 m image ,
dust continuum emission from the outer disk is spatially resolved for the first time .
the disk orientation agrees with that determined from spatially and spectrally resolved sub - mm co emission , suggesting the gas and dust are co - spatial .
the deconvolved image does not show a central clearing in the dust ; the gaussian hwhm along the disk major axis is @xmath30 au .
the [ ] 158 @xmath0 m emission line was detected at the @xmath187 level in a preliminary shallow spectrum and at the @xmath188 level in a deep follow - up spectrum .
no other emission lines were detected , including [ ] 63 @xmath0 m .
the integrated line flux is @xmath189 , giving an ionized carbon gas mass @xmath103 . with various assumptions
, this suggests a lower limit on the total atomic gas mass of @xmath109 .
simple modeling of the new sed confirms the two - component structure of the disk .
we find blackbody temperatures of 175 k for the inner dust and 62 k for the outer dust , implying characteristic radii of @xmath156 au and @xmath157 au .
preliminary thermochemical modeling of the combined gas and dust disk indicates that a `` primordial - like '' protoplanetary disk model can not simultaneously reproduce all the gas observations .
the gas may instead be secondary material coming from the destruction of comet - like ices .
the next step in our work on the 49 cet system is fitting the sed with an advanced dust disk modeling code ( grater ; * ? ? ?
this code includes realistic grain properties and can take into account the powerful constraints provided by resolved imaging , as was done in our modeling of the hd181327 debris disk @xcite .
the new dust disk model will provide a more accurate dust mass and can be input to prodimo for further exploration of the disk parameter space .
we will also experiment with prodimo runs in `` debris disk mode , '' adopting time - dependent chemistry and non - solar starting compositions . through these and other studies
, we hope to resolve the nature of the interesting 49 cet disk .
_ herschel _ is an esa space observatory with science instruments provided by european - led principal investigator consortia and with important participation from nasa .
support for this work was provided by the nasa _
_ science center through an award issued by jpl / caltech .
a. roberge also acknowledges support by the goddard center for astrobiology , part of the nasa astrobiology institute .
augereau thanks the cnes - pnp for financial support . c. eiroa , g. meeus , and b. montesinos
were partly supported by spanish grant aya 2011 - 26202 . | we present far - ir / sub - mm imaging and spectroscopy of 49 ceti , an unusual circumstellar disk around a nearby young a1v star .
the system is famous for showing the dust properties of a debris disk , but the gas properties of a low - mass protoplanetary disk .
the data were acquired with the _ herschel space observatory _ pacs and spire instruments , largely as part of the `` gas in protoplanetary systems '' ( gasps ) open time key programme . disk dust emission is detected in images at 70 , 160 , 250 , 350 , and 500 @xmath0 m ; 49 cet is significantly extended in the 70 @xmath0 m image , spatially resolving the outer dust disk for the first time .
spectra covering small wavelength ranges centered on eight atomic and molecular emission lines were obtained , including [ ] 63 @xmath0 m and [ ] 158 @xmath0 m .
the line was detected at the @xmath1 level the first detection of atomic emission from the disk .
no other emission lines were seen , despite the fact that the line is the brightest one observed in _
herschel _
protoplanetary disk spectra ( e.g. * ? ? ?
* ; * ? ? ?
* ) .
we present an estimate of the amount of circumstellar atomic gas implied by the emission .
the new far - ir / sub - mm data fills in a large gap in the previous spectral energy distribution ( sed ) of 49 cet .
a simple model of the new sed confirms the two - component structure of the disk : warm inner dust and cold outer dust that produces most of the observed excess .
finally , we discuss preliminary thermochemical modeling of the 49 cet gas / dust disk and our attempts to match several observational results simultaneously .
although we are not yet successful in doing so , our investigations shed light on the evolutionary status of the 49 cet gas , which might not be primordial gas but rather secondary gas coming from comets . |
since the discoveries of @xmath4 and @xmath5 more than thirty years ago , a great deal of information on heavy quarkonium levels and their transitions has been accumulated @xcite .
on one hand , the known levels of charmonium and bottomonium include @xmath0 , @xmath6 , @xmath7 , @xmath1 , @xmath8 , @xmath9 , @xmath10 , and @xmath11 states , with the labels @xmath12 , @xmath13 , @xmath14 corresponding to relative orbital angular momentum @xmath15 = 0 , 1 , 2 between quark and antiquark . on the other hand ,
the numerous transitions between heavy quarkonium states are classified into strong and radiative decays , which shed light on aspects of quantum chromodynamics ( qcd ) in both the perturbative and the non - perturbative regimes ( for a recent review see @xcite ) .
recently the more precise @xmath16 , @xmath17 , and @xmath18 mass measurements have been reported @xcite , and the errors of their relevant decay widths have decreased @xcite . a thorough understanding of their properties , such as their quark distribution amplitudes , which are the universal non - perturbative objects , will be of great benefit when analyzing the hard exclusive processes with heavy quarkonium production .
it has been known that heavy quarkonium is relevant for a non - relativistic treatment @xcite .
although non - relativistic qcd ( nrqcd ) is a powerful theoretical tool for separating the high energy modes from low energy contributions , in most cases the calculation of low energy hadronic matrix elements has relied on model - dependent non - perturbative methods . therefore , various methods have been employed in heavy quarkonium physics , such as lattice qcd , quark - potential model , etc . @xcite .
the light - front quark model offers many insights into the internal structures of the bound states . in this study ,
heavy quarkonium is explored within a quark model on the light front .
light - front qcd is a promising analytic method for solving the non - perturbative problems of hadron physics @xcite , and may be the only possible method by which the low energy quark model and the high energy parton model can be reconciled . for hard processes with a large momentum transfer , light - front qcd reduces to perturbative qcd ( pqcd ) which factorizes the physical quantity into a convolution of the hard scattering kernel and the quark distribution amplitudes .
the basic ingredient in light - front qcd is the relativistic hadron wave function .
it generalizes the distribution amplitudes by including the transverse momentum distributions , and it contains all the information of a hadron from its constituents .
the hadronic quantities are represented by the overlap of wave functions and can be derived in principle .
the light - front wave function is manifestly a lorentz invariant as it is expressed in terms of the internal momentum fraction variables which are independent of the total hadron momentum .
moreover , the fully relativistic treatment of quark spins and the center - of - mass motion can be carried out using the so - called melosh rotation @xcite .
this treatment has been successfully applied to calculate many phenomenologically important meson decay constants and hadronic form factors @xcite .
in addition , the covariant light - front approach @xcite has also been applied to ground - state @xmath19-wave mesons , which include @xmath20 pseudoscalar mesons @xmath21 and @xmath22 vector mesons @xmath23 ( see ref .
@xcite ) .
as mentioned above , the quark distribution amplitude or light - cone wave function ( lcwf ) absorbs the non - perturbative dynamics and is the key ingredient of any hard exclusive process with hadron production . in the literature , there are many theoretical studies @xcite of this issue .
the main purpose of this study is the calculation of the quark distribution amplitudes of pseudoscalar and vector heavy quarkonium states by integrating the transverse momenta of momentum distribution amplitudes within the light - front approach .
as to the so - called @xmath3-moments which parameterize the quark distribution amplitudes , they are also calculated analytically . the remainder of this paper is organized as follows .
section ii comprises brief reviews of the light - front framework and the light - front analysis for the decay constants of pseudoscalar ( @xmath13 ) and vector ( @xmath24 ) mesons ; the processes @xmath25 are given . in section iii , the quark distribution amplitudes and the @xmath3-moments are calculated . in section iv , the numerical results and discussions are presented .
finally , the conclusions are given in section v.
in heavy quarkonium , the valence quarks have equal masses , @xmath26 , with @xmath27 the mass of heavy quark @xmath28 or @xmath29 .
thus , the formulae in this section lead to simplifications for the quarkonium system .
the momentum of a particle is given in terms of the light - front component by @xmath30 where @xmath31 and @xmath32 , and the light - front vector is written as @xmath33 .
the longitudinal component @xmath34 is restricted to positive values , i.e. , @xmath35 for the massive particle . in this way , the physical vacuum of light - front qcd is trivial except for the zero longitudinal momentum modes ( zero - mode ) .
a meson with total momentum @xmath13 and two constituents , quark and anti - quark whose momenta are @xmath36 and @xmath37 , respectively will be studied . in order to describe the internal motion of the constituents , it is crucial to introduce the intrinsic variables @xmath38 through & & p_1^+=x p^+ , p_1=x p_+ p _ ; + & & p_2^+=(1-x ) p^+ , p_2=(1-x ) p_- p _ , where @xmath39 is the light - front momentum fraction .
the invariant mass @xmath40 of the constituents and the relative momentum in @xmath41 direction @xmath42 can be written as [ eq : mpz ] m_0 ^ 2= , p_z=(x-)m_0 .
the invariant mass @xmath40 of @xmath43 is generally different from the mass @xmath44 of meson which satisfies @xmath45 .
this is due to the fact that the meson , quark and anti - quark can not be on - shell simultaneously .
the momenta @xmath46 and @xmath42 constitute a momentum vector @xmath47 , which represents the relative momenta in the transverse and @xmath41 directions , respectively .
the energy of the quark and antiquark @xmath48 can be obtained from their relative momenta , e=. it is straightforward to find that x= , e=. as shown in ref .
@xcite , one can pass to the light - front approach by integrating the @xmath49 component of the internal momentum in the covariant feynman momentum loop integrals .
the feynman rules for the meson - quark - anti - quark vertices were then needed to calculate the amplitudes which related to the decay constant and m1 transition . in the following formulations ,
we follow the notation in @xcite .
the vertices @xmath50 for the incoming meson @xmath44 are given as [ eq : hm ] h_p _ 5 & & for p , + ih_v & & for v. the feynman rules which are derived from quantum field theory and eq .
( [ eq : hm ] ) can be used to write down the relevant matrix elements .
the integration of the @xmath49 component will force the quark or anti - quark to be on its mass shell .
the specific form of the covariant vertex function for the on - shell ( anti)quark can be determined by comparing it to the conventional vertex function , which can be written as |m(p , ^2s+1l_j , j_z ) = & & \{d^3p_1}\{d^3p_2 } 2(2)^3 ^3 ( p -p_1-p_2 ) + & & _ _
1,_2 ^ss_z(p_1,p_2,_1,_2 ) |q_1(p_1,_1 ) |q_2(p_2,_2 ) .
[ lfmbs ] the momentum - space wave function @xmath51 can be expressed as ^ss_z(p_1 , p_2,_1,_2 ) = r^ss_z__1_2(x , p _ ) _ ( x , p _ ) , where @xmath52 is the light - front momentum distribution amplitude for the @xmath19-wave meson and can be chosen to be normalizable , i.e. , it satisfies [ eq : norm1 ] |(x , p_)|^2=1 , and @xmath53 constructs a state of definite spin ( @xmath54 ) out of light - front helicity ( @xmath55 ) eigenstates . in practice , it is more convenient to use the covariant form for @xmath53 @xcite : r^ss_z__1_2(x , k _ )
= |u(p_1,_1)v(p_2,_2 ) , [ covariant ] where & & = _ 5 ( pseudoscalar , s=0 ) , + & & = -(s_z)+ ( vector , s=1 ) .
all details are shown in appendix a of ref .
the function @xmath56 and the parameter @xmath57 are reduced to @xmath58 and @xmath59 , respectively , and they are written by [ eq : htowf ] h_p&=&h_v=(m^2-m_0 ^ 2)(x , p _ ) , + w_v&=&m_0 + 2 m . in principle
, @xmath60 is obtained by solving the light - front qcd bound - state equation @xmath61 , which is the familiar schr " odinger equation in ordinary quantum mechanics and @xmath62 is the light - front hamiltonian .
however , except in some simple cases , the full solution has remained a challenge .
there are several popular phenomenological light - front momentum distribution amplitudes that have been employed to describe various hadronic structures in the literature .
a widely used one is the gaussian - type chosen here for the @xmath0 heavy quarkonium state @xcite : ^1s(x , p_)=4 ( ) ^3/4 exp ( - ) .
[ gaussian1s ] in addition , the momentum distribution amplitude for the @xmath1 heavy quarkonium state is considered .
( [ eq : norm1 ] ) and ( [ gaussian1s ] ) can be rewritten as & & |(p)|^2 = 1 , + & & ^1s(p)=4 ( ) ^3/4 exp ( - ) , [ gaussian1s_p3 ] where @xmath63 . the fourier transition , or the conjugate coordinate wave function of eq .
( [ gaussian1s_p3 ] ) is the ground - state solution of the harmonic - oscillator ( ho ) problem : ^1s(r)=()^3/4 exp ( - ) .
[ coordinate1s ] then , for consistency , the relevant excited - state solutions ^1p_m(r)&= & r exp ( -)y_1m(,),[coordinate1p ] + ^2s(r)&=&()^3/4 ( -)exp ( - ) , [ coordinate2s ] ( @xmath64 s are the spherical harmonics ) , are applied to the excited meson states .
this suggestion has been given by the authors of ref .
@xcite . therefore , the fourier transforms of eqs .
( [ coordinate1p ] ) and ( [ coordinate2s ] ) can be rewritten as ^1p_m(x , p_)&=&4 ( ) ^3/4 exp ( - ) , [ gaussian1p ] + ^2s(x , p_)&=&4()^3/4 ( -)exp ( - ) , [ gaussian2s ] ( @xmath65 , and @xmath66 ) and can be treated as the momentum distribution amplitudes of the @xmath6 and @xmath1 meson states , respectively .
in fact , eq . ( [ gaussian1p ] ) has been used for the @xmath67-wave mesons in ref .
this paper further analyzes the momentum distribution amplitudes as shown in eqs .
( [ gaussian1s ] ) and ( [ gaussian2s ] ) .
the formulae of the decay constants @xmath68 and m1 transition form factor @xmath69 are needed in the latter analysis .
however , their derivations have been done in refs .
the definitions and formulae for them are provided here .
the decay constants of mesons @xmath68 are defined by the matrix elements for @xmath13 and @xmath24 mesons 0|a_|p(p ) & = & if_p p _ , + 0|v_|v(p , ) & = & m_v f_v _ , where @xmath70 is the momentum of meson and @xmath71 is the polarization vector of @xmath24 meson .
the formula for @xmath72 is f_p & = & dx d^2p _
m + & = & dx d^2 p _ _ p(x , p _ ) , [ eq : dcp ] where @xmath73 is the color number and @xmath27 denotes the quark mass . as to the formula for @xmath74 , we considered the case with the transverse polarization ( ) =( _ p_,0 , _ ) , _ = ( 1,i ) . and obtain f_v & = & dx d^2p _ + & = & dx d^2 p _ _ v(x , p _ ) .
[ eq : dcv ] this expression can be shown to be in agreement with eq . ( 2.22 ) of ref . @xcite .
since the momentum distribution function is even in @xmath42 , a quality defined in eq .
( [ eq : mpz ] ) , it follows that dx d^2p _
( x-)m_0=0 .
therefore , dx d^2p _ xm_0 ^ 2=dx d^2p _ . for the transition @xmath25 , a more general process @xmath76 where the final photon is off - shell is considered .
the @xmath77 transition is parameterized in term of a vector current form factor @xmath75 by _
= ie_^q^p^v(q^2 ) , where @xmath78 is the amplitude of the @xmath77 process . @xmath13
( @xmath79 ) is the momentum ( polarization vector ) of the initial vector meson , @xmath80 denotes the momentum of the final pseudoscalar meson , and the momentum transfer @xmath81 .
the formula for the form factor @xmath75 is [ eq : vq2 ] v(q^2)&=&dx d^2p _ + & & , where @xmath82 , @xmath83 , @xmath84 , and @xmath85 .
the rate for @xmath25 is [ eq : vpr2 ] ( vp)=|v(0)|^2 .
in this section , the momentum distribution amplitudes of pseudoscalar and vector heavy quarkonium states are analyzed .
the forms of quark distribution amplitude and @xmath3-moments can be derived from them .
the quark distribution amplitude of the pseudoscalar heavy quarkonium state can be defined as follows @xcite : 0||c(z)^_5[z ,- z]c(-z)|p = if_p p^^1_-1de^i(pz)(,),[lcwf ] where @xmath87 and @xmath88 is an energy scale which separates the perturbative and non - perturbative regimes .
the factor @xmath89 $ ] is defined as [ z ,- z]=pexp[iq^z_-zdx^a_(x ) ] , which makes the matrix element eq .
( [ lcwf ] ) gauge invariant .
the quark distribution amplitude @xmath90 is normalized as ^1_-1d(,)=1 , and it can be expanded @xcite in gegenbauer polynomials @xmath91 as ( , ) = _ as ( ) , where @xmath92 is the asymptotic quark distribution amplitude and @xmath93 the gegenbauer moments which describe to what degree the quark distribution amplitude deviates from the asymptotic one . @xmath91s
have the orthogonality integrals ^1_-1 ( 1-^2)c^3/2_m ( ) c^3/2_n ( ) d= _ mn .
[ orthogonal ] then @xmath94 can be obtained by using the above orthogonality integrals as a_n ( ) & = & ^1_-1c^3/2_n()(,)d . an alternative approach to parameterize quark distribution amplitude
is to calculate the so - called @xmath3-moments , ^n_=^1_-1 d ^n ( , ) .
it is easy to find the relations between the gegenbauer moments and the @xmath3-moments . here
we list them up to @xmath95 a_2 & = & -[1 - 5 ^2 ] , + a_4 & = & [ 1 - 14 ^2 + 21 ^ 4 ] , + a_6 & = & -[5 - 135 ^2 + 495 ^ 4 - 429 ^6].[anmoments ] all the @xmath96-odd moments are vanishing because this paper only focuses on the @xmath19-wave momentum distribution amplitudes , which are @xmath3-even functions .
there are some similar procedures @xcite by which the quark distribution amplitude can be obtained from the equal time wave function . within the framework of this study , the decay constant eq .
( [ eq : dcp ] ) can be rewritten as 1=^1_0 dx and we defined lcwf @xmath97 as _
p(x,)=^p_^2 < ^2 d^2p _ _ p(x , p _ ) where @xmath98 .
if the momentum distribution amplitudes eqs .
( [ gaussian1s ] ) and ( [ gaussian2s ] ) are applied as the non - perturbative inputs , the suppression of the gaussian function allows us to do the @xmath2 integrals up to infinity with no loss of accuracy .
thus , the results can be obtained as follows : ^1s_p()&= & ( ) ^5/4 e^d , [ lc1s ] + ^2s_p()&=&()^5/4 e^d \ { -(+d ) } , [ lc2s ] where @xmath99 $ ] is the incomplete gamma function = ^_y t^r-1e^-t dt , [ gamma ] and @xmath100 . the incomplete gamma function may be expressed quite elegantly in terms of the confluent hypergeometric function = -r^-1y^r _ 1f_1(r;r+1;-y),[hpgeo ] where _
if_j(r_1,r_2, ... ,r_i;r_1,r_2, ... ,r_j;y)=^_n=0,[1f1 ] and @xmath101 is the pochhammer symbol . after fixing the parameters @xmath27 and @xmath102 , the behaviors of the quark distribution amplitudes , eqs .
( [ lc1s ] ) and ( [ lc2s ] ) , are determined and compared with those of other theoretical groups in section iv .
it is worth to mention that eqs .
( [ lc1s ] ) and ( [ lc2s ] ) have incorrect asymptotic behavior because the gaussian - type wave functions are primarily only suitable for the low energy region . at energy scales accessible at current experiments
, a quark distribution amplitude can be far from the asymptotic form .
the worth of the asymptotic form is that it provides a criterion . in other words
, one can use the gegenbauer moments ( or @xmath3-moments ) to describe to what degree his quark distribution amplitude deviates from the asymptotic one .
in fact , the @xmath3 integrals of @xmath103 can also be analytically performed , and the decay constants @xmath104 can be expressed as f_p^1s & = & ( ) ^5/4 m e^d \ { _ 1f_1(- ; ; -d ) - _ 1f_1 ( ; ; -d ) } , [ 1sanalytic ] + f_p^2s & = & 3()^5/4 m e^d \{-2 d _ 1f_1(- ; ; -d ) + _ 1f_1(- ; ; -d)}.[2sanalytic ] it is evident that @xmath104 is only dependent on the values of @xmath27 and @xmath105 .
( [ 1sanalytic ] ) and ( [ 2sanalytic ] ) can be expanded in terms of @xmath106 as f_p^1s & ~ & c_1(1 + 3d+d^2+d^3+ ... )+c_1 d^3/4 ( 1+d+d^2+d^3+ ... ),[1sseries ] + f_p^2s & ~&-c_2d(1+d+d^2+d^3 ... )+ c_2 d^3/4(1 + 2 d+d^2+d^3+ ...
),[2sseries ] where all @xmath107 are positive constants . however , for a typical value of @xmath106 , all the series in eqs .
( [ 1sseries ] ) and ( [ 2sseries ] ) converge very slowly .
in addition , the @xmath3- moments can be expressed analytically as ^2 ^ 1s_p & = & a^1s_p\ { _ 1f_1(- ; , -d)- _ 1f_1(- , , -d ) } , + ^4 ^ 1s_p & = & a^1s_p\ { _ 1f_1(- ; , -d)- _ 1f_1(- , , -d ) } , + ^6 ^ 1s_p & = & a^1s_p\ { _ 1f_1(- ; , -d)- _ 1f_1},[ximoment1s ] and ^2 ^ 2s_p & = & a^2s_p\ { _ 1f_1(-,;-d ) + _ 1f_1(-,;-d ) } , + ^4 ^ 2s_p & = & a^2s_p\ { _ 1f_1(-,;-d ) + _ 1f_1(-,;-d ) } , + ^6 ^ 2s_p & = & a^2s_p\ { _ 1f_1(-,;-d ) + _ 1f_1(-,;-d ) } , [ ximoment2s ] where a^1s_p = ( ) ^5/4 e^d , a^2s_p = ( ) ^5/4 e^d .
the derivations of eqs .
( [ ximoment1s ] ) and ( [ ximoment2s ] ) have used the formula
_ 1 f_1(a;b;c)= , which is easily checked from the definition of the confluent hypergeometric function eq .
( [ 1f1 ] ) .
similar to the case of @xmath86 , eq . ( [ eq : dcv ] ) can be rewritten and @xmath109 defined as _
v(x ) = d^2 p _ .
[ lcwfvector ] however , the @xmath2 integrals in eq .
( [ lcwfvector ] ) can not be analytically performed .
the crux is the third term in the square bracket , that is , the term proportional to @xmath110 .
this term may be rewritten and expanded as [ vwapp ] = , because , in the non - relativistic limit , @xmath111 .
then , we defined the approximate quark distribution amplitudes as & & _v(x ) = d^2 p _ , [ phiv0 ] + & & _
v(x ) = d^2 p _ .[phiv1 ] after applying the momentum distribution function in eq .
( [ gaussian1s ] ) and fitting the parameters , we found that not only the center values of the parameters @xmath102 of eqs .
( [ phiv0 ] ) and ( [ phiv1 ] ) were both in the error bar of @xmath112 ( see table i ) , but also that their curves were almost the same .
in addition , a similar situation also existed in the case of the @xmath1 states .
therefore , we have only shown the results from the form @xmath113 . by performing the @xmath2 integrals , it can be obtained that ^1s_v ( ) & = & ( ) ^5/4 \{d+(3+^2 ) } , [ analyphiv1s ] + ^2s_v ( ) & = & ( ) ^5/4 \{d+(3+^2 ) + & & -(+d)(d+(3+^2))}. [ analyphiv2s ] the above derivations are based on the formula : -(r-1)=y(-(r-2 ) ) , which is easily checked from the definition of incomplete gamma function , eq .
( [ gamma ] ) .
the @xmath3-moments of @xmath114 can also be obtained analytically ^2 ^ 1s_v & = & a^1s_v\ { + & & + } , + ^4 ^ 1s_v & = & a^1s_v\ { + & & + } , ^6 ^ 1s_v & = & a^1s_v\ { + & & + } , [ ximomentv1s ] and ^2 ^ 2s_v & = & a^2s_v\ { + & & + } , + ^4 ^ 2s_v & = & a^2s_v\ { + & & + } , + ^6 ^ 2s_v & = & a^2s_v\ { + & & + } , [ ximomentv2s ] where a^1s_v=()^5/4 , a^2s_v=()^5/4 .
in this work , the numerical results were calculated for the quark distribution amplitudes @xmath115 and @xmath3-moments for the pseudoscalar and vector heavy quarkonium .
first , it was necessary to determine the parameters appearing in the momentum distribution functions . in total , there are five for charmonium , @xmath116 , @xmath117 , @xmath118 , @xmath112 , @xmath119 , and five ones for bottomonium : @xmath120 , @xmath121 , @xmath122 , @xmath123 , @xmath124 .
the decay constants of vector heavy quarkonium states were determined first .
the decay constant @xmath74 is related to the electromagnetic decay of vector meson @xmath125 by @xcite [ eq : dcvexp ] ( ve^+e^-)=c_v f_v^2 , where @xmath126 is the square of the electric charge of heavy quark . from the data @xcite
, we obtained the values f_j/&=&416 7 mev , f_=298 8 mev , [ fvectorc ] + f_&=&715 5 mev , f_=497 4 mev.[fvectorb ] for the decay constants of @xmath127 , the two decay modes @xmath128 and @xmath129 are considered using the following relation = ( ) ^2 ||^2 |(1-)+ |^2,[btoccratio ] where @xmath130 and @xmath131 are related to the wilson coefficients and can be determined to have the value @xmath132 @xcite , _
abc=^1/2 , and the form factors @xmath133 are defined by the lorentz decomposition of the matrix element k(p)||s _
b | b(p+q)=f_+(q^2 ) ( 2p+q)_+ f_-(q^2)q_. by the results of ref .
@xcite one calculated the form factors @xmath133 within the covariant light - front quark model . combining the above with the experimental values , we can obtain f__c=421 38 mev .
[ fetac ] this result leads to the ratio ( ) ^2=1.020.22,[thisworkratio ] which is consistent with the van royen - weisskopf formula @xcite , ( ) ^2= 1.04,[vrwratio ] when the approximation @xmath134 is used . for the decay constant @xmath135 , however , a similar estimation of eq .
( [ btoccratio ] ) has a large uncertainty because the error of @xmath136 was greater than @xmath137 @xcite .
given the consistency between eq .
( [ thisworkratio ] ) and ( [ vrwratio ] ) , the approximation @xmath138 was used , and the corresponding van royen - weisskopf formula , ( ) ^2= 1.01 , [ appratio ] is applied and the decay constant @xmath139mev is obtained .
furthermore , an additional constraint is needed .
the new data @xmath140 @xcite are chosen to calculate the form factor @xmath69 in eq .
( [ eq : vpr2 ] ) . for the charm sector , there were one assumption ( eq .
( [ appratio ] ) ) and four constraints ( the decay constants ( [ fvectorc ] ) , ( [ fetac ] ) , and the new data @xmath141 ) were used to fix the five parameters appearing in the momentum distribution amplitudes .
the results are listed in table i. .[tab : para ] fixed parameters @xmath116 and @xmath102 s for charmonium states ( in units of gev ) . @xmath142 and @xmath143 are the parameters appearing in equations ( [ phiv0 ] ) and ( [ phiv1 ] ) . [ cols="^,^,^,^,^,^",options="header " , ]
this study performed the transverse momenta @xmath2 integrals of formulae for the decay constants @xmath68 of @xmath0 and @xmath1 heavy quarkonium states and then obtained their quark distribution amplitudes @xmath144 .
in addition , the @xmath3-moments @xmath145 were also obtained by integrating out @xmath3 in @xmath144 . for each heavy quarkonium state ,
the five parameters @xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 which appeared in the momentum distribution amplitude were determined .
this study first extracted the decay constants @xmath151 from the experimental data of the leptonic decay @xmath152 , and it then used the van royen - weisskopf formula to obtain the decay constants @xmath153 .
these decay constants were used as constraints to fix the above parameters .
then , the curves of the quark distribution amplitudes for the @xmath0 and @xmath1 heavy quarkonium states were plotted by the fixed parameters .
it was found that , for the @xmath0 charmonium state , the momentum fraction @xmath39 in the function used by ref .
@xcite was more centered on @xmath154 than the one in the gaussian - type wave function .
in addition , the @xmath39-distribution of the pseudoscalar bottomonium was almost the same as that of the vector bottomonium .
the reason for this was that the differences between these heavy quarkonium states , which arise from @xmath155 corrections , become small when @xmath146 is large .
finally , the numerical results of the @xmath3-moments were calculated and compared with the experimental data and other theoretical predictions . * acknowledgements *
+ the author would like to thank hsiang - nan li , tsung - wen yeh , and alexey luchinsky for their helpful discussions .
this work is supported in part by the national science council of r.o.c . under grant no nsc-96 - 2112-m-017 - 002-my3 . | in this paper , the quark distribution amplitudes of @xmath0 and @xmath1 heavy quarkonium states are studied in terms of gaussian - type wave functions .
the transverse momenta @xmath2 integrals of the formulae for the decay constant are performed analytically .
then the quark distribution amplitudes are obtained .
in addition , the @xmath3-moments are also calculated .
after fixing the relevant parameters appearing in the quark distribution amplitude , the curves of the quark distribution amplitude for @xmath0 and @xmath1 heavy quarkonium states are plotted .
finally , the numerical results of this approach are compared with the other theoretical predictions . |
let @xmath2 . let @xmath0 be the hyperbolic n - space and @xmath11 its group of isometries .
suppose that @xmath12 is a discrete , torsion free subgroup .
the action of @xmath13 on @xmath0 extends to a continuous action on the compactification of @xmath0 by the sphere at infinity @xmath14 .
the limit set @xmath15 is the set of all accumulation points of the orbit of a point in @xmath0 .
@xmath15 does not depend on the choice of the point and is a subset of @xmath14 ( see @xcite or @xcite for the case @xmath16 ) .
the domain of discontinuity @xmath5 is the complement of @xmath15 in @xmath14 .
the quotient space @xmath7 is a manifold because @xmath13 is torsion free and hence acts without fixed points on @xmath0 ( see @xcite ) .
let @xmath17 be the projection map .
let @xmath18 be the topological metric on @xmath0 induced by its riemannian metric and @xmath19 be the topological metric on @xmath20 induced by its riemannian metric . in this paper
we prove the following theorem .
[ thm;main ] suppose that @xmath3 is a discrete , torsion free subgroup and @xmath21 .
then there exists an open neighborhood @xmath8 of @xmath4 in @xmath9 such that @xmath22 the author would like to thank the referee for many valuable suggestions , in particular , those which simplified the proof of lemma [ lem;lipschitz ] .
suppose that @xmath2 and @xmath3 is a discrete , torsion free subgroup .
let @xmath23 be the compactification of poincar ball model of @xmath0 by the sphere at infinity @xmath14 .
let @xmath24 and @xmath25 .
we write @xmath26 to denote the topological euclidean metric on @xmath27 .
we will make use of the following equation ( see @xcite ) .
suppose that @xmath21 .
due to theorem [ thm;crucial ] , we can choose an open neighborhood @xmath29 of @xmath4 in @xmath37 such that @xmath30 for all @xmath31 . choose @xmath38 such that @xmath39 .
let @xmath40 .
suppose that @xmath41 and @xmath31 .
notice that @xmath42 and @xmath43 . since @xmath44
, we have @xmath45 .
therefore @xmath46 hence @xmath47 a mbius transformation of @xmath48 is a finite composition of reflections of @xmath48 in spheres or planes .
a mbius transformation of @xmath27 is a mbius transformation of @xmath48 that leaves @xmath27 invariant .
we will also make use of the following theorem ( see @xcite , 4.4 , 4.5 ) .
[ mobius ] suppose that @xmath49 and @xmath31 .
then @xmath50 extends to @xmath51 , a unique mbius transformation of @xmath27 ( which we also refer to as @xmath50 ) , where @xmath52 is an orthogonal transformation of @xmath53 and @xmath54 is a reflection in a sphere which is orthogonal to @xmath14 . to get a contradiction ,
suppose that @xmath64 is an infinite set .
recall that @xmath13 does not contain any element which has a fixed point in @xmath0 except the identity . therefore if @xmath65 , then @xmath66 .
hence @xmath67 any orthogonal transformation is an euclidean isometry which fixes @xmath72 .
therefore , for all @xmath73 , from theorem [ mobius ] we have @xmath74 therefore if @xmath73 , then @xmath75 thus , by eq .
( [ eq;b1 ] ) , @xmath76 has an accumulation point in @xmath77 .
this is a contradiction to the fact @xmath60 .
therefore @xmath78 is a finite set . hence @xmath79 .
suppose that @xmath21 . by theorem [ thm;crucial ]
, we can choose an open neighborhood @xmath29 of @xmath4 in @xmath9 such that @xmath30 for all @xmath31 . choose @xmath81 such that @xmath39 .
let @xmath82 . to get a contradiction ,
suppose that for all @xmath83 , there exist @xmath84 , @xmath85 and @xmath86 such that @xmath87 we may assume that @xmath88 for all @xmath83 . for each @xmath83
, let @xmath89 be as in theorem [ mobius ] , where @xmath90 is the reflection in the sphere @xmath91 . by proposition
[ prop;bounded ] , we can choose a constant @xmath92 such that @xmath93 for all @xmath83 .
therefore from theorem [ thm;inversion ] , we have @xmath94 therefore from eq .
( [ eq;l2 ] ) , we have @xmath95 thus @xmath96 .
hence there exists a limit point @xmath56 of @xmath97 such that @xmath98 choose a subsequence @xmath99 of @xmath100 such that @xmath101 .
since @xmath102 and @xmath103 is orthogonal to @xmath14 , we have @xmath104 therefore by theorem [ thm;inversion ] , we have @xmath105 thus @xmath106 contradicting eq .
( [ eq;l1 ] ) .
hence there exists a constant @xmath35 such that @xmath107 suppose that @xmath21 . by lemma [ lem;lipschitz ]
, there exist open neighborhood @xmath8 of @xmath4 and @xmath92 such that @xmath109 note that we may assume that @xmath110 and the boundary of @xmath8 is a sphere orthogonal to @xmath14 , intersected with @xmath27 .
suppose that @xmath112 and @xmath49 . since @xmath110 , we have @xmath113 .
consider @xmath114 and @xmath115 as vectors in @xmath116 and let @xmath117 so that @xmath111 ( see figure [ fig ] )
. since @xmath118 , we have @xmath119 and @xmath120 . therefore by eq .
( [ eq;lip ] ) , we have @xmath121 let @xmath122
. then @xmath123 . by applying proposition [ prop;good ] followed by lemma [ lem;trivial ]
, we can choose an open neighborhood @xmath8 of @xmath4 in @xmath9 and a constant @xmath35 such that @xmath126 for all @xmath127 and @xmath31 .
notice that eq .
( [ eq;m2 ] ) is true for @xmath133 , too .
hence @xmath134 for all @xmath127 and @xmath49 .
therefore from eq .
( [ eq;qmetric ] ) , we have @xmath135 it is clear from eq .
( [ eq;qmetric ] ) , that @xmath136 for any @xmath127 .
therefore @xmath137 for all @xmath127 . | let @xmath0 be the hyperbolic @xmath1-space with @xmath2 .
suppose that @xmath3 is a discrete , torsion free subgroup and @xmath4 is a point in the domain of discontinuity @xmath5 .
let @xmath6 be the projection map from @xmath0 to the quotient manifold @xmath7 . in this paper
we prove that there exists an open neighborhood @xmath8 of @xmath4 in @xmath9 such that @xmath6 is an isometry on @xmath10 . |
in a dual beam optical tweezer setup ( fig . [ sys ] ) the protein is covalently connected to double - stranded dna handles that are attached to glass or polystyrene beads in two optical traps . for small displacements of the beads from the trap centers
@xcite , the trap potentials are harmonic , with strengths @xmath3 along the lateral plane , and a weaker axial strength @xmath4 , where @xmath5 @xcite . for simplicity , we take both traps to have equal strengths , though our method can be generalized to an asymmetric setup .
the trap centers are separated from each other along the @xmath6 axis , with trap 1 at @xmath7 and trap 2 at @xmath8 .
as the bead - handle - protein ( bhp ) system fluctuates in equilibrium , the positions of the bead centers , @xmath9 and @xmath10 , vary in time .
we assume that the experimentalist can collect a time series of the @xmath11 components of the bead positions , @xmath12 and @xmath13 . denote the mean of each time series as @xmath14 and @xmath15 .
we assume that the trap centers are sufficiently far apart that the whole system is under tension , which implies that the mean bead displacements are non - zero , @xmath16 , where @xmath17 is the mean tension along @xmath6 .
we focus on the case where there is no feedback mechanism to maintain a constant force , so the instantaneous tension in the system changes as the total end - to - end extension component @xmath18 [ fig .
[ sys ] ] varies .
though we choose one particular passive setup , the theory can be adapted to other types of passive optical tweezer systems @xcite , where the force is approximately constant ( in which case we could skip the transformation into the constant - force ensemble described below ) .
the mean tension @xmath17 , a measure of the overall force scale , can be tuned at the start of the experiment by making the trap separation @xmath19 larger ( leading to higher @xmath17 ) or smaller ( leading to lower @xmath17 ) . because @xmath20 , the precise relationship between @xmath19 and @xmath17 requires knowing the mean total extension @xmath21 , which depends among other things on the details of the energy landscape .
hence , we can not in general calculate beforehand what @xmath17 will be for a given @xmath19
. however , one of the advantages of our approach is that we can combine data from different experimental runs ( each having a different @xmath19 and @xmath17 ) to accurately construct the protein free energy profile .
this combination is carried out through the weighted histogram analysis method ( wham ) @xcite ( see supplementary information ( si ) for details ) , in a spirit similar to earlier work in the context of optical tweezers @xcite .
we first solve the problem of obtaining the protein landscape based on a single observed trajectory of bead - to - bead separations specified as @xmath22 as a function of @xmath1 .
the key quantity in the construction procedure is @xmath23 , the equilibrium probability distribution of @xmath22 within the external trap potential , which can be directly derived from the experimental time series .
the imperfect nature of the measured data , due to noise and low - pass filtering effects in the recording apparatus , will distort @xmath24 , but we have developed a technique to model and approximately correct for these issues ( see finite bandwidth scaling ( fbs ) in the _ methods _ ) .
once we have an experimental estimate for @xmath24 , the objective is to find @xmath25 , the intrinsic distribution of the protein end - to - end extension component @xmath2 at some _ constant _ force @xmath26 , whose value we are free to choose .
( we will use tilde notation to denote probabilities in the constant - force ensemble . ) the intrinsic protein free energy profile is @xmath27 .
the procedure , obtained from rigorous theoretical underpinnings described in detail in the si , consists of two steps : 1 .
_ transformation into the constant - force ensemble_. given @xmath24 , we obtain the total system end - to - end distribution at a constant @xmath26 using , @xmath28 where @xmath29 and @xmath30 is a normalization constant .
the equation above applies in the case of a single experimental trajectory at a particular trap separation @xmath19 .
_ extraction of the intrinsic protein distribution_. in the constant - force ensemble , @xmath31 , relates the total end - to - end fluctuations @xmath32 to the end - to - end distributions for the individual components @xmath33 , where @xmath34 denotes bead ( b ) , handle ( h ) , or protein ( p ) , and @xmath35 is a 1d convolution operator . for the beads ,
`` end - to - end '' refers to the extension between the bead center and the handle attachment point , projected along @xmath36 . in fourier space
the convolution has the form : @xmath37 where @xmath38 is the fourier transform of @xmath39 .
here @xmath40 , which is the result of convolving all the bead and handle distributions , acts as the main point spread function relating the intrinsic protein distribution @xmath41 to @xmath42 .
since @xmath40 can be modeled from a theoretical description of the handles and beads , we can solve for @xmath41 using eq . and hence find @xmath43 , the intrinsic free energy profile of the protein .
the derivation of the procedure ( given in the si , along with technical aspects of its numerical implementation ) shows the conditions under which the two step method works .
the mathematical approximation underlying step 1 becomes exact if either of the following hold : ( i ) @xmath44 ; ( ii ) the full 3d total system end - to - end probability is separable into a product of distributions for longitudinal ( @xmath6 ) and transverse ( @xmath45 , @xmath46 ) components .
in general , condition ( ii ) is not physically sensible @xcite . however , if @xmath47 is the typical length scale describing transverse fluctuations , then condition ( i ) is approximately valid when @xmath48 .
if this condition breaks down , accurate construction of the intrinsic energy landscape can not be performed without knowledge of the transverse behavior .
however , in the simulation and experimental results below , the force scales are such that transverse fluctuations are small , @xmath49 , so to ensure condition ( i ) is met , we require that @xmath50 pn / nm at @xmath51 k. we use the experimental value @xmath52 in our test cases @xcite , which is well under the upper limit . in principle , one can choose any @xmath26 , the force value of the constant force ensemble where we carry out the analysis . in practice , @xmath26 should be chosen from among the range of forces that is sampled in equilibrium during the actual experiment , since this will minimize statistical errors in the final constructed landscape .
for example , setting @xmath53 , the mean tension , is a reasonable choice .
step 2 depends on knowledge of @xmath54 , and thus the individual constant - force distributions of the beads and the handles in fourier space .
the point spread function is characterized by : the bead radius @xmath55 , the handle contour length @xmath56 , the handle persistence length @xmath57 , and the handle elastic stretching modulus @xmath58 . in @xmath59 we also include the covalent linkers which attach the handles to the beads and protein .
if we model these linkers as short , stiff harmonic springs , we have two additional parameters : the linker stiffness @xmath60 and natural length @xmath61 . using the extensible semiflexible chain as a model for the handles
, we exploit an exact mapping between this model and the propagator for the motion of a quantum particle on the surface of a unit sphere @xcite to calculate the handle fourier - space distribution to arbitrary numerical precision . together with analytical results for the bead and linker distributions
, we can thus directly solve for @xmath54 . to verify that the analytical model for the point - spread function can accurately describe handle / bead fluctuations over a range of forces , we have analyzed data from control experiments on a system involving only dsdna handles attached to beads , where @xmath62 ( si ) .
the theory simultaneously fits results for several experimental quantities measured on the same system : the distributions @xmath40 derived from three different trap separations , corresponding to mean forces @xmath63 pn , and a force - extension curve .
the accuracy of the model @xmath40 is @xmath64 , within the experimental error margins .
we first apply the theory to a problem for which the intrinsic free energy profiles at arbitrary force are known exactly .
the generalized rouse model ( grm ) hairpin ( see si for details ) is a two - state folder whose full 3d equilibrium end - to - end distributions are analytically solvable .
a representative grm distribution @xmath65 at @xmath66 pn is plotted in fig .
[ grma](a ) .
since @xmath65 is cylindrically symmetric , the top panel shows a projection onto the @xmath67 plane , while the bottom panel shows the further projection onto the @xmath11 coordinate .
the two peaks correspond to the native ( n ) state at small @xmath11 , and the unfolded ( u ) state at large @xmath11 . in order to model the optical tweezer system , we add handles and beads to the grm hairpin , whose probabilities
@xmath68 and @xmath69 ( including transverse fluctuations ) are illustrated in fig .
[ grma](b ) and ( c ) .
the full 3d behavior is derived in an analogous manner to the theory mentioned above for the 1d fourier - space distribution @xmath70 of the beads / handles ; the only difference is that the transverse degrees of freedom are not integrated out .
the 3d convolution of the system components , plus the optical trap contribution , gives the total distribution @xmath71 in fig .
[ grma](d ) .
the bead , handle , linker , and trap parameters are listed in si table s1 . from @xmath71 one
can calculate the mean total @xmath11 extension and the mean tension , which in this case are @xmath72 nm , @xmath73 pn . the @xmath74 probability projection in the bottom panel of ( d ) is the information accessible in an experiment , and the computation of the intrinsic distribution in the bottom panel of ( a ) is the ultimate goal of the construction procedure . comparing ( a ) and ( d ) , two effects of the apparatus are visible : the grm peaks have been partially blurred into each other , and the transverse ( @xmath75 ) fluctuations have been enhanced .
the handles provide the dominant contribution to both these effects . figs . [
grma](e ) through ( g ) illustrate the construction procedure for the grm optical tweezer system .
panel ( e ) corresponds to step 1 , with a transformation of the distribution @xmath71 ( whose varying force scale is shown along the top axis ) into @xmath76 at constant force @xmath66 pn .
step 2 uses the exact @xmath40 , shown in real - space in panel ( f ) , and produces the intrinsic distribution @xmath65 , drawn as a solid line in ( g ) . the agreement with the exact analytical result ( dashed line )
is extremely close , with a median error of @xmath77 over the range shown .
this deviation is due to the approximation in step 1 , discussed above , as well as the numerical implementation of the deconvolution procedure .
as shown in our previous study @xcite , the smaller the ratio @xmath78 for the handles , the more the features of the protein energy landscape get blurred by the handle fluctuations . since the experimentally measured total distribution always distorts to some extent the intrinsic protein free energy profile due to the finite duration and sampling of the system trajectory , more flexible handles will exacerbate the signal - to - noise problem . to illustrate this effect
, we performed brownian dynamics simulations of the grm in the optical tweezer setup , with handles modeled as extensible , semiflexible bead - spring chains ( see si for details ) . in fig .
[ grmb](a ) we compare the free energy @xmath79 for a fixed @xmath80 nm and a varying @xmath78 , derived from the simulation trajectories , and the exact intrinsic grm result @xmath81 at @xmath26 . when the handles are very flexible , with @xmath82 , the energy barrier between the native and unfolded states almost entirely disappears in @xmath83 , with the noise making the precise barrier shape difficult to resolve .
remarkably , even with this extreme level of distortion , using our theory we still recover a reasonable estimate of the intrinsic landscape [ fig .
[ grmb](b ) ] . for each @xmath83 in fig .
[ grmb](a ) , panel fig .
[ grmb](b ) compares the result of the construction procedure and the exact answer for @xmath84 .
clearly some information is lost as @xmath78 becomes smaller , since the @xmath85 system does not yield as accurate a result as the ones with stiffer handles .
however in all cases the basic features of the exact @xmath86 are reproduced .
thus , the theoretical - based method works remarkably well over a wide range of handle parameters .
this conclusion is generally valid even when other parameters are varied ( see fig .
s3 in the si for tests at various @xmath26 and @xmath87 ) .
the excellent agreement between the constructed and intrinsic free energy profiles for the exactly solvable grm hairpin over a wide range of handle and trap experimental variables establishes the robustness of the theory . to demonstrate that the theory can be used to produce equilibrium intrinsic free energy profiles with multiple states from mechanical folding trajectories
, we performed simulations of a protein in an optical tweezer setup .
the simulations were designed to mirror the single - molecule experiment reported in ref .
@xcite , and to this end we studied a coiled - coil , lz26 @xcite , based on three repeats of the leucine zipper domain from the yeast transcriptional factor gcn4 @xcite ( see _ methods _ ) .
the simple linear unzipping of the two strands of lz26 allows us to map the end - to - end extension to the protein configuration .
furthermore , the energy heterogeneity of the native bonds that form the `` teeth '' of the zipper leads to a non - trivial folding landscape with at least two intermediate states @xcite .
the more complex landscape of lz26 thus provides an additional stringent test of the proposed theory .
the native ( n ) structure of lz26 is illustrated on the right in fig .
[ lz26 ] ( from a simulation snapshot ) , with the two alpha - helical strands running from n - terminus at the bottom to c - terminus at the top . in the experiment
a handle is attached to the n - terminus of each strand , and this is where the strands begin to unzip under applied force . to prevent complete strand separation , the c - termini are cross - linked through a disulfide bridge between two cysteine residues .
each alpha - helix coil consists of a series of seven - residue heptad repeats , with positions labeled a through g. for the leucine zipper the a and d positions are the `` teeth '' , consisting of mostly hydrophobic residues ( valine and leucine ) which have strong non - covalent interactions with their counterparts on the other strand . the exceptions to the hydrophobic pattern are the three hydrophilic asparagine residues in @xmath88 positions on each strand ( marked in blue in the structure snapshots on the right of fig .
[ lz26 ] ) . as has been seen experimentally @xcite ( and shown below through simulations ) , the weaker interaction of these asparagine pairs is crucial in determining the properties of the intermediate folding states , a point we will return to in more detail in the discussion . in analyzing the lz26 leucine zipper system , we performed coarse - grained simulations using the self - organized polymer ( sop ) model @xcite ( full details in the si , with selected parameters summarized in table s1 ) .
the intrinsic free energy profile @xmath89 at @xmath90 pn is shown in fig .
[ lz26](a ) .
the four prominent wells in @xmath43 as a function of @xmath2 correspond to four stages in the progressive unzipping of lz26 . at @xmath91
pn all the states are populated , and the system fluctuates in equilibrium between the wells .
the transition barrier between n and i1 exhibits a shallow dip that may correspond to an additional , very transiently populated intermediate .
since this dip is much smaller than @xmath92 , we do not count it as a distinct state .
like in the grm example , adding the optical tweezer apparatus to the sop simulation significantly distorts the measured probability distributions . in the first row of fig .
[ deconv ] sample simulation trajectory fragments are shown both for the protein - only case [ fig .
[ deconv](a ) ] at constant force @xmath93 pn , and within the full optical tweezer system [ fig .
[ deconv](c ) ] with @xmath94 nm . for the latter case we plot both @xmath0 ( purple ) and @xmath95 ( gray ) , allowing us to see how the bead separation tracks changes in the protein extension .
the probability distributions @xmath41 and @xmath71 are plotted in fig .
[ deconv](b ) and ( d ) respectively . in fig .
[ deconv](e ) , the distribution @xmath71 within the optical tweezer system is plotted for @xmath94 nm .
though we only illustrate this particular @xmath19 value , @xmath96 trajectories are generated at different @xmath19 and combined together using wham @xcite ( see si ) to produce a single @xmath42 at a constant force @xmath90 pn [ fig .
[ deconv](e ) ] .
we can then use our theoretical method to recover the protein free energy @xmath43 [ fig .
[ deconv](f ) ] .
despite numerical errors due to limited statistical sampling ( both in the protein - only and total system runs ) , there is remarkable agreement between the constructed result and @xmath43 derived from protein - only simulations .
this is particularly striking given that the total system free energy @xmath97 , plotted for comparison in panels ( f ) , shows how severely the handles / beads blur the energy landscape , reducing the energy barriers to a degree that the n state is difficult to resolve .
the signature of n in @xmath98 is a slight change in the curvature at higher energies on the left of the i1 well .
however despite this , we still recover a basin of attraction representing the n state in the constructed @xmath99 .
overall , the results in ( f ) show that our theory can accurately produce the intrinsic free energy profiles using only the simulated folding trajectories as input , thus proving a self - consistency check of the method for a system with multiple intermediates . as a final test of the efficacy of the theory we used the experimental time series data @xcite to obtain @xmath43 .
the data consists of two independent runs with the lz26 leucine zipper , using the same handle / bead parameters for each run ( see table s1 ) , but at different trap separations @xmath19 .
we project the deconvolved landscape from each trajectory onto the mid - point force @xmath26 where the two most populated states ( i1 and u ) have equal probabilities in @xmath41 .
the values of @xmath26 derived from the two runs are the same within error bounds : @xmath100 and @xmath101 pn .
the detailed deconvolution steps are shown for one run in the last row of fig .
[ deconv ] , and the final result , the intrinsic free energy profile @xmath43 , is shown for both runs in fig .
[ deconv](h ) ( solid and dotted blue curves respectively ) .
accounting for error due to finite trajectory length and uncertainties in the apparatus parameters , the median total uncertainty in each of the reconstructed landscapes is about @xmath102 @xmath103 in the @xmath11 range shown ( see si for full error analysis ) .
the landscapes from the two independent runs have a median difference of @xmath104 @xmath92 , and hence the method gives consistent results between runs , up to a small experimental uncertainty , an important test of its practical utility .
the reproducibility of @xmath43 is a testament to the stability of the dual optical tweezer setup , allowing us to sample extensively from the energy landscape : each trajectory lasted for more than 100 s , and thus collected @xmath105 of the various types of transitions between protein states ( the slowest transition , @xmath106 , occurred on time scales of @xmath107 s ) .
comparison between the experimental @xmath43 in panel ( h ) and the simulation result in ( f ) reveals a notable difference : the landscape constructed using the experimental data does not have four identifiable basins .
the n state may not be discernible in the experiment because of the limited resolution of the apparatus ( see below ) .
the spacing between the i1 and i2 wells is similar in the simulation and experiment ( @xmath108 nm ) , but that between i2 and u is @xmath109 nm in the simulation versus 25 nm in the experiment .
this is likely due to a larger helix content in the unfolded state for the simulation case .
our theory for the point spread function @xmath110 can be used to understand the interplay of physical effects that relate the intrinsic protein distribution to the total system . to quantify the various contributions to @xmath110 , we calculated its variance .
since variances of probability distributions combine additively upon convolution , we break down the variance of @xmath111 into the individual bead , handle , and linker contributions . fig .
[ grmb](c ) shows the fraction of the variance associated with each component as a function of the handle elastic stretching modulus @xmath58 at @xmath91 pn , with @xmath112 nm , @xmath113 nm , @xmath114 nm ( the approximate experimental parameters from ref .
@xcite ) . for any given value of @xmath58 ,
the height of each of the four colored slices represents four fractions . though not directly measured in ref .
@xcite , we have assumed @xmath115 kcal / mol@xmath116nm@xmath117 , @xmath118 nm for the linkers .
the handle contribution is itself broken down into the `` elastic '' part , defined as the extra variance due to the finite stretching modulus @xmath58 , compared to an inextensible ( @xmath119 ) worm - like chain ( wlc ) , and the remainder , which we call the wlc part .
for the case of ref .
@xcite , @xmath120 pn . since the length extension relative to the wlc result is @xmath121 , we expect finite handle extensibility to play a small role . however , the elastic contribution to the total @xmath122 variance at this @xmath58 is 43% , comparable to the wlc contribution of 48% .
hence , in predicting @xmath40 correctly it is important to account for both the bending rigidity and elasticity of the handles , which are exactly modeled in our approach .
the folding landscape of lz26 is apparently closely related to the pattern of residue - residue contact energies between the two strands of the zipper @xcite .
sop simulations give us a detailed picture of this relationship .
the average fraction of intact inter - strand ( `` zipper '' ) bonds vs. extension @xmath11 , in fig .
[ lz26](b ) is a monotonic curve , starting with the fully closed structure on top ( n state , bond fraction near 1 ) to the fully open structure at the bottom ( u state , bond fraction near 0 ) .
listed along this curve are the individual residues at the @xmath88 and @xmath123 positions of the heptads in the sequence .
several features stand out : the transition barriers between the states show a steeper rate of zipper bond unraveling compared to the well regions .
the change of slope from steep to more gradual descent occurs near the location of the asparagine residues in the sequence , and the the well minima of i1 , i2 , and u occur one or two residues after the asparagines .
the correlation between well minima locations and asparagines agrees with the experimental landscape @xcite , underscoring the importance of the weak , hydrophilic asparagine bonds that interrupt the hydrophobic valine / leucine pattern at the a / d positions .
the sequence of rescaled bt @xcite energies used for the a / d native contacts is plotted in fig . [ lz26](c ) .
the a / d bonds are all @xmath124 @xmath103 , except for the asparagines , which are less stable at @xmath125 @xmath92 .
the difference in the number of wells in the simulation and experimental free energy landscapes of the leucine zipper is related to finite time and spatial resolution .
the measured time series is subject to noise ( environmental vibrations of the optical elements , detector shot noise ) , as well as low - pass filtering due to `` parasitic '' effects in the photodiodes and the nature of the electronic amplification circuits @xcite .
the standard experimental protocol often involves additional low - pass filtering as a way of removing noise and smoothing trajectories : for the leucine zipper every five data points ( originally recorded at 10 @xmath126s intervals ) are averaged together during collection to give a time step of 50 @xmath126s @xcite ; in other cases similar effects are achieved using bessel filters @xcite .
noise broadens the measured distribution of bead separations , while low - pass filtering narrows it .
we developed the fbs technique ( _ methods _ and si ) , based on the details of the specific apparatus used in the experiment , to estimate and correct for the distortions .
for our system , the fbs theory provides an excellent description , as we have verified in tests using both numerical simulations and experimental data ( with and without the additional filtering ) .
however the fbs theory can only apply corrections to peaks ( i.e. distinct protein states ) that we observe in the measured probability distributions .
there is the possibility of protein states leaving no discernible signature in the recorded distribution .
the n state in the leucine zipper is only connected to the i1 state in the folding pathway . in the simulations , where the n state is directly observed , it has short mean lifetimes ( @xmath127 @xmath126s in the studied force range ) , and the @xmath128 change involves the shortest mean extension difference ( @xmath129 nm ) among all the transitions .
if the n state in the actual protein has similar properties , it could be impossible to resolve it in the experimental data for two different reasons : ( i ) regardless of any additional filtering , the intrinsic low - pass characteristics of the apparatus filter out states with very short lifetimes . for our lot setup , the effective low - pass filter time - scale for the detectors / electronics is @xmath130 @xmath126s ( si ) , which is at the cutting edge of current technology .
thus , states with lifetimes @xmath131 will not appear as distinct peaks in the measured distribution .
( ii ) independent of the filtering issues in detection / recording , environmental background noise in the time series also poses a problem , particularly since we measure bead displacements , and these have signal amplitudes at high frequencies that are generally attenuated compared to the intrinsic amplitudes of the protein conformational changes .
the reason for this is that the beads have much larger hydrodynamic drag than dsdna handles or proteins , and their characteristic relaxation times @xmath132 in the optical traps may be comparable to or larger than the lifetime of a particular protein state . the bead can not fully respond to force changes on time scales shorter than its relaxation time @xcite . for example , @xmath133 @xmath126s in the leucine zipper experiment . if the lifetime of the n state at a particular force is much smaller than @xmath132 , protein transitions from i1@xmath134n@xmath134i1 will generally occur before the bead can relax into the n state equilibrium position .
if the bead displacements associated with these transitions are smaller than the noise amplitude in the time series , the entire excursion to the n state will be lost to the noise .
we can illustrate the finite response time of the bead using simulations where resolution is not limited by noise or apparatus filtering , allowing us to illustrate the relationship between @xmath0 and @xmath95 , compared in two different trajectory fragments in fig .
[ deconv](c ) .
triangles in the figure indicate times where the protein makes a transition between states .
changes in protein extension during these transitions are very rapid , and the bead generally mirrors these changes with a small time lag , as seen in the enlarged trajectory interval at @xmath135 @xmath126s . when the protein makes sharp , extremely brief excursions ( like a visit to the n state from i1 in the enlarged interval @xmath136 @xmath126s ) , the corresponding changes in bead separation are smaller and much less well - defined . in the presence of noise , such tiny changes would be obscured .
thus , we surmise that the n state is not observable due to some combination of apparatus filtering , noise , and finite bead response time .
hence , the theory applied to the experimental data produces a landscape with only i1 , i2 , and u wells , as opposed to the four wells produced from the simulation data .
our labeling of the basins in the landscape agrees with the earlier state identification @xcite , and provides an explanation for why the n state was not resolved .
extraction of the energy landscape of biomolecules using lot data is complicated because accurate analysis depends on correcting for distortions due to system components on the measured result .
we have solved this problem completely by developing a theoretically - based construction method that accounts for these factors . through an array of tests involving an analytically solvable hairpin model , coarse - grained protein simulations , and experimental data , we have demonstrated the robustness of the technique in a range of realistic scenarios .
the method works for arbitrarily complicated landscapes , as demonstrated by the analysis of the leucine zipper experimental data , producing consistent results when the same protein is studied under different force scales .
probability distributions derived from experimental time series of bead - bead separations are corrupted by noise , low - pass filtering due to the apparatus , and in some cases additional filtering due to the data processing protocol .
we developed fbs theory to model and correct for these effects ( see si for details ) , using information encoded in time series autocorrelations , together with earlier spectral characterization of the dual trap optical tweezer detector and electronic systems @xcite .
all the experimental distributions @xmath71 in the main text were first processed by fbs .
we use a variant of the coarse - grained self - organized polymer ( sop ) model @xcite , where each of the 176 residues in lz26 is represented by a bead centered at the @xmath137 position ( see si for details . ) the @xmath34-helical secondary structure is stabilized by interactions which mimic @xmath138 hydrogen bonding @xcite .
we use residue - dependent energies for tertiary interactions @xcite .
we simulate ( see si for details ) trajectories for both the protein alone and the full optical tweezer setup using an overdamped brownian dynamics ( bd ) algorithm @xcite .
the handles used in the lot setup [ fig .
[ sys ] ] are modeled as semiflexible chains .
m.h . was a ruth l. kirschstein national research service postdoctoral fellow , supported by a grant from the national institute of general medical sciences ( 1 f32 gm 97756 - 1 ) .
d.t . was supported by a grant from the national institutes of health ( gm 089685 ) .
part of the work was done while d.t .
was in tum as a senior humboldt fellow .
10 thirumalai d , hyeon c ( 2005 ) rna and protein folding : common themes and variations .
_ biochemistry _ 44:49574970 .
onuchic j , luthey - schulten z , wolynes pg ( 1997 ) theory of protein folding : the energy landscape perspective .
chem . _ 48:539600 .
dill ka , ozkan sb , shell ms , weikl tr ( 2008 ) the protein folding problem .
_ 37:289316 .
thirumalai d , obrien ep , morrison g , hyeon c ( 2010 ) theoretical perspectives on protein folding .
_ 39:159183 .
guo z , thirumalai d ( 1995 ) kinetics of protein folding : nucleation mechanism , time scales , and pathways .
_ biopolymers _ 36:83102 .
klimov dk , thirumalai d ( 2005 ) symmetric connectivity of secondary structure elements enhances the diversity of folding pathways . _
_ 353:11711186 .
stigler j , ziegler f , gieseke a , gebhardt jcm , rief m ( 2011 ) the complex folding network of single calmodulin molecules .
_ science _ 334:512516 .
woodside mt , et al .
( 2006 ) direct measurement of the full , sequence - dependent folding landscape of a nucleic acid . _ science _ 314:10011004 .
woodside mt , et al .
( 2006 ) nanomechanical measurements of the sequence - dependent folding landscapes of single nucleic acid hairpins .
usa _ 103:61906195 .
neupane k , yu h , foster dan , wang f , woodside mt ( 2011 ) single - molecule force spectroscopy of the add adenine riboswitch relates folding to regulatory mechanism . _
nucleic acids res . _ 39:76777687 .
gebhardt jcm , bornschlgl t , rief m ( 2010 ) full distance - resolved folding energy landscape of one single protein molecule .
sci . usa _ 107:20132018 .
elms pj , chodera jd , bustamante c , marqusee s ( 2012 ) the molten globule state is unusually deformable under mechanical force .
usa _ 109:37963801 .
hyeon c , thirumalai d ( 2006 ) forced - unfolding and force - quench refolding of rna hairpins .
_ biophys . j. _ 90:34103427 .
manosas m , et al .
( 2007 ) force unfolding kinetics of rna using optical tweezers .
ii . modeling experiments .
_ biophys . j. _ 92:30103021 .
von hansen y , mehlich a , pelz b , rief m , netz rr ( 2012 ) auto- and cross - power spectral analysis of dual trap optical tweezer experiments using bayesian inference .
_ 83:095116 .
yu h , et al .
( 2012 ) direct observation of multiple misfolding pathways in a single prion protein molecule .
usa _ 109:52835288 .
hummer g , szabo a ( 2010 ) free energy profiles from single - molecule pulling experiments .
u. s. a. _ 107:2144121446 .
hinczewski m , von hansen y , netz rr ( 2010 ) deconvolution of dynamic mechanical networks . _ proc .
usa _ 107:2149321498 .
hyeon c , morrison g , thirumalai d ( 2008 ) force - dependent hopping rates of rna hairpins can be estimated from accurate measurement of the folding landscapes .
usa _ 105:96049609 .
greenleaf wj , woodside mt , abbondanzieri ea , block sm ( 2005 ) passive all - optical force clamp for high - resolution laser trapping . _
_ 95:208102 .
neuman kc , block sm ( 2004 ) optical trapping . _
_ 75:27872809 .
ferrenberg am , swendsen rh ( 1989 ) optimized monte - carlo data - analysis . _
_ 63:11951198 .
shirts mr , chodera jd ( 2008 ) statistically optimal analysis of samples from multiple equilibrium states . _ j. chem .
_ 129:124105 .
m , brawn - cinani b , la porta a ( 2011 ) measuring the folding landscape of a harmonically constrained biopolymer . _
100:27362744 .
kierfeld j , niamploy o , sa - yakanit v , lipowsky r ( 2004 ) stretching of semiflexible polymers with elastic bonds . _
phys . j. e _ 14:1734 .
bornschlgl t , rief m ( 2006 ) single molecule unzipping of coiled coils : sequence resolved stability profiles .
_ 96:118102 . oshea ek , klemm jd , kim ps , alber t ( 1991 ) x - ray structure of the gcn4 leucine zipper , a 2-stranded , parallel coiled coil .
_ science _ 254:539544 .
bornschlgl t , rief m ( 2008 ) single - molecule dynamics of mechanical coiled - coil unzipping .
_ langmuir _ 24:13381342 .
hyeon c , dima ri , thirumalai d ( 2006 ) pathways and kinetic barriers in mechanical unfolding and refolding of rna and proteins .
_ structure _ 14:16331645 .
betancourt mr , thirumalai d ( 1999 ) pair potentials for protein folding : choice of reference states and sensitivity of predicted native states to variations in the interaction schemes . _
protein sci .
_ 8:361369 .
mickler m , et al .
( 2007 ) revealing the bifurcation in the unfolding pathways of gfp by using single - molecule experiments and simulations .
natl . acad .
usa _ 104:2026820273 .
denesyuk na , thirumalai d ( 2011 ) crowding promotes the switch from hairpin to pseudoknot conformation in human telomerase rna .
_ 133:1185811861 .
ermak dl , mccammon ja ( 1978 ) brownian dynamics with hydrodynamic interactions .
_ j. chem .
_ 69:13521360 .
= 20.74pt * supplementary information for `` from mechanical folding trajectories to intrinsic energy landscapes of biopolymers ''
* + = 17.28pt michael hinczewski , j. christof m. gebhardt , matthias rief , and d. thirumalai = 14.4pt .parameters used in the grm model and sop simulation of the lz26 leucine zipper , together with the corresponding quantities from two experimental runs @xcite [ cols="<,^,^,^,^",options="header " , ]
we begin with the hamiltonian for the beads in the traps ( fig . 1 in the main text ) , which allows us to introduce the relevant variables of the system .
if the displacements of the beads from the trap centers are small ( @xmath139 nm for a laser of 1064 nm wavelength and bead radii @xmath140 @xcite ) , the trap hamiltonian can be approximately written as : @xmath141,\ ] ] where @xmath142 is the position of the @xmath143th bead center , @xmath144 , @xmath145 , @xmath146 are the trap strengths along each coordinate direction , and the two traps are positioned at @xmath7 and @xmath8 respectively .
given the cylindrical symmetry of the optical traps around the @xmath46 axis , we take @xmath147 and @xmath4 , where the weaker axial trapping is reduced by a factor @xmath148 @xcite .
we assume both traps have equal strengths , though our method can be generalized to an asymmetric configuration , where the two traps have different strengths @xmath149 . in this case
the reconstruction procedure derived below is valid with the substitution @xmath150 .
we rewrite the hamiltonian in eq . by defining a total end - to - end coordinate @xmath151 , and a total center - of - mass coordinate @xmath152 . in terms of these variables , @xmath153 becomes : @xmath154 the variables @xmath22 and @xmath19 are explicitly labeled in fig . 1 of the main text . the equilibrium probability @xmath155 of finding the beads at positions with a given @xmath156 and @xmath157 can be expressed as : @xmath158 where @xmath29 , @xmath159 is a normalization constant , and @xmath160 is the equilibrium probability of the total bead - handle - protein system having bead separation @xmath157 in the absence of the external trapping potential or any applied force . by translational symmetry
@xmath161 is independent of the center - of - mass coordinates , and by rotational symmetry @xmath162 .
thus , if we introduce cylindrical coordinates @xmath163 , where @xmath164 , @xmath165 , there is no angular dependence , so that @xmath166 .
we are ultimately interested in the marginal probability @xmath167 , which can be derived from the experimental time series and forms the starting point of our theoretical procedure to obtain the desired free energy profile .
we obtain @xmath24 from @xmath168 by integrating over the @xmath156 , @xmath169 and @xmath170 degrees of freedom : @xmath171 here @xmath172 and @xmath30 are constants that have absorbed the result of integrating over @xmath156 and @xmath169 respectively , and @xmath173 is a modified bessel function of the first kind . up to the third line the calculation in eq .
is exact . in the last step
we make the problem fully one - dimensional , by approximately relating @xmath24 to @xmath174 , defined as @xmath175 .
we are forced to make this crucial approximation , because experiments have access only to the @xmath6 fluctuations through @xmath24 , but generally do not have complete information about the transverse components . as mentioned in the main text , the last step in eq .
becomes exact if : ( i ) @xmath44 ; or ( ii ) when @xmath176 is separable in the form @xmath177 for some function @xmath178 .
though condition ( ii ) is not expected to be generally valid , we can approximately satisfy ( i ) when @xmath179 , where @xmath47 is the typical length scale of total system fluctuations transverse to @xmath6 .
thus , for sufficiently soft traps , we have in eq . a useful relation between the @xmath6 marginal probabilities of the total system with and without the external trapping potentials .
since @xmath180 is the total end - to - end @xmath11-component distribution in the absence of any external trapping potential or applied force , the corresponding distribution for the total system with constant tension @xmath26 applied to the beads along @xmath6 is given by @xmath181 . substituting for @xmath174 using eq
, we find the following relation for @xmath32 , which constitutes step 1 of our construction procedure in the main text : @xmath182 the quantity @xmath24 on the right - hand side can be derived from the experimental time series , and thus eq .
allows us to obtain an equilibrium distribution in the constant force ensemble , @xmath183 , directly from the folding trajectories . in the constant force ensemble
, @xmath42 is just a 1d convolution of the probabilities of the individual system components : @xmath184 where @xmath35 denotes the convolution operator .
the probability @xmath185 is the equilibrium distribution of @xmath186 at constant force @xmath26 , where @xmath187 denotes a bead , handle , or protein .
the quantity @xmath186 is the end - to - end distance of @xmath187 along @xmath6 . using the notation in fig .
1 of the main text , we can give a few examples : for the protein @xmath188 ; for the left handle @xmath189 ; for the left bead @xmath190 . in fourier space eq .
, which is the key equation for step 2 of the construction procedure in the main text , has a simple form : @xmath191 where @xmath192 is the fourier transform of @xmath193 .
here @xmath194 is the fourier transform of the convolution of all the bead and handle distributions .
if the left and right handles ( or analogously the beads ) had distinct properties ( i.e. different sizes ) then the factor @xmath195 in @xmath54 would be replaced by the product @xmath196 of the distinct handle terms .
given the rotational properties of the beads and modeling the handles as semiflexible polymers , we can derive a numerically exact form for the fourier components @xmath197 , and hence by inversion the corresponding real space distribution .
this will allow us to directly recover @xmath41 from @xmath42 , without resorting to an experimental estimate for the point spread function , which is problematic due to the varying force conditions that arise in optical traps with non - zero stiffness .
the first step in finding @xmath198 is to obtain an expression for the fourier - space bead probability @xmath199 .
taking as an example the left bead in fig . 1
, let @xmath200 be the vector between the bead center and the point on the bead surface that is attached to the handle .
this vector has a fixed length @xmath55 given by the bead radius , but its direction can fluctuate , subject to a constant force @xmath26 along @xmath6 .
the equilibrium distribution @xmath201 is given by : @xmath202 with the delta function enforcing the constraint @xmath203 , and the normalization constant @xmath204 .
the quantity @xmath205 is the fourier transform of @xmath201 evaluated at @xmath206 : @xmath207 though the fourier components @xmath208 of the semiflexible handle distribution do not have a simple analytic expression , they can be calculated numerically to arbitrary accuracy .
the hamiltonian for the semiflexible handle polymer with contour length @xmath56 , persistence length @xmath57 , and elastic stretching modulus @xmath58 , can be exactly mapped onto the propagator of a quantum particle on the surface of a unit sphere @xcite . following the approach in ref .
@xcite , we describe the polymer as a continuous spatial contour @xmath209 in terms of an unstretched arc length @xmath210 which runs from @xmath211 to @xmath212 . at each point
@xmath210 we define a unit tangent vector @xmath213 .
the end - to - end distance @xmath214 can be written as , @xmath215 where @xmath216 is the local relative bond length extension .
for an inextensible ( @xmath119 ) worm - like chain , @xmath217 for all @xmath210 , which corresponds to all bonds in the chain having fixed length .
for finite @xmath58 , the @xmath218 are additional degrees of freedom in the system , which together with the unit tangent vectors @xmath213 completely define the contour .
the hamiltonian @xmath219 for the semiflexible polymer under tension is , @xmath220 , \end{split}\ ] ] where @xmath29 and we have used eq . for the end - to - end distance . the first term in eq
. corresponds to a bending energy parameterized by the persistence length @xmath57 , the second term is due to an applied mechanical force @xmath221 along @xmath6 , and the third term describes the stretching energy of the bonds , with elastic modulus @xmath58 . for prestretching tension @xmath26 , @xmath222 , but for convenience we will extend the definition of @xmath223 to include arbitrary @xmath178 in order to obtain the fourier components of the end - to - end probability distribution below .
the partition function of the polymer ( with free end boundary conditions ) can be expressed as a path integral over all possible configurations of @xmath213 and @xmath218 , with the constraint that @xmath224 at each @xmath210 : @xmath225,\\ & \equiv \int { \cal d}{\mathbf{u}}(s ) \prod_s \delta({\mathbf{u}}^2(s)-1)\ , \exp\left[-\beta { \cal h}_\text{eff}({\mathbf{u}}(s))\right ] \end{split}\ ] ] up to some normalization constant . in the second line we have carried out the path integral over @xmath218 exactly to express @xmath226 in terms of an effective hamiltonian @xmath227 depending on the tangent vectors alone , @xmath228.\ ] ]
the probability of finding the polymer in a configuration with an end - to - end extension @xmath229 along @xmath6 is given by @xcite : @xmath230\\ & = \frac{1}{z_\text{h}(\beta f_0 ) } \int { \cal d}{\mathbf{u}}(s ) \prod_s \delta({\mathbf{u}}^2(s)-1)\,\\ & \qquad \int { \cal d}{\mathbf{\epsilon}}(s)\,\int \frac{dk}{2\pi } e^{ik\left ( z_\text{h } - \int_0^l ds\,(1+\epsilon(s))u_z(s ) \right ) } \exp\left[-\beta { \cal h}({\mathbf{u}}(s),\epsilon(s))\right]\\ & \equiv \int \frac{dk}{2\pi}\ , e^{ik z_\text{h } } \tilde{\cal p}_\text{h}(k;f_0 ) , \end{split}\ ] ] where the fourier components of the probability distribution are : @xmath231 in order to evaluate @xmath208 , we need to calculate @xmath226 .
let us define the propagator @xmath232 as the path integral over all configurations with initial tangent @xmath233 and final tangent @xmath234 : @xmath235 this is related to the partition function through @xmath236 , where the integrations are over the unit sphere @xmath237 . the quantum hamiltonian corresponding to @xmath238 is @xmath239 describing a particle on the surface of a unit sphere , with @xmath240 defining the @xmath6 direction .
the propagator @xmath241 can be written in terms of the quantum eigenvalues @xmath242 and eigenstates @xmath243 of @xmath244 : @xmath245 where we have expanded the eigenstates in the basis of spherical harmonics , @xmath246 .
the coefficients @xmath247 are the components of the @xmath248th eigenvector of the hamiltonian @xmath249 in the @xmath250 basis . the partition function @xmath226 becomes : @xmath251 in the last step we have written the expression as a single component of the exponentiated matrix @xmath252 in the @xmath253 spherical harmonic basis , where @xmath254 denotes a state @xmath255 . since the hamiltonian matrix in the @xmath256 subspace does not couple to @xmath257 components , we only need @xmath256 matrix elements to evaluate @xmath226 .
the list of non - zero matrix entries in the @xmath256 subspace is : @xmath258 to carry out the matrix exponent , we truncate the matrix at @xmath259 , which is sufficiently large for numerical accuracy . in some experimental setups ,
covalent linkers are attached on both ends of each handle , connecting the handle to the neighboring bead and protein , as schematically drawn in fig .
[ link](a ) .
the effect of linkers can be absorbed into the theory by modifying @xmath260 .
the simplest representation of a linker is a harmonic spring with stiffness @xmath60 and natural length @xmath61 . with one of these added at each end of the handle , eq .
becomes : @xmath261 where @xmath262 .
\end{split}\ ] ] the expressions given by eq . and completely determine the fourier - transformed point spread function @xmath197 at all @xmath263 .
naively , one could use eq .
to write : @xmath264 since @xmath265 is derivable from the experimental times series , this would immediately yield @xmath266 , and after inversion the ultimate goal , @xmath267 . however , this direct deconvolution in fourier space is numerically unstable @xcite , due to the effects of round - off noise and the denominator in the equation for @xmath268 approaching zero at large @xmath263 . to work around this problem
, we implement the deconvolution in real space , by solving the following integral equation for @xmath41 ( the real space version of eq .
): @xmath269 one way to approach eq .
is to approximate the integral as a matrix - vector product by discretizing the @xmath22 and @xmath2 ranges .
however , the convolution matrix corresponding to @xmath40 is generally ill - conditioned , so direct inversion to find a solution is unfeasible . alternatively , to obtain robust , smooth results for the deconvolution , we can rewrite eq . by representing the three quantities
@xmath270 , @xmath40 , and @xmath76 in terms of suitable fitting functions . since these are all probability distributions , in practice we can approximate them to arbitrary precision as sums of gaussians @xmath271 , @xmath272 where @xmath273 , bh , or tot .
the number of gaussians needed for each distribution , @xmath274 , is chosen depending on the problem .
the two sets of parameters @xmath275 and @xmath276 ( which implicitly depend on @xmath26 ) are computed by fitting to the known functions @xmath42 and @xmath40 .
the goal of the procedure is then to use eq . to solve for the parameter set @xmath277 describing the unknown function @xmath41 .
for the cases discussed in the main text , choosing @xmath278 was sufficient to find solutions @xmath41 such that the left and right - hand sides of eq . had a median deviation @xmath279 over all @xmath22 where @xmath280 .
the details of the solution procedure are as follows : we choose @xmath281 , so that eq .
can be approximated as a one - to - one convolution mapping each gaussian in @xmath41 into a corresponding gaussian in @xmath42 . for all @xmath282 , eq .
describes the following relationships between the amplitudes , positions and variances of the gaussians : @xmath283 the approximation is exact when the point - spread function @xmath122 is precisely a single gaussian , but is generally valid whenever @xmath40 is close to gaussian ( as is the case for the bead - handle system , where the corrections introduced by choosing @xmath284 are small ) .
eq . can be inverted to yield the desired parameter set @xmath277 : @xmath285,\\ \end{split}\ ] ] where we have used the fact that @xmath286 due to normalization .
. the same set of best - fit parameters is used for all the theory curves . ( a ) force @xmath287 vs. total extension @xmath11 from an experimental pulling trajectory , compared to the theoretical mean extension as a function of force ; ( b - d ) total bead - handle probability end - to - end distance distributions @xmath40 ( blue solid curves ) derived from three experimental runs at different constant trap separations , corresponding to mean forces of @xmath288 , @xmath289 , and @xmath290 pn respectively . in each case the experimental data is corrected for noise / filtering effects using the fbs method ( sec . [ sec : fbs ] ) , and transformed into the constant force ensemble using eq . [ 1 ] of the main text , with @xmath26 chosen to be equal to the mean force value in the trajectory .
the distance scale is centered at @xmath291 , the mean extension .
the light blue shaded region around each blue curve represents the standard error margin for every point in the distribution ( 68% confidence band ) . for comparison ,
the experimental results omitting the fbs corrections are shown as gray dashed lines.,scaledwidth=98.0% ] in order to check that the theoretical model of the point - spread function @xmath40 derived in secs .
[ sec : bead]-[sec : hand ] is an accurate description of the handle and bead response in experiments , we analyzed control experiments on a system with only dsdna handles and beads .
bead radii are @xmath292 nm , the trap strength is @xmath293 pn / nm , and the handle parameters are extracted from the theoretical best - fit described below .
four distinct experimental data sets are collected ( fig .
[ hand_test ] ) : the first is from a pulling setup , where the trap separation is varied to give a trajectory of force @xmath287 vs. total extension @xmath11 ( fig . [ hand_test](a ) , blue curve ) ; the other data sets are trajectories of extension @xmath11 as a function of time collected at three different constant trap separations .
these three trajectories can be binned , and projected onto the constant force ensemble using the same method ( eq . ) as described above for the full system , yielding probability distributions @xmath294 for the total end - to - end extension of the bead - handle system ( fig .
[ hand_test](b - d ) , blue curves ) . the constant force value @xmath26 for each projection is chosen equal to the mean force in each of the three trajectories , namely @xmath295 , @xmath289 , and @xmath290 pn .
the experimental data were collected at 100 khz , with no additional time averaging beyond the electronic filtering intrinsic to the detection and recording apparatus .
prior to the projection onto the constant force ensemble , the fbs method ( sec .
[ sec : fbs ] ) was used to approximately correct the raw experimental data for distortions due to electronic filtering and noise . in the absence of these corrections , the @xmath294 from the raw data is given by the dashed curves in fig .
[ hand_test](b - d ) . the standard error margin ( 68% confidence interval ) for each point in the @xmath40 distribution
is marked by a light blue band , reflecting uncertainties in apparatus and fbs parameters , as well as statistical error due to sampling .
details of the error estimation procedure are in sec .
[ sec : err ] .
the median standard error in the @xmath11 range shown varies from @xmath296 between the three trajectories .
we use the theoretical model of secs .
[ sec : bead]-[sec : hand ] to simultaneously fit all four experimental data sets with a single set of handle parameters , yielding best fit values : @xmath297 nm , @xmath298 nm , and @xmath299 pn . the theory has excellent agreement with all the experimental results , with median deviations in @xmath294 for the @xmath11 range shown in fig .
[ hand_test](b - d ) varying from @xmath300 between the three trajectories , comparable to the standard error margins .
the comparison between theory and experiments firmly establishes the remarkable accuracy of our theory in quantitatively describing the bead - handle system .
the grm model @xcite , illustrated schematically in fig . [ link](b ) , is a gaussian chain with @xmath301 monomers , connected by @xmath302 harmonic springs with an average extension @xmath88 .
a conformation of the grm is specified by the monomer positions @xmath303 , @xmath304 .
to get behavior reminiscent of hairpin unzipping , an additional harmonic bond potential , @xmath305 , is added between the end - points @xmath306 and @xmath307 ; the force due to this potential is non - zero only if the end - point separation is within a cutoff distance , @xmath308 . under a constant external tension , @xmath309 ,
the grm hamiltonian is @xmath310 where @xmath311 , and @xmath312 is the unit step function .
we choose parameters : @xmath313 , @xmath314 @xmath315 nm ( 11.9 pn ) , @xmath316 nm , @xmath317 nm , @xmath318 @xmath315nm@xmath117 ( 0.37 pn / nm ) .
if we write the end - to - end vector @xmath319 in cylindrical coordinates as @xmath320 , the exact probability distribution for this vector in equilibrium under constant force @xmath321 is given by : @xmath322 where @xmath323 is a normalization constant .
this distribution , projected onto the @xmath324 plane , is illustrated in the top panel of fig .
2(a ) in the main text .
the peak at small @xmath11 corresponds to the `` folded '' hairpin state ( f ) with an intact end - point bond , while the peak at larger @xmath11 is the unfolded ( u ) state . integrating @xmath325 over @xmath75 and @xmath326
one obtains the marginal probability @xmath327 , @xmath328 with normalization constant @xmath329 .
@xmath327 is plotted in the lower panel of fig .
2(a ) .
after deconvolution for three different values of the force @xmath26 : ( a ) 9.9 pn ; ( b ) 11.9 pn ; ( c ) 14 pn . in each case , results for two different trap strengths @xmath330 , @xmath331 pn / nm are shown as solid lines of different color .
the exact analytical solutions are drawn as dashed lines .
the @xmath11 scale is plotted relative to @xmath332 , the location of the minimum in the free energy . ] in fig .
3(b ) in the main text we showed that the deconvolution results for the grm are robust when varying the handle parameters . in fig .
[ grmextra ] we demonstrate that the same conclusion holds when either the force @xmath26 or the trap strength @xmath87 are varied .
the weighted histogram analysis method @xcite ( wham ) is a powerful tool in analyzing optical tweezer experiments . by combining trajectories generated at different trap separations @xmath19 ( resulting in different force scales @xmath17 ) , one can sample the full extent of the protein free energy landscape , and use wham to construct a single energy profile using all the trajectory data , as has been previously done in ref .
@xcite ( and in a related , but different manner in ref .
@xcite . ) in the context of our theory , wham modifies step 1 of our procedure , allowing us to derive the equilibrium probability @xmath183 at constant force @xmath26 based on information from multiple experimental trajectories .
consider a set of @xmath333 experimental runs , where the @xmath143th trajectory consists of @xmath334 data points and has a trap separation @xmath335 . except for @xmath335 ,
all other system parameters are kept the same between runs . for
each run one can calculate the normalized histogram of total end - to - end distances @xmath22 , yielding a probability distribution @xmath336 .
this distribution is related to @xmath180 , the unbiased @xmath22 probability in the absence of a trapping potential or external force , through eq . .
inverting that equation , we can write @xmath337 where @xmath338 is a normalization constant , @xmath339 and @xmath340 . in the case of one trajectory ( @xmath341 ) ,
eq . is a way to estimate @xmath180 , from which one can calculate @xmath342 .
this is just the standard step 1 procedure described earlier . when @xmath343 , eq .
provides a different estimate of @xmath174 for each @xmath143 , which ideally should be combined to give a single best approximation .
the wham method resolves this problem , yielding a best estimate for @xmath174 of the form : @xmath344 where @xmath159 is a normalization constant .
the unknown parameters @xmath345 are given by : @xmath346.\ ] ] eqs . and are a coupled system of equations for @xmath180 and @xmath345 .
we solve these by making an initial guess for the set @xmath347 , substituting it into eq . to find @xmath180 , and using this estimate for @xmath180 in eq . to find a new set of @xmath347 .
the process is iterated until we converge to a self - consistent solution to both equations .
once we have a best estimate of @xmath174 , we can calculate @xmath183 as above , completing step 1 of the construction .
the amino acid sequence for a single @xmath34-helical strand of the lz26 coiled coil is as follows ( grouped into heptad repeats ) : mcqleqk veellqk nyhleqe varlkql vgeleqk veellqk nyhleqe varlkql vgeleqk veellqk nyhleqe varlkql vgec .
the sequence is the same as in ref .
@xcite , except that we have left out four residues at the beginning and three from the final heptad , for a total of 88 residues per strand . as in the experiment @xcite ,
the handles are attached at the cysteine in position b of the first heptad , and the cross - linking between strands is at the cysteine in position d of the last heptad .
( for consistency when comparing simulations with or without the handles / traps , end - to - end distance for the protein is always measured between the two n - terminal cysteines . )
although the crystal structure is not available for lz26 , it is believed to be similar to three gcn4 leucine zipper domains ( pdb i d code 2zta ) @xcite in series .
thus , we constructed a model for the native structure based on gcn4 , connecting the leucine zipper segments in such a way that the distances between neighboring @xmath137 positions and angles of superhelical coiling formed a continuous pattern as one moves along lz26 .
going from n- to c - terminus on one strand and returning c- to n - terminus on the other , let us label the residues @xmath348 , where @xmath349 , where @xmath350 corresponds to one strand , and @xmath351 corresponds to the other .
every non - neighboring pair of residues , @xmath352 where @xmath353 , is assigned to one of three sets : @xmath354 ( secondary structure pairs ) , @xmath355 ( tertiary structure pairs ) , and @xmath356 ( remainder ) .
the set @xmath354 consists of all pairs where @xmath357 and @xmath143 , @xmath358 share the same strand , representing residues interacting through @xmath34-helical hydrogen bonding .
the set @xmath355 consists of all pairs where @xmath143 , @xmath358 are on different strands , and the distance between the two residues in the native structure , @xmath359 , is below a cutoff : @xmath360 nm . these pairs are involved in tertiary interactions between the two @xmath34-helical coils .
all other non - neighboring pairs that do not satisfy the criteria for @xmath354 or @xmath355 fall into the set @xmath356 , and only interact via repulsive lennard - jones potentials .
the variant sop hamiltonian for lz26 has the form : @xmath361\delta_{i , i+1,i+2}\\ & - \sum_{(i , j ) \in
{ \cal s } } \epsilon_\text{hb}\left[1+k^\text{hb}_\text{bond } ( r_{i , j } - r_{i , j}^0)^2 + k^\text{hb}_\text{ang}\left ( ( \theta_{i , j , j-1 } - \theta_{i , j , j-1}^0)^2 + ( \theta_{i+1,i , j}-\theta_{i+1,i , j}^0)^2\right.\right.\\ & \qquad\qquad\qquad\left.\left.+ ( \phi_{i+1,i , j , j-1 } - \phi_{i+1,i , j , j-1}^0)^2\right)\right]^{-1}\\ & + \sum_{(i , j ) \in { \cal t } } \chi |\epsilon_\text{bt}(i , j)-\epsilon_s| v_\text{lj}\left(\frac{r^0_{i , j}}{r_{i , j}}\right ) + \sum_{(i , j ) \in { \cal r } } \epsilon_\text{rep } \left(\frac{\sigma}{r_{i , j}}\right)^6 . \end{split}\ ] ] the first term is the nearest - neighbor bond potential , where @xmath362 is the distance between residues @xmath143 and @xmath358 , and the spring constant @xmath363 kcal / mol@xmath116nm@xmath117 .
the second term is the bond angle potential , with the spring constant @xmath364 kcal / mol .
the angle between the bonds @xmath365 and @xmath366 is @xmath367 , and the equilibrium value @xmath368 , a typical bond angle in protein structures @xcite .
the factor @xmath369 if @xmath143 , @xmath358 , @xmath263 are all on the same strand , 0 otherwise .
the relative softness of the bond angle potential , together with the form of the secondary structure interactions detailed below , ensure that the two strands in the unfolded lz26 ( with all inter - strand tertiary contacts broken ) have a persistence length of @xmath370 nm , consistent with experimental measurements @xcite . the third term in eq .
accounts for the effects of hydrogen bonding along the @xmath34-helical backbone , and is based on a similar form developed for rna @xcite . we mimic the directionality - dependence of hydrogen bonds by making the bond energy depend not only on the distance @xmath362 , but also on bond and dihedral angles defined by the four residues @xmath371 , @xmath143 , @xmath358 , and @xmath372 , with @xmath373 . for each @xmath352
there are two bond angles , @xmath374 and @xmath375 , and one dihedral angle , @xmath376 .
the equilibrium values of the angles , denoted by a superscript 0 , are calculated from the corresponding quantities in the native structure .
only when the distance , bond angles , and dihedral angles are all simultaneously equal to the equilibrium values does the hydrogen bond potential reach its energy minimum @xmath377 , where @xmath378 .
thus , the minimum is reached only when the entire @xmath138 strand segment adopts a structure resembling a single @xmath34-helical turn .
the @xmath34-helical propensity of an @xmath138 segment is determined by the energy scale @xmath379 and the sensitivity parameters @xmath380 , @xmath381 .
larger values for the sensitivity parameters increase the brittleness of the @xmath34-helix , making it more likely to be destablized due to thermal fluctuations . to calibrate the parameters , we define a helix function @xmath382 for any @xmath383 , @xmath384 reflecting the rms deviation of the bond distances and angles from their equilibrium values .
we use @xmath382 as a measure of helix content , by counting the fraction of pairs in @xmath354 where @xmath382 is less than a cutoff @xmath385 .
it is known from thermal denaturation experiments on gcn4 @xcite that the individual @xmath34 helices upon unzipping are unstable , with @xmath386 helical content .
in contrast , the tertiary contacts in the coiled - coil structure stabilize helix formation , resulting in a much higher helical content of @xmath387 .
we expect qualitatively similar behavior in lz26 in the case of force denaturation , and thus tune the sensitivity parameters to yield a large difference in the helix content between the unfolded and folded states .
the parameter values are set at @xmath388 kcal / mol , @xmath389 nm@xmath390 , @xmath391 rad@xmath390 .
for these values we find a helix content of 3% and 82% respectively for the unfolded and folded states of lz26 under a constant force of @xmath93 pn . the fourth term in eq
. describes tertiary interactions between the two strands of lz26 , @xmath392 .
these have a residue - dependent energy @xmath393 . here
@xmath394 is an overall prefactor , @xmath395 is the betancourt - thirumalai ( bt ) contact energy for residues @xmath143 and @xmath358 @xcite , and @xmath396 shifts the zero of the energy scale @xcite . to get a leucine zipper that unfolds at the experimental force scale of @xmath397 pn
, we choose @xmath398 and energy shift @xmath399 @xmath92 .
the tertiary interactions use a modified lennard - jones potential of the form : @xmath400 this has the standard 12 - 6 form at large distances , but a softer short - range repulsive core , increasing with the inverse 6th rather than 12th power .
the choice of the softer potential is made to allow for a longer simulation time step , while not having a significant impact on the large - scale dynamics of the system @xcite . the final term in eq .
describes purely repulsive interactions among the remaining non - neighboring pairs , @xmath401 , with energy factor @xmath402 kcal / mol and range @xmath403 nm .
we use the inverse 6th power in the repulsive potential for the same reasons as above .
each double - stranded dna handle is modeled as a chain of @xmath404 beads of radius @xmath316 nm , corresponding to a contour length @xmath405 .
the handle hamiltonian is : @xmath406 \end{split}\ ] ] where @xmath407 are the distances between neighboring beads , @xmath363 kcal / mol@xmath116nm@xmath117 , @xmath57 is the persistence length , and @xmath408 are angles between consecutive bonds . the two terms are stretching and bending energies respectively .
the handle elastic modulus @xmath409 pn . for the persistence length
we consider , @xmath114 nm , at the applied tension due to the traps , the handles ( and unfolded portions of the protein ) are almost fully extended , and there is negligible probability of the chain overlapping itself or protein residues in the vicinity of the handle attachment point .
hence , there is no need to include excluded volume interactions for the handles .
the covalent linkers that attach the handles either to the cysteine residue at the protein n - terminus or a point on the bead surface are modeled as simple harmonic springs with strength @xmath410 and length @xmath118 nm .
let @xmath411 be the mobility of a sphere of radius @xmath412 nm , where @xmath413 mpa@xmath116s is the viscosity of water at @xmath414 k. this will be the mobility of our dna handle beads , while for the large polystyrene beads the corresponding mobility is @xmath415 .
the rotational diffusion of the polystyrene bead is characterized by a mobility @xmath416 .
for the protein residues we choose a mobility @xmath417 , corresponding to an effective hydrodynamic radius of @xmath418 nm . the characteristic brownian dynamics time scale associated with @xmath419 is @xmath420 ns . to avoid numerical errors ,
our simulation time step @xmath421 should be a small fraction of @xmath422 , and we obtained reliable results using @xmath423 ps . for lz26 both with and without handles / beads , we ran @xmath96 long trajectories at various force conditions ( or trap separations ) , totaling to @xmath424 simulation time steps , or @xmath425 ms , with data collection every @xmath426 steps .
( in the case of the simulations involving the grm hairpin instead of the protein , the time step @xmath427 ps , and the total trajectory data for each grm parameter set corresponded to @xmath428 ms . )
before the data from optical tweezer experiments can be used to reconstruct the intrinsic biomolecule free energy landscape , one must consider the inevitable distortions due to noise , the electronic systems involved in data recording , and any additional filtering done as part of the collection protocol .
we have developed a method , finite bandwidth scaling ( fbs ) , to correct for these distortions . in the following we first derive the basic fbs scaling relations , and then verify them using both simulation and experimental data sets .
understanding how the time series of bead positions is distorted as part of the measurement process requires a detailed spectral analysis of all components in the dual optical tweezer apparatus .
the spectral properties of the experimental system used to collect the data in our work have been extensively characterized by von hansen _ et .
@xcite , allowing us to develop a simplified theory which approximates the most important sources of distortion .
our theory fits all the experimental data sets under consideration , but it can be easily modified to include additional complications that we ignore ( for example crosstalk between the two laser traps ) as well as the details of other experimental setups .
let @xmath429 be the trajectory of bead - bead separations along the @xmath36-axis recorded during the experiment .
this raw data set is based on the signal from the silicon photodiode devices that measure the deflection of the lasers due to bead displacements .
this output is then processed and amplified by the electronic system used in the recording apparatus .
if @xmath0 is the actual trajectory of bead displacements , inaccessible to the experimentalist , the recorded output @xmath429 is related to @xmath0 as : @xmath430 the deviation of @xmath429 from @xmath0 stems from two main effects : ( i ) an additive noise component @xmath431 , which includes environmental noise like vibrations of the optical elements in the apparatus and electronic noise in the detectors @xcite . for simplicity , we model the noise as gaussian white noise with zero mean and variance equal to @xmath432 : @xmath433 , @xmath434 , and @xmath435 , where @xmath436 denotes an equilibrium ensemble average ; ( ii ) convolution with a kernel function @xmath437 , which reflects the filtering properties of the photodiodes and electronics .
any additional time averaging or filtering carried out by the experimentalist on the recorded data series will be considered explicitly later on , and is not included in @xmath437 .
the analysis of ref .
@xcite yielded the following form for the filter kernel in the frequency domain , @xmath438\frac{1}{b_8(i \omega \tau_{bf})},\ ] ] where for our lot setup @xmath439 , @xmath440 @xmath126s , @xmath441 @xmath126s , and @xmath442 is the 8th order butterworth polynomial .
the term in the square brackets above originates in a physical phenomenon known as `` parasitic filtering ''
@xcite , arising from the transparency of the silicon in the photodiode to the laser light with wavelength 1064 nm used in the experiment : a fraction @xmath443 of the photocurrent from the detector is produced with a lag time @xmath444 relative to the photon signal .
the second term in eq . , involving the butterworth polynomial , is due to the subsequent electronic amplification of the signal from the detector , which acts like a butterworth lowpass filter with characteristic timescale @xmath445 , such that at the frequency @xmath446 the signal amplitude is attenuated by 3 db . since the form of eq .
is too complicated for use in our analytical theory , we will approximate @xmath437 as a generic first - order low - pass filter , exploiting the fact that both the parasitic and electronic terms act to attenuate high - frequency portions of the signal , @xmath447 where @xmath448 @xmath126s .
this effective filtering timescale @xmath449 is derived by demanding that eq .
exhibit the same degree of attenuation at @xmath446 as eq . .
though these distortions are expressed in the frequency domain , they have observable consequences for the equilibrium probability distribution of bead - bead separations . as an example , consider the raw autocorrelation function @xmath450 , where @xmath451 is the mean recorded bead - bead separation .
the variance of the raw probability distribution @xmath452 is equal to @xmath453 . from eqs . and
, the raw autocorrelation is related to the true one , @xmath454 , by : @xmath455 the first term in eq . , due to noise , tends to increase the variance @xmath453 relative to @xmath456 .
the second term , due to filtering , is always less than @xmath456 , since it is an average over @xmath457 , and @xmath458 .
noise broadens the measured distribution , and filtering narrows it .
however without knowing the amplitude of the noise @xmath432 , it is unclear whether the filtering due to the detectors and electronics under- or over - compensates for the noise , and how far @xmath452 deviates from the true distribution @xmath24 .
thus , we need a way to estimate @xmath432 .
the situation is even more complicated since the experimentalist may choose to apply additional filtering on the recorded data , for example as a way of manually removing noise and unwanted high frequency components of the signal ( since the dynamics of interest typically occur at frequencies much lower than imposed filter cutoff ) . for the gcn4 leucine zipper
, the data sets recorded at 100 khz ( corresponding to a sampling time step @xmath459 @xmath126s ) were subsequently filtered in real time during collection by averaging every 5 consecutive time steps together .
such averaging acts like a low - pass filter , and so has narrowing effects on the equilibrium probability distribution qualitatively similar to the filtering described above .
some type of additional filtering of this kind is a common experimental practice ( see refs .
@xcite,@xcite , and @xcite for recent examples , involving either an averaging or 8 pole bessel filter ) .
it turns out , however , that we can take advantage of the filtering protocol : by varying the degree of filtering we will use it to approximately extrapolate features of the true probability distribution . let us concentrate on the simple case of filtering the recorded data by averaging every @xmath248 consecutive points into a single value .
if the collection time step is @xmath460 , the original raw data is represented by the recorded time series @xmath461 , where @xmath462 for @xmath463 .
the averaged data is a time series @xmath464 , where @xmath465 .
for the averaged time series we will focus on two quantities , both related directly to its autocorrelation @xmath466 : the variance @xmath467 , and the mean - squared displacement ( msd ) between consecutive points , @xmath468 . in a more complicated fashion ,
these two quantities can also be expressed in terms of the original autocorrelation @xmath469 before averaging : @xmath470 we know that @xmath471 is related to the unknown true correlation @xmath472 through eq .
, so we can complete the theoretical description by specifying a form for @xmath472 .
a generic correlation function can be expanded as a sum of exponentials , @xmath473 , with relaxation times @xmath474 .
we will be interested in correlations on the shortest accessible time - scales , @xmath475 , so we plug the expression for @xmath472 into eq . and expand for small @xmath1 , keeping the contribution from the @xmath476 exponential and lowest order corrections from the @xmath477 terms : @xmath478 where @xmath479 , @xmath480 .
if necessary , the expansion can be extended to higher orders , but the above form was sufficient to fit all the simulation and experimental cases which we analyze below .
- completely define the variance @xmath481 and msd @xmath482 in terms of five unknown parameters : @xmath432 , @xmath483 , @xmath476 , @xmath484 , and @xmath485 . by averaging the recorded time series @xmath486 for different values of @xmath248 ( varying the effective filter bandwidth )
, we construct curves of @xmath481 and msd @xmath482 as a function of @xmath248 .
fitting these curves to eqs . -
, we can then extract the unknown parameters .
this allows us to estimate the true variance of the probability distribution , @xmath487 since we are using properties of time series at different effective bandwidths to gain information about the true , `` infinite '' bandwidth limit , we call our method _ finite bandwidth scaling _ ( fbs )
. the analogy is to finite size scaling @xcite , where thermodynamic properties of systems on finite lattices are extrapolated to the infinite lattice limit .
one of the nice features of fbs is that the scaling analysis can be carried out even when we can only calculate @xmath481 and @xmath482 for a subset of @xmath248 values .
for example , in the leucine zipper case below , the available time series corresponds to @xmath488 , since the data was time averaged during collection . from the @xmath488 data we can construct trajectories for @xmath489 .
this subset is sufficient for the fbs extrapolation .
once we know @xmath456 , how can we use it to approximately reconstruct the true distribution @xmath24 ? keep in mind that the variance @xmath490 .
the simplest estimate for @xmath24 is to start with the measured , averaged distribution @xmath491 for some @xmath248 , and deform it in one of two ways , changing its variance by an amount @xmath492 : ( i ) if @xmath493 , we carry out a convolution with a normalized gaussian of variance @xmath494 , @xmath495 ( ii ) if @xmath496 , we do a deconvolution instead , solving @xmath497 for @xmath71 .
the latter can be carried out using the numerical deconvolution technique described in sec .
[ sec : numdec ] .
after the deformation , the estimated @xmath71 will by construction have the correct variance @xmath456 .
we should recover roughly the same @xmath71 starting from @xmath498 for any @xmath248 in the range where the fbs scaling is valid , as we will demonstrate in the examples below . in systems with multiple states , where there is more than one peak in the measured distribution
, it is more accurate to carry out the fbs analysis separately on each state , and apply the corresponding specific deformation for each peak .
this can be done with the aid of hidden markov model @xcite partitioning of the time series , as described in the next section for the case of the grm and leucine zipper .
the fbs method has several limitations : ( i ) using a gaussian to deform @xmath491 into @xmath71 is an assumption , since all we strictly know about the actual point - spread function is the variance @xmath494 .
the smaller the variance , the more valid the assumption , since the potential non - gaussian contributions to the point - spread function become less significant . we can also test the assumption from our measured data , by checking whether the @xmath491 for various @xmath248 can be mapped to each other by gaussian deformations . for all the systems analyzed in our work
this is indeed the case .
( ii ) gaussian deformations map individual peaks into slightly broader or narrower peaks , but do not produce new peaks .
hence , if there is a state with a very short lifetime that is smeared out by the filtering ( either the parasitic , electronic , or additional filtering ) , yielding no distinct peak in @xmath491 , the fbs method will not be able to reconstruct its properties .
whether or not the experimentalist chooses to do additional averaging , the intrinsic time resolution @xmath449 of the apparatus puts fundamental constraints on what we can learn from the measured time series .
transitions occurring on timescales faster than @xmath449 will be lost to us .
( iii ) in a similar way , the characteristic relaxation times of the trapped beads also impose limits . to illustrate this ,
take two protein states @xmath499 and @xmath500 , which have a small difference in their mean end - to - end distance along the force direction , and assume @xmath499 is only accessible from @xmath500 . if the mean lifetime of @xmath499 is much smaller than @xmath500 , such that it is shorter than the bead relaxation time , transitions like @xmath501 will correspond to only negligible excursions in the measured trajectory of bead displacements , since the beads do not have enough time to relax to the equilibrium position associated with @xmath499 before the protein returns to @xmath500 .
if the mean - squared distance of the excursions is smaller than the noise amplitude in the recorded time series , the existence of state @xmath499 will be hidden from the experimentalist , regardless of the apparatus filtering timescale @xmath449 . in summary ,
the distribution produced by fbs is an approximation to the truth : the method can correct distortions produced by noise and filtering , but it only works for states in the energy landscape which leave some signature of themselves in the measured time series .
@xmath432 & @xmath483 & @xmath476 & @xmath484 & @xmath485 + & [ @xmath126s ] & [ @xmath126s ] & [ nm@xmath502s ] & [ nm@xmath117 ] & [ @xmath126s ] & [ nm@xmath117 ] & [ nm@xmath117/@xmath126s ] + + state @xmath301 & 0.024 & 0 & 0 & @xmath504 & @xmath505 & @xmath506 & @xmath507 + state @xmath508 & 0.024 & 0 & 0 & @xmath509 & @xmath510 & @xmath511 & @xmath512 + + @xmath513 pn & 10 & 7 & @xmath514 & @xmath515 & @xmath516 & @xmath517 & @xmath518 + @xmath519 pn & 10 & 7 & @xmath514 & @xmath520 & @xmath521 & @xmath522 & @xmath523 + @xmath524 pn & 10 & 7 & @xmath514 & @xmath525 & @xmath526 & @xmath527 & @xmath528 + + state i1 & 10 & 7 & @xmath514 & @xmath529 & @xmath530 & @xmath531 & @xmath532 + state i2 & 10 & 7 & @xmath514 & @xmath533 & @xmath534 & @xmath535 & @xmath536 + state u & 10 & 7 & @xmath514 & @xmath537 & @xmath538 & @xmath539 & @xmath540 + + state i1 & 10 & 7 & @xmath514 & @xmath541 & @xmath542 & @xmath531 & @xmath543 + state i2 & 10 & 7 & @xmath514 & @xmath544 & @xmath545 & @xmath546 & @xmath547 + state u & 10 & 7 & @xmath514 & @xmath548 & @xmath549 & @xmath517 & @xmath550 + nm ; all other parameters as in table s1 ) . (
a ) the probability distribution of the bead - bead separation from the raw simulation data , @xmath551 ( gray ) , and the decomposition into individual gaussian peaks corresponding to the n ( blue ) and u ( red ) states .
distances are measured with respect to @xmath291 , the mean bead - bead separation .
( b ) a sample time series fragment from the simulation , with the individual data points colored according to their assignment to the n ( blue ) and u ( red ) states by hidden markov model analysis .
( c ) for the raw time series filtered by averaging together every @xmath248 data points , the variance @xmath481 as a function of @xmath248 .
the time series corresponding to each state , n and u , is analyzed separately , and plotted as blue and red points respectively , with bars denoting standard error due to finite sampling .
the solid curves are the fbs theoretical fits to @xmath481 [ eq . ] .
best - fit fbs parameters are listed in table s2 .
( d ) analogous to ( c ) , except showing the msd function @xmath482 for consecutive pairs of points in the averaged time series .
( e ) the raw distributions @xmath552 for the averaged time series at @xmath553 , 50 , 100 , 150 .
( f ) solid curves : the distribution @xmath554 estimated by applying the appropriate fbs correction to the raw distributions in ( e ) .
there are four curves , but due to overlapping they appear as one .
points : the raw distribution @xmath491 for @xmath553 ( no averaging ) , which for the grm case is the true distribution , since there are no noise or apparatus filtering effects in the simulation . ] as a first test of the fbs theory , we analyze a brownian dynamics simulation trajectory of the grm model in an optical tweezer setup ( fig . [ grm_avg ] ) .
the trap separation @xmath555 nm , and all the other parameters are listed in table s1 .
a computer simulation has perfect recording of data , with no environmental noise or apparatus filtering effects , hence it can test the fbs theory of eqs . - in the limit @xmath556 . in this case
the true distribution is just the @xmath553 raw distribution @xmath557 , plotted in fig .
[ grm_avg](a ) ( gray curve ) .
if the fbs scaling is valid , we should be able to map any distribution for @xmath558 onto the @xmath553 result by applying the fbs correction procedure described above .
the grm model exhibits two states , native n and unfolded u , which have distinct dynamical properties .
hence it is more accurate to apply the fbs method separately to just those portions of the time series belonging to each state . partitioning the time series by state requires estimating the most likely sequence of states that corresponds to the data .
hidden markov modeling ( hmm ) @xcite is a general tool for this task .
the probability distribution can be accurately decomposed into gaussians corresponding to each state , as depicted in fig .
[ grm_avg](a ) , which define likelihoods for any observed @xmath22 data point in the trajectory to belong to one or the other state .
we then process the entire trajectory through the baum - welch algorithm @xcite , to find optimal values for the unknown transition probabilities between states , and finally construct the most likely state sequence using the viterbi algorithm @xcite .
[ grm_avg](b ) shows a fragment of the trajectory , colored according to the state assignment resulting from hmm . the variance @xmath481 and msd @xmath482 are then calculated as a function of @xmath248 from the trajectory fragments belonging to a certain state . for a given @xmath248 ,
the calculation involves averaging over data points within a time window up to @xmath559 in length , so getting good statistics requires having many fragments longer than @xmath559 .
this will be true so long as @xmath559 is much smaller than the mean lifetime of the state , putting a practical upper bound on @xmath248 .
[ grm_avg](c ) and ( d ) plot the results for @xmath481 and @xmath482 respectively ( blue points : n state ; red points : u state ) .
bars represent statistical standard error due to finite sample size , as determined through jackknife estimation @xcite .
the solid curves are fits to eqs . - ,
from which we extract the fbs parameter values listed in table s2 . with these values in hand , we can carry out the correction procedure : fig .
[ grm_avg](d ) shows the raw distributions @xmath498 for @xmath560 ( solid curves ) , and panel ( e ) shows that same distributions after they have been corrected according to the method outlined above ( @xmath553 needs no correction , but is included for comparison )
. the greater the degree of averaging ( increasing @xmath248 ) , the narrower the peaks in @xmath498 .
however , the fbs method compensates for this , and all the distributions in ( e ) have collapsed onto a single estimate for the true @xmath71 . as expected
, this estimate agrees very well with the @xmath553 result @xmath561 ( cyan points ) . for apparatus parameters ) .
we analyze three separate trajectories at different constant trap separations , corresponding to mean forces @xmath562 , @xmath289 , and @xmath290 pn .
the results in each panel are labeled by the @xmath26 value of the trajectory .
( a ) for the raw experimental time series filtered by averaging together every @xmath248 data points , the variance @xmath481 as a function of @xmath248 .
the results are plotted as points , with bars denoting standard error due to finite sampling .
the solid curves are the fbs theoretical fits to @xmath481 [ eq . ] .
best - fit fbs parameters are listed in table s2 .
( b ) analogous to ( a ) , except showing the msd function @xmath482 for consecutive pairs of points in the averaged time series .
( c ) the raw distributions @xmath563 for the averaged time series at @xmath553 , 2 , 5 , 10 .
each row corresponds to a different trajectory .
( d ) solid curves : the distribution @xmath71 estimated by applying the appropriate fbs correction to the raw distributions in ( c ) . for each trajectory
there are four curves , but due to overlapping they appear as one .
dashed curves : the raw distribution @xmath491 for @xmath553 .
though this distribution is free of any additional time averaging carried out on the recorded time series , it is subject to parasitic and electronic filtering effects intrinsic to the apparatus .
these distortions are corrected by fbs , and hence the dashed and solid curves are distinct . ]
the second test of the fbs theory is on experimental data for a system with only dsdna handles and beads , discussed in sec .
[ sec : ps ] .
fbs results for three different trajectories ( corresponding to three values of the mean force @xmath26 ) are presented in fig .
[ hand_avg ] .
these data sets were recorded with a sampling rate of 100 khz ( @xmath564 @xmath126s ) , with no additional averaging beyond the unavoidable filtering effects of the detectors and electronics . as a consequence of these effects ,
@xmath565 is not the same as the true distribution , and the deviation grows larger as @xmath248 is increased .
the fbs best - fit results are shown in table s2 . in the fitting the noise amplitude @xmath432
is constrained to be the same among all three trajectories , since they are all collected on the same equipment .
like in the previous example , @xmath498 for various @xmath248 can all be collapsed onto a single estimate for @xmath71 through the fbs method . in fig .
[ hand_avg](e ) this estimate ( solid curves ) is compared to @xmath565 ( dashed curves ) , to emphasize that the distortions due to apparatus filtering are small but noticeable .
( gray ) , and the decomposition into individual gaussian peaks corresponding to the i1 ( red ) , i2 ( green ) , and u ( blue ) states .
distances are measured with respect to @xmath566 , the position of the i1 peak .
( b ) a sample time series fragment from the experiment , with the individual data points colored according to their assignment to the i1 ( red ) , i2 ( green ) , and u ( blue ) states by hidden markov model analysis . ( c ) for the raw time
series filtered by averaging together every @xmath248 data points , the variance @xmath481 as a function of @xmath248 .
the time series corresponding to each state is analyzed separately , and plotted as points in distinct colors , with bars denoting standard error due to finite sampling .
the solid curves are the fbs theoretical fits to @xmath481 [ eq . ] .
best - fit fbs parameters are listed in table s2 .
( d ) analogous to ( c ) , except showing the msd function @xmath482 for consecutive pairs of points in the averaged time series .
( e ) the raw distributions @xmath563 for the averaged time series at @xmath488 , 10 , 15 , 20 , 25 .
( f ) solid curves : the distribution @xmath567 estimated by applying the appropriate fbs correction to the raw distributions in ( e ) .
there are five curves , but due to overlapping they appear as one .
dashed curve : the raw distribution @xmath491 for @xmath488 . ]
the final test is on the gcn4 leucine zipper experimental time series ( trajectory 1 , with parameters described in table s1 ) . as mentioned earlier , here we only can construct averaged data sets for @xmath568 : the @xmath553 trajectory , at the original @xmath564 @xmath126s sampling interval , is not available . despite this limitation ,
the fbs scaling analysis works nicely , with results summarized in fig .
[ riefa_avg ] and table s2 .
we took advantage of the fact that the handle - only data sets , collected with the same optical tweezer setup as the leucine zipper ( except with no protein ) , had direct information about @xmath553 timescales , and thus probed higher frequencies than were accessible in the leucine zipper data .
since going to higher frequencies gives us better estimates of the background noise , we set the noise amplitude @xmath432 in the leucine zipper case to the best - fit value from the handle - only analysis .
all other fbs parameters were fit individually for each state ( i1 , i2 , and u ) in the leucine zipper distribution . with fbs corrections ,
@xmath491 for @xmath248 going up to 25 ( effective bandwidths as low as 4 khz ) all collapse onto a single estimate of the true @xmath71 .
the free energy reconstruction is only as good as the data on which it is based : the recorded time series which is the input , and the information about the apparatus which is used to analyze the time series and predict the intrinsic landscape .
both of these are subject to uncertainties , which will propagate into the final result .
let us first consider statistical uncertainties due to the finite length of the trajectories from which the input probability distribution @xmath561 is determined .
one of the advantages of the double optical trap setup is that it is exceptionally stable , allowing for data collection over periods @xmath569 s. in the case of the leucine zipper , the slowest transition ( from u to i2 ) occurs on timescales of @xmath107 s in the force range of interest , so even a single trajectory contains @xmath570 of the rarest observed conformational changes . .
( a - b ) probability distributions @xmath551 of the raw time series for total bead - bead separation ( black curves ) collected during two experimental runs ( left / right columns ; see table s1 for parameters ) .
distances are measured with respect to @xmath566 , the position of the i1 peak .
the dark blue band corresponds to the standard error ( 68% confidence interval ) for each point in the distribution , due to finite sampling .
( c - d ) the corresponding intrinsic protein free energy @xmath43 ( black curves ) , as calculated using the procedure described in the main text .
the free energies are in the constant force ensemble , at the mid - point force value @xmath26 where the probability of being in states i1 and u is equal ( @xmath571 pn for run 1 , @xmath572 pn for run 2 ) .
the dark blue band represents standard error ( 68% confidence interval ) including just the uncertainty due to finite sampling ; the wider light blue band is the standard error including all sources of uncertainty ( sampling and apparatus parameters ) . ]
thus the distribution @xmath551 has small statistical uncertainties .
to quantitatively estimate the error , we use a block bootstrap method @xcite in the following manner : the trajectory is divided into blocks of length larger than the longest autocorrelation time , and a synthetic data set of the same length is generated by sampling with replacement from this set of blocks . using a large number of synthetic data sets ( @xmath573 )
we can determine confidence intervals for each point in the @xmath561 distribution .
the number of blocks is varied until convergence is achieved in the error estimate .
the results are shown in fig .
[ err](a - b ) for two leucine zipper experimental trajectories ( parameters as in table s1 ) .
the @xmath551 distributions ( black curves ) are surrounded by dark blue bands which represent the 68% confidence interval , or standard error margin . the median error in the @xmath11 range where @xmath574 , is 10% and 19% respectively for the two trajectories . in reconstructing the intrinsic free energy landscape @xmath99 ,
this statistical error is compounded by uncertainties in all the apparatus parameters that are used in the analysis : bead radii , trap strengths , handle properties , as listed in table s1 , as well as uncertainties in the fbs parameters used to correct the raw distributions ( table s2 ) .
we perform a monte carlo error estimate , by sampling from gaussian distributions of these parameters with standard deviations given by the uncertainties , and for each parameter set performing the complete free energy reconstruction on the entire ensemble of synthetic data sets generated by the block bootstrap . in order to analyze the shape differences among the reconstructed landscapes ,
every @xmath43 is projected to the mid - point value of @xmath26 where the probabilities of states i1 and u are equal .
( @xmath571 pn and @xmath101 pn from trajectories 1 and 2 respectively . )
though computationally intensive , this procedure allows us to estimate 68% confidence intervals for @xmath43 shown as light blue bands for the two trajectories in fig .
[ err](c - d ) . for comparison ,
if one assumed no uncertainty in the apparatus parameters , one would get the much narrower dark blue bands , representing just the error in @xmath99 from the finite sampling of @xmath561 .
clearly , the uncertainties in the apparatus parameters are the predominant source of error . with both apparatus and sampling uncertainties included , the median standard error over the @xmath11 range where @xmath575 is @xmath576 in both trajectories .
this corresponds to @xmath577 deviations in the shape of the landscape .
the median difference between @xmath43 estimated from the two trajectories in this range is @xmath578 , and hence our free energy analysis gives a consistent result , within standard error , between the two different experimental runs .
klimov dk , betancourt mr , thirumalai d ( 1998 ) virtual atom representation of hydrogen bonds in minimal off - lattice models of alpha helices : effect on stability , cooperativity and kinetics .
_ fold . des . _ 3:481496 .
betancourt mr , thirumalai d ( 1999 ) pair potentials for protein folding : choice of reference states and sensitivity of predicted native states to variations in the interaction schemes . _ protein sci .
_ 8:361369 .
berg - sorensen k , peterman ejg , weber t , schmidt cf , flyvbjerg h ( 2006 ) power spectrum analysis for optical tweezers .
ii : laser wavelength dependence of parasitic filtering , and how to achieve high bandwidth . _
_ 77:063106 .
ritchie db , foster dan , woodside mt ( 2012 ) programmed-1 frameshifting efficiency correlates with rna pseudoknot conformational plasticity , not resistance to mechanical unfolding .
usa _ 109:1616716172 . | in single molecule laser optical tweezer ( lot ) pulling experiments a protein or rna is juxtaposed between dna handles that are attached to beads in optical traps .
the lot generates folding trajectories under force in terms of time - dependent changes in the distance between the beads .
how to construct the full intrinsic folding landscape ( without the handles and the beads ) from the measured time series is a major unsolved problem . by using rigorous theoretical methods which account for fluctuations of the dna handles , rotation of the optical beads , variations in applied tension due to finite trap stiffness , as well as environmental noise and the limited bandwidth of the apparatus
we provide a tractable method to derive intrinsic free energy profiles .
we validate the method by showing that the exactly calculable intrinsic free energy profile for a generalized rouse model , which mimics the two - state behavior in nucleic acid hairpins , can be accurately extracted from simulated time series in a lot setup regardless of the stiffness of the handles .
we next apply the approach to trajectories from coarse grained lot molecular simulations of a coiled - coil protein based on the gcn4 leucine zipper , and obtain a free energy landscape that is in quantitative agreement with simulations performed without the beads and handles .
finally , we extract the intrinsic free energy landscape from experimental lot measurements for the leucine zipper , which is independent of the trap parameters . the energy landscape perspective has provided a conceptual framework to describe how rna @xcite and proteins @xcite fold .
some of the key theoretical predictions , such as folding of proteins and rna by the kinetic partitioning mechanism @xcite and the diversity of folding routes @xcite , have been confirmed by a number of experiments @xcite .
more refined comparisons require mapping the full folding landscape of biomolecules , which has been difficult to achieve .
the situation has dramatically changed with advances in laser optical tweezer ( lot ) experiments , which have been used to obtain free energy profiles as a function of the extension of biomolecules under tension @xcite . the usefulness of the lot technique , however , hinges on the crucial assumption that information about the fluctuating biomolecule can be accurately recovered from the raw experimental data , namely the time - dependent changes in the positions of the beads in the optical traps , attached to the biomolecule by double - stranded dna handles [ fig .
[ sys ] ] .
thus , we only have access to the intrinsic folding landscape of the biomolecule ( in the absence of handles and beads ) indirectly through the bead - bead separation along the force direction . many extraneous factors , such as fluctuations of the handles @xcite , rotation of the beads , and the varying applied tension due to finite trap stiffness , can severely distort the intrinsic folding landscape .
moreover , the detectors and electronic systems used in the data collection have finite response times , leading to filtering of high frequency components in the signal @xcite .
ad hoc attempts have been made to account for handle effects based on experimental estimates of stretched dna properties , employing techniques similar to image deconvolution @xcite .
theory has been used to extract free energy information from nonequilibrium pulling experiments @xcite , and to determine the intrinsic power spectrum of protein fluctuations @xcite from lot data . however , to date there has been no comprehensive theory to model and correct for all the systematic instrumental distortions of the underlying folding landscapes of proteins and rna . a crucial unsolved problem is how can one construct the intrinsic free energy profile of a biomolecule using the measured folding trajectories in the presence of beads and handles ( the total separation @xmath0 in fig .
[ sys ] as a function of time @xmath1 ) . here
, we solve this problem using a rigorous theoretical procedure .
besides @xmath0 , the only input needed in our theory are the bead radii , the trap strengths and positions , and handle characteristics such as the contour length , the persistence length , and the elastic stretch modulus .
the output is the intrinsic free energy as a function of the biomolecular extension ( @xmath2 in fig .
[ sys ] ) in the constant force ensemble .
we validate our approach using two systems : ( i ) a generalized rouse model ( grm ) hairpin @xcite , which has an analytically solvable double - well energy landscape under force ; this allows a direct test of the method ; ( ii ) a double - stranded coiled - coil protein based on the yeast transcriptional factor gcn4 leucine zipper domain , whose folding landscape was studied using a lot experiment @xcite .
we first use coarse - grained molecular simulations to obtain the intrinsic free energy landscape of the isolated protein at a constant force .
we then simulate mechanical folding trajectories using the full lot setup , from which we quantitatively recover the intrinsic free energy landscape of gcn4 , thus further establishing the efficacy of our theory .
finally , we apply our theory to experimentally generated data , and show that we can get reliable estimates for the protein energy profile independent of the optical trap parameters . |
the problem of nonlinear dynamics in double well potential has been first addressed by jensen @xcite who considered light power spatial oscillations in two coupled nonlinear waveguides , which resemble josephson oscillations @xcite in the _ spatial domain_. the latter macroscopic quantum tunneling effect , originally discovered in superconducting junctions , is caused by the global phase coherence between electrons in the different layers .
more recently the similar realization of a bosonic josephson junction has been reported for a bose - einstein condensate embedded in a macroscopic double harmonic well potential @xcite .
the difference with the ordinary josephson junction behavior is that the oscillations of atomic population imbalance are suppressed for high imbalance values and a self - trapping regime emerges @xcite .
the nonlinear dynamics of bosonic junctions , described by the gross - pitaevskii equation ( gpe ) @xcite , is usually mapped to a simpler system characterized by two degrees of freedom ( population imbalance and phase difference ) while the nonlinear properties of the wave function within the single well are neglected . in this approach
the symmetric and antisymmetric stationary solutions of gpe are used as a basis to build a global wave function @xcite .
this description allows to show that for higher nonlinearities the symmetric solutions become unstable and degenerate to an asymmetric stationary ( approximate ) solution of the gpe corresponding to a new self - trapping regime @xcite . on the other hand ,
considering the double square well potential ( instead of harmonic one ) , we discover that , in a wide range of nonlinearities , the system can either remain trapped mostly in one of the wells , or swing periodically from right to left and back .
the switching from one state to the other is triggered by a slight local variation of the potential barrier between the wells .
the _ coexistence _ of oscillatory and self - trapping regimes corresponds to the simultaneous presence of josephson oscillations and of an asymmetric solution of the gpe .
our result differs from known behaviors of bosonic josephson junctions , where the presence of oscillatory or self - trapping regimes is uniquely determined by the parameters of the system .
the resulting switching property should have a straightforward experimental realization in waveguide arrays , which constitute truly one - dimensional systems and are particularly convenient for the observation of nonlinear effects @xcite and in engineered optical lattices of bose - einstein condensates @xcite .
let us write gpe with double square well potential as follows : @xmath0 where @xmath1 is the double square well ( represented in fig .
[ fig:0.25 ] ) with a total width @xmath2 and the potential barrier height and width @xmath3 and @xmath4 , respectively .
the stationary solution of are sought as @xmath5 with a real - valued function @xmath6 found in terms of jacobi elliptic functions @xcite @xmath7 , \nonumber\\ l < x < l\ : \ & \phi = a\ { { \rm cn}}[\gamma_a(x - l)+{\mathbb k}(k_a),k_a ] , \label{4}\\ -l
< x < l\ : \ & \phi = c\ { { \rm dn}}[\gamma_c(x - x_0),k_c ] , \nonumber\end{aligned}\ ] ] with the parameters given in terms of the amplitudes by @xmath8 where @xmath9 denotes the complete elliptic integral of the first kind and by construction the above expressions verify the vanishing boundary values in @xmath10 .
the solutions are then given in terms of five parameters ( @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 ) , four of which are determined by the continuity conditions in @xmath16 . thus the conserved total injected power ( nonlinearity parameter ) @xmath17 completely determines the solutions .
another useful conserved quantity is the total energy @xmath18 given by @xmath19 in the weakly nonlinear limit ( small @xmath20 ) , the solutions are symmetric ( odd or even ) .
the even solution @xmath21 corresponds to @xmath22 in , while odd solution @xmath23 corresponds to @xmath24 . for higher powers , namely above a threshold value , an asymmetric solution @xmath25 also exists for which @xmath26 .
these analytical solutions are represented in fig .
[ fig:0.25 ] .
to plot the solutions we stick with the following parameter values : the width of the rectangular double well potential is @xmath27 , the barrier width is @xmath28 and its height is @xmath29 . we derive the complete set of solutions and display the dependence of their amplitudes on the total power @xmath17 in the main plot of fig .
[ fig : bist ] . below the threshold value
@xmath30 only the symmetric ( odd and even ) solutions exist and their amplitudes almost superpose . at the threshold value a new solution appears which is asymmetric with amplitudes @xmath11 and @xmath12 in the two wells , respectively , represented by the upper and lower branches in fig . [
fig : bist ] .
the regime of josephson oscillation is usually understood on the basis of _ coupled mode _ approach as follows . using the symmetric and antisymmetric solutions , one builds a variational anzatz by seeking the solution @xmath31 under the form @xmath32 the functions @xmath33 and @xmath34 are interpreted as the _ probabilities _ to find the system localized either on the left or on the right part of the double square well . by construction ,
the overlap of @xmath35 with @xmath36 is negligible , consequently , the projection of the gpe successively on @xmath35 and @xmath36 provides the coupled mode equations @xcite @xmath37 with coupling constant @xmath38 and nonlinearity parameter @xmath39 defined by @xmath40dx } { \int\phi_1 ^ 2\,dx } , \quad d=\frac { \int\phi_1 ^ 4\,dx}{\int\phi_1 ^ 2\,dx}.\ ] ] an explicit solution of in terms of jacobi elliptic functions has been found in @xcite and used in bose - einstein condensates in @xcite .
it is a good approximation for the system in a double harmonic potential well @xcite and correctly describes the oscillatory regime in our case . indeed , when the power is initially injected into one array , say @xmath41 , @xmath42 , we obtain for @xmath43 @xmath44,\quad \label{7a}\ ] ] since @xmath45 oscillates around the value @xmath46 , this expression describes an oscillation of light intensity between the left and the right wells .
the period of this oscillation is @xmath47 and has been checked on various numerical shots at different total input power . in summary , while the self - trapping regime is directly interpreted in terms of the asymmetric solution , the interpretation of the josephson oscillation regime needs to call to the coupled mode approach , which in turn fails to explain the observed _ coexistence _ of both regimes .
such a coexistence , however , is understood in terms of the energy which can be evaluated , at given total power @xmath20 , both for the symmetric solution @xmath48 and for the asymmetric solution @xmath49 . as shown in the inset of fig .
[ fig : bist ] these two energy values @xmath50 and @xmath51 turn out to be very close up to the total power value @xmath52 .
consequently , switching from a regime to the other is allowed at fixed power .
in particular , in the numerical experiments of fig . [
fig : oscpin ] , total power and energy are the same before and after the local variation of the potential barrier value . it is worth to remark that a similar analysis in the case of harmonic double well potentials @xcite shows that the energy of the asymmetric solution ( when this solution exists ) is significantly smaller than the energy of the symmetric solution .
in such a situation , it is thus impossible to switch from a self - trapped state to an oscillatory regime when keeping both the energy and the total power constant .
now our aim in this section is to suggest the realistic experiments on bec and waveguide arrays , which are engineered in such a way to mimic two weakly coupled chains of jj s ( see fig .
[ becexp ] ) . using such an experimental set - up
, we demonstrate the feasibility of the efficient control of a switch between oscillating and self- trapping states of the systems .
we show that our problem reduces to the gross - pitaevskii equation ( gpe ) @xcite in a double square well , which displays very different properties from the previously considered double harmonic well potential @xcite .
our results are broadly applicable and open the way to the experimental study of these phenomena in the dynamics of those systems .
we start the consideration of the case of bec in an optical lattice , for which a one - dimensional hamiltonian has a following form : @xmath53 where @xmath54 is atomic mass , @xmath55 is the scattering length corresponding to the attractive atom - atom interactions and @xmath56 is the transversal oscillation length , which implicitly takes into account the real three dimensionality of the system @xcite , @xmath57 being the transversal frequency of the trap .
the optical lattice potential is @xmath58 where @xmath59 is the wavenumber of the laser beams that create the optical lattice and @xmath3 is the height of the additional spatial energy barrier placed in the middle of the optical lattice . besides that ,
dirichlet boundary conditions with @xmath60 are chosen in order to describe the large confining barriers at both ends of the bec .
these boundary conditions could be realized experimentally by an additional optical lattice with larger amplitude and larger lattice constant , as shown in fig.[becexp ] .
introducing a dimensionless length scale @xmath61 and time @xmath62 , where @xmath63 and @xmath64 is the recoil energy @xcite , we can rewrite as follows @xmath65 where the normalized wave - function , @xmath66 , is introduced @xcite .
the dimensionless potential @xmath67 still has the form with the following dimensionless depths of the optical lattice @xmath68 @xmath69 being the dimensionless nonlinearity parameter .
we have performed numerical simulations of eq . with 12 wells ( 6 wells on each side of the barrier as presented in fig .
[ becexp ] ) and the parameters @xmath70 ( in physical units this means that the depth of the optical lattice is @xmath71 ) , @xmath72 and we fix the nonlinearity to the value @xmath73 , i.e. , we choose attractive interactions .
the dynamics is similar for repulsive interatomic forces ( see the discussion below ) .
the phenomenon we study in this letter does not depend significantly on the actual size of the system , if at least 3 lattice sites are present at each side of the barrier . as seen from the left panel of fig .
[ becosc ] , if one prepares the condensate in a self - trapped state it remains there until we apply the pulselike time variation of the barrier displayed in the inset .
after that action , the condensate goes into the oscillating tunneling regime . on the other hand , preparing the condensate in the oscillating tunneling regime ( right graph in fig .
[ becosc ] ) one can easily arrive at a self - trapping state by varying again the energy barrier in the middle as displayed in the inset .
let us mention that , as far as the energy of the barrier is changed adiabatically , the total energy of the condensate does not vary , i.e. the self - trapped and tunneling oscillatory regimes have the _ same _ energy .
this is quite different from what happens in a double harmonic well potential @xcite .
the point is that , in the double harmonic well , the asymmetric stationary solution is characterized by a smaller energy than the symmetric solution and this difference increases sharply with increasing nonlinearity .
hence , a drastic energy injection is required in order to realize the transition between the two regimes ; whilst in our case the transition is simply achieved only by varying pulsewise the energy barrier . below
we argue that this happens because our case effectively reduces to the case of a double square well potential ( see the inset of fig . [ becexp ] and the reduction procedure below ) for which asymmetric and symmetric stationary solutions carry almost the same energies in a wide range of the nonlinearity parameter .
now we proceed to reducing eq . to a discrete nonlinear schrdinger equation ( dnls ) .
we discretize it via a tight - binding approximation @xcite , representing the wave function @xmath74 as @xmath75 where @xmath76 is a normalized isolated wave function in an optical lattice in the fully linear case @xmath77 and could be expressed in terms of wannier functions ( see , e.g. , @xcite ) . for clarity ,
we use here its approximation for a harmonic trap centered at the points @xmath78 ( @xmath79 varies from 1 to @xmath80 , the number of wells ) . in the context of the evolution equation
@xmath76 has the form @xmath81 for @xmath82 , and one should substitute @xmath83 by @xmath84 in the above expression in order to get an approximate formula for the wave function for @xmath85 .
assuming further that the overlap of the wave functions in neighboring sites is small , we get from the following dnls equation for the sites @xmath82 @xmath86 while for @xmath85 we have @xmath87 where we assume pinned boundary conditions .
the constants @xmath88 , @xmath89 , @xmath90 and @xmath91 are easily computed from the following expressions ( @xmath92 ) : @xmath93d\tilde x , \nonumber \\
q_1&=&-\int\left[\frac{\partial\phi_1}{\partial \tilde x}\frac{\partial\phi_{-1}}{\partial \tilde x}+(\tilde v+\tilde v_0)\cos^2(\tilde x/2)\phi_1\phi_{-1}\right]d\tilde x , \nonumber \\
u&=&g\int\phi_j^4\,d\tilde x\simeq u_1=g\int\phi_{\pm 1}^4\,d\tilde x \;.
\label{9 } \end{aligned}\ ] ] in order to characterize the solutions of eqs . and
, we follow the same procedure used in ref .
@xcite , which goes through a continuum approximation .
assuming that @xmath94 we finally arrive at @xmath95 where now @xmath96 is a continuous variable , @xmath97 is a double square well potential with a barrier height @xmath98 and width @xmath99 , @xmath100 obeys pinned boundary conditions @xmath101 ( @xmath2 is a width of a double square well potential ) and the nonlinearity parameter is given by @xmath102 . expressing @xmath103 and @xmath104 and mentioning that total power @xmath20
is connected with nonlinearity parameter @xmath105 as @xmath106 , we see the equation is the same as and thus all the above consideration of peculiarities of double square well potential directly applies to the considered bec lattices . in case of the waveguide systems the situation is even simpler . particularly , as well known an array of adjacent waveguides coupled by power exchange is modeled by the discrete nonlinear schrdinger equation ( dnls ) @xcite which reads @xmath107 where waveguides discrete positions are labelled by the index @xmath96 ( @xmath108 ) , and the complex field @xmath109 results from the projection of the electric field envelope on the eigenmode of the individual waveguide .
it is normalized to a unit onsite nonlinearity . the linear refractive index @xmath110 is set to @xmath80 for all @xmath111 , and to @xmath112 for @xmath113 .
the coupling constant between two adjacent waveguides is @xmath88 and @xmath114 and @xmath115 are the light frequency and velocity .
vanishing boundary conditions @xmath116 model a strongly evanescent field outside the waveguides .
considering now @xmath117 as being a virtual grid spacing we may represent @xmath118 by the function @xmath119 in the continuous variable @xmath120 . as a result the dnls model maps to the with a double square well potential considered initially .
a new coherent state in square double well potential has been discovered .
this coherent state has the property of being _
bistable _ : one can easily switch from oscillatory to self - trapping regimes and back .
this nontrivial behavior may have interesting applications in various weakly linked extended systems , such as bose - einstein condensates , waveguide or josephson junctions arrays , which deserve further studies . in the region of nonlinearities where the asymmetric solution coexists with the symmetric and asymmetric stationary solutions , we have induced the switch from one regime to the other by varying the height of the barrier . in view of a real experiment one could induce such flips by varying the refractive index of the central waveguide ( in the context of weakly linked waveguide arrays ) or by pulswize change of optical barrier potential ( in case of bec ) .
* acknowledgements : * we would like to thank f.t .
arecchi , e. arimondo , a. montina and o. morsch for useful discussions .
r. kh . acknowledges support by marie - curie international incoming fellowship award ( mif1-ct-2005 - 021328 ) and nato grant ( fel.rig.980767 ) , s. r. acknowledges financial support under the prin05 grant on _ dynamics and thermodynamics of systems with long - range interactions _ and s.w .
is funded by the alexander von humboldt foundation ( feodor - lynen program ) .
jensen , ieee j. quantum electron .
* 18 * , 1580 , ( 1982 ) .
b. d. josephson , phys .
* 1 * , 251 ( 1962 ) .
anderson , j.w .
rowell , phys .
* 10 * , 230 ( 1963 ) .
m. albiez , r. gati , j. folling , s. hunsmann , m. cristiani , m. k. oberthaler , phys .
, * 95 * , 010402 ( 2005 ) .
a. smerzi , s. fantoni , s. giovanazzi , s. r. shenoy , phys .
* 79 * , 4950 ( 1997 ) .
s. raghavan , a. smerzi , s. fantoni , s. r. shenoy , phys .
a , * 59 * , 620 ( 1999 ) .
l. p. pitaevskii , sov .
jetp , * 13 * , 451 , ( 1961 ) ; e. p. gross , nuovo cimento , * 20 * , 454 , ( 1961 ) ; j. math . phys . , * 4 * , 195 , ( 1963 ) .
e. a. ostrovskaya et .
al . , phys .
a , * 61 * , 031601(r ) , ( 2000 ) .
d. ananikian , t. bergeman , phys .
a , * 73 * , 013604 , ( 2006 ) .
a. montina , f.t .
arecchi , phys .
a , * 66 * , 013605 , ( 2002 ) .
t. kapitula , p.g .
kevrekidis , nonlinearity , * 18 * , 2491 , ( 2005 ) .
kevrekidis , zhigang chen , b.a .
malomed , d.j .
frantzeskakis , m.i .
weinstein , phys .
lett . a , * 340 * , 275 , ( 2005 ) .
eisenberg et .
al . , phys .
lett . , * 81 * , 3383 , ( 1998 ) .
r. morandotti , h.s .
eisenberg , y. silberberg , m. sorel , j.s .
aitchison , phys .
lett . , * 86 * , 3296 , ( 2001 ) .
d. mandelik et .
al . , phys .
lett . , * 90 * , 053902 , ( 2003 ) ; phys .
lett . , * 92 * , 093904 , ( 2004 ) .
sukhorukov , d. neshev , w. krolikowski , y.s .
kivshar , phys .
lett . , * 92 * , 093901 , ( 2004 ) .
fleischer et al . , phys .
lett . , * 90 * , 023902 , ( 2003 ) .
fleischer et al . ,
nature , * 422 * , 147 , ( 2003 ) .
a. fratalocchi , g. assanto , phys .
e * 73 * , 046603 ( 2006 ) ; a. fratalocchi et .
al . , opt .
express , * 13 * , 1808 , ( 2005 ) .
cataliotti , s. burger , c. fort , p. maddaloni , f. minardi , a. trombettoni , a. smerzi , m. inguscio , science , * 293 * , 843,(2001 ) .
m. anderlini , et al .
, j. phys .
b , * 39 * , s199 ( 2006 ) . p.f .
byrd , m.d .
friedman , _ handbook of elliptic integrals for engineers and physicists _ , springer ( berlin 1954 ) .
t. bergeman , m.g .
moore , m. olshanii , phys .
* 91 * , 163201 , ( 2003 ) .
i. bloch , j. phys .
b * 38 * , s629 ( 2005 ) ; o. morsch , m. oberthaler , rev .
mod . phys . * 78 * , 179 ( 2006 ) .
l. carr , m.j .
holland , and b.a .
malomed , j. phys .
b * 38 * , 3217 ( 2005 ) ; s. wimberger , p. schlagheck , and r. mannella , j. phys .
b * 39 * , 729 ( 2006 ) ; p. schlagheck , t. paul , phys . rev .
a , * 73 * , 023619 ( 2006 ) .
a. smerzi , a. trombettoni , phys .
a * 68 * , 023613 ( 2003 ) .
slater , phys . rev . * 87 * , 807 ( 1952 ) .
r. khomeriki , j. leon , s. ruffo , phys .
* 97 * , 143902 ( 2006 ) .
christodoulides , r.i .
joseph , optics lett .
, * 13 * , 794 , ( 1988 ) .
ablowitz , z.h .
musslimani , phys .
* 87 * , 254102 , ( 2001 ) ; phys .
e , * 65 * , 056618 , ( 2002 ) . | considering the coherent nonlinear dynamics in double square well potential we find the example of _ coexistence _ of josephson oscillations with a self - trapping regime .
this macroscopic bistability is explained by proving analytically the simultaneous existence of symmetric , antisymmetric and asymmetric stationary solutions of the associated gross - pitaevskii equation .
the effect is illustrated and confirmed by numerical simulations .
this property allows to make suggestions on possible experiments using bose - einstein condensates in engineered optical lattices or weakly coupled optical waveguide arrays . |
the complex scaling method @xcite invented in early 70-s remains one of the most effective approaches to calculation of resonances in few - body systems .
this method is applicable to an @xmath0-body problem in the case where inter - particle interaction potentials are analytic functions of coordinates .
the complex scaling gives a possibility to rotate the continuous spectrum of an @xmath0-body hamiltonian in such a way that certain sectors of unphysical sheets neighboring the physical one turn into a part of the physical sheet for the resulting non - selfadjoint operator .
resonances appear to be complex eigenvalues of this operator @xcite while the binding energies stay fixed during the scaling transformations .
therefore , when searching for the resonances within the complex scaling approach one may apply the methods which are usually employed to locate the binding energies
. some reviews of the literature on the complex scaling and its many applications can be found , in particular , in @xcite .
here we only mention that there is a rigorous mathematical proof @xcite that for a rather wide class of interaction potentials the resonances given by the complex scaling method coincide with the `` true scattering resonances '' , i.e. the poles of the analytically continued scattering matrix in the unphysical sheets . along with the complex scaling ,
various different methods are also used for calculations of the resonances . among the methods developed to calculate directly the scattering - matrix resonances we ,
first , mention the approach based on the momentum space faddeev integral equations @xcite ( see , e.g. , ref .
@xcite and references cited therein ) . in this approach one
numerically solves the equations continued into an unphysical sheet and , thus , the three - body resonances arise as the poles of the continued t - matrix .
another approach to calculation of the scattering - matrix resonances is based on the explicit representations @xcite for the analytically continued t- and s - matrices in terms of the physical sheet . from these representations
one infers that the three - body resonances can be found as zeros of certain truncations of the scattering matrix only taken in the physical sheet .
such an approach can be employed even in the coordinate space @xcite . to the best of our knowledge
there are no published works applying the complex scaling to the faddeev equations .
therefore , we consider the present investigation as a first attempt undertaken in this direction .
however , the purpose of our work is rather two - fold . on the one hand
, we make the complex scaling of the faddeev differential equations . on the other hand
we compare the complex scaling method with the scattering - matrix approach suggested in @xcite .
we do this making use of both the approaches to examine resonances in a model system of three bosons having the nucleon masses and in the three - nucleon ( @xmath1 ) system itself .
first , we recall that , after the scaling transformation , the three - body schrdinger operator reads as follows @xcite @xmath2 where @xmath3 is the scaling parameter with @xmath4 .
by @xmath5 we understand the six dimensional laplacian in @xmath6 where @xmath7 are the standard jacobi variables , @xmath8 .
notation @xmath9 is used for the two - body potentials which are assumed to depend on @xmath10 but not on @xmath11 . the corresponding scaled faddeev equations which we solve read @xmath12 \phi^{(\alpha)}(z;x)+ v_\alpha({{\rm e}}^\vartheta
x_\alpha)\sum_{\beta\neq\alpha}\phi^{(\beta)}(z;x ) = f_\alpha(x),}\\ \label{faddinhom } & & \qquad\qquad\qquad{\alpha=1,2,3 . } \ ] ] here @xmath13 is an arbitrary three - component vector with components @xmath14 belonging to the three - body hilbert space @xmath15 . the partial - wave version of the equations ( [ faddinhom ] ) for a system of three identical bosons at the zero total angular momentum @xmath16 reads @xmath17 where @xmath18 , @xmath19 and @xmath20 denotes the partial - wave kinetic energy operator , @xmath21 while @xmath22 stands for the partial - wave component of the total wave function , @xmath23 here , @xmath24 and @xmath25 .
explicit expression for the geometric function @xmath26 can be found , e.g. , in @xcite .
the partial - wave equations ( [ fadpartcor ] ) are supplied with the boundary conditions @xmath27 for compactly supported inhomogeneous terms @xmath28 the partial - wave faddeev component @xmath29 also satisfies the asymptotic condition @xmath30 \\ & + & \displaystyle\frac{\exp({\rm i } \sqrt{z}\,\,{{{\rm e}}^{{{\rm i}}\theta}}\rho)}{\sqrt{\rho } } \left[a_l(z;y / x)+o\left(\rho^{-1/2}\right)\right ] , \end{array}\ ] ] for simplicity it is assumed in this formula that the two - boson subsystem has only one bound state with the energy @xmath31 , and @xmath32 represents its wave function .
the values of @xmath33 and @xmath34 are the main asymptotical coefficients effectively describing the contributions to @xmath35 from the elastic @xmath36 and breakup @xmath37 channels , respectively . hereafter , by @xmath38 , @xmath39
, we understand the main ( arithmetic ) branch of the function @xmath40 . in the scaling method a resonance
is looked for as the energy @xmath41 which produces a pole to the quadratic form @xmath42^{-1 } f , f\right>}\ , \ ] ] where @xmath43 is the non - selfadjoint operator resulting from the complex - scaling transformation of the faddeev operator .
the latter operator is just the operator constituted by the l.h.s .
parts of eqs .
( [ faddinhom ] ) .
the resonance energies should not , of course , depend on the scaling parameter @xmath44 and on the choice of the terms @xmath28 .
in the scattering - matrix approach we solve the same partial - wave faddeev equations ( [ fadpartcor ] ) with the same boundary conditions ( [ bc0 ] ) and ( [ hebs ] ) but for @xmath45 and @xmath46 the resonances are looked for as zeroes of the truncated scattering - matrix ( see @xcite for details ) @xmath47 , where the @xmath48 elastic scattering amplitude @xmath49 for complex energies @xmath41 in the physical sheet is extracted from the asymptotics ( [ hebs ] ) .
for numerical solution of the boundary - value problem ( [ fadpartcor][hebs ] ) we employ its finite - difference approximation in the hyperradius - hyperangle coordinates .
a detail description of the finite - difference algorithm used can be found in ref .
in the table we present our results obtained for a complex - scaling resonance in the model three - body system which consists of identical bosons having the nucleon mass . to describe interaction between them
we employ a gauss - type potential of ref .
@xcite @xmath50 + v_b \exp[-\mu_b ( x - x_b)^2]\ ] ] with @xmath51mev , @xmath52@xmath53 , @xmath54fm , @xmath55@xmath53 and @xmath56 .
the figures in the table correspond to the roots of the inverse function @xmath57^{-1}$ ] for @xmath16 and @xmath58 only taken into account . in the present calculation
we have taken up to 400 knots in both hyperradius and hyperangle variables while for the cut - off hyperradius we take 40fm .
one observes from the table that the position of the resonance depends very weakly on the scaling parameter @xmath44 which confirms a good numerical quality of our results .
we compare the resonance values of the table to the resonance value @xmath59mev obtained for the same three - boson system with exactly the same potentials but in the completely different scattering - matrix approach of ref .
we see that , indeed , both the complex scaling and the scattering matrix approaches give the same result .
we also watched the trajectory of the above resonance when the barrier amplitude @xmath60 varied ( see .
[ fig - trajectory1 ] ) . while the complex scaling method was applicable it gave practically the same positions for the resonance . for the barrier
amplitudes @xmath60 smaller than 1.0 only the scattering - matrix approach allows to locate the resonance ( which finally , for @xmath61 , turns into a virtual level ) . as to the @xmath1 system in the @xmath62state where we employed the mti
iii @xcite potential model , both the methods applied give no resonances on the two - body unphysical sheet ( see @xcite ) .
moreover , we have found no resonances in the part of the three - body sheet accessible via the complex scaling method .
thus , at least in the framework of the mti
iii model we can not confirm the experimental result of ref .
@xcite in which the point @xmath63mev was interpreted as a resonance corresponding to an exited state of the triton @xmath64h .
the triton virtual state can be only calculated within the scattering - matrix method but not in the scaling approach .
our present improved scattering - matrix result for the triton virtual state is @xmath65mev ( i.e. the virtual level lies 0.47mev below the two - body threshold ) .
this result has been obtained with the mti - iii potential on a grid having 1000 knots in both hyperradial and hyperradial variables and with the value of cut - off hyperradius equal to 120fm .
notice that some values for the virtual - state energy obtained by different authors can be found in @xcite and all of these values are about 0.5mev below the two - body threshold . | in this work we compare two different approaches to calculation of the three - body resonances on the basis of faddeev differential equations .
the first one is the complex scaling approach .
the second method is based on an immediate calculation of resonances as zeros of the three - body scattering matrix continued to the physical sheet . ,
, three - body systems , complex scaling , resonances |
let @xmath0 be a @xmath1-finite complete measure space and @xmath2 be the field of complex numbers .
by @xmath3 , we denote the linear space of all equivalence classes of @xmath4-valued @xmath5-measurable functions on @xmath6 that are identified @xmath7-a.e . and
are considered as column vectors .
let @xmath8 denote the linear space of all functions in @xmath9 that are finite a.e . with the topology of convergence in measure on the sets of finite measure , it is a metrizable space . the @xmath4-valued banach function space @xmath10 is defined as @xmath11 where @xmath12 is a function norm on @xmath10 such that for each @xmath13 , we have 1 .
@xmath14 for @xmath7-a.e .
@xmath15 , 2 . @xmath16-a.e . for each @xmath17 , and 3 .
@xmath18 with @xmath19 implies that @xmath20 , for each @xmath21 , and @xmath22 for some constant @xmath23 , depending on @xmath24 but independent of @xmath25 , where @xmath26 is the characteristic function of the set @xmath24 .
@xmath27 a function @xmath25 in a banach function space @xmath10 is said to have _ absolutely continuous norm _ in @xmath10 if @xmath28 for every sequence @xmath29 of @xmath7-measurable sets in @xmath6 satisfying @xmath30-a.e . , where @xmath31 means that @xmath32-a.e .
let @xmath33 be the set of all functions in @xmath10 having _ absolutely continuous norm_. if @xmath34 , then we say that @xmath10 has _ absolutely continuous norm_. let @xmath35 be the closure of the set of all @xmath7-simple functions in @xmath10 .
then , we have @xmath36 throughout this paper , we assume that @xmath37 , that is , the simple functions are dense in @xmath10 . in case @xmath10 has _ absolutely continuous norm _ , we have @xmath38 and so its banach space dual @xmath39 and its associate space @xmath40 coincide , where @xmath40 is defined as @xmath41 and @xmath42 where @xmath43 denotes the corresponding space of equivalence classes of @xmath4-valued functions considered as row vectors .
note that @xmath44 is the usual product of a row matrix formed by @xmath45 into a column matrix formed by @xmath46 .
the monotone convergence theorem holds in every banach function space @xmath10 in the form of the weak fatou property ( see axiom ( ii ) ) . also note that @xmath33 is the largest subspace of @xmath10 for which the suitable dominated convergence theorem holds ( see proposition 3.6 , p. 16 of @xcite ) .
so this fact can be used to easily generalise those results on @xmath47-spaces to the general banach function spaces having absolutely continuous norm in which the dominated convergence theorem is required .
it is due to this fact that our results in 3 follow on similar lines as in @xmath47-spaces without any extra effort .
for details on banach function spaces , we refer to @xcite . for a measurable function @xmath48 ,
the set of all @xmath49 matrices over @xmath2 , the multiplication transformation @xmath50 is defined as @xmath51 using the arguments given in the introduction on p. 517 of @xcite and p. 163
, theorem 1.2 of @xcite we can easily prove the next result . the multiplication operator @xmath52 is a bounded operator on a banach function space @xmath10 if and only if @xmath53 , the space of all @xmath54-valued essentially - bounded measurable functions .
moreover , we have @xmath55 } \sup_{x \in \omega } \| q(x ) \|,\\[.3pc ] \|m_{u}\|_{x\mapsto x } & = \inf \{m \geq 0\hbox{\rm : } \ \mu ( \ { x \in \omega\hbox{\rm : } \ \| u(x ) \| > m \ } ) = 0 \},\hskip -1 cm \phantom{0}\end{aligned}\ ] ] where @xmath56 is the operator norm of @xmath57 in @xmath58 induced by @xmath59 and @xmath60 $ ] is the equivalence class of all measurable functions that are @xmath7-a.e . equal to @xmath61 .
note that for @xmath62 is the space of essentially bounded @xmath2-valued measurable functions denoted as @xmath63 . for details on matrix analysis , see @xcite .
@xmath27 for @xmath64 , an operator @xmath65 defined on @xmath66 by @xmath67 for @xmath68 and for all @xmath69 is called a matrix multiplication operator .
we consider an abstract banach space - valued linear initial value problem of the form @xmath70
v(0 ) & = x , \end{aligned}\ ] ] where the independent variable @xmath71 represents time , @xmath72 is a function with values in a @xmath4-valued banach function space @xmath73 , where @xmath4 denotes the @xmath74-dimensional complex space .
since all the norms on @xmath4 are equivalent , we choose @xmath75 .
note that this norm induces the matrix norm @xmath76 on @xmath77 , that is , the norm of a matrix is given by the maximal absolute sum of its rows . also , @xmath78 is a linear operator and @xmath79 is the initial value .
this problem is called an abstract cauchy problem ( acp ) associated with a matrix multiplication operator @xmath80 and the initial value @xmath81 . a function
@xmath82 forms a solution of acp , if @xmath61 is a continuously differentiable function with respect to @xmath10 and @xmath83 , for each @xmath84 ( see @xcite ) .
if the operator @xmath85 is the generator of a strongly continuous semigroup @xmath86 , then for each @xmath87 , the function @xmath88 is the unique solution of acp .
see @xcite and @xcite for details on acp and semigroups .
note that every well - posed cauchy problem is solved by a strongly continuous semigroup and each such semigroup is the solution semigroup of a well - posed cauchy problem ( see @xcite ) . in this paper , we extend the results of @xcite to general banach function spaces .
note that the separable banach function spaces form a subclass of the absolutely continuous ones .
the examples of banach function spaces having absolutely continuous norm are @xmath47-spaces , orlicz spaces with @xmath89-conditions @xcite , lorentz spaces @xcite , separable orlicz
lorentz spaces @xcite , etc . in the second section
, we study the compactness , closedness , invertibility and fredholmness properties of a multiplication operator @xmath52 on a banach function space @xmath10 . in the third section ,
we discuss the abstract cauchy problem associated with the operator @xmath65 on @xmath10 .
for a measurable function @xmath90 , we call the set @xmath91 its essential range and define the associated multiplication operator @xmath52 on the space @xmath10 by @xmath92 for each @xmath25 in the domain @xmath93 in case @xmath52 is a bounded linear operator on @xmath10 , we have @xmath94 . [
th : one]@xmath27 let @xmath65 be the multiplication operator on a banach function space @xmath10 induced by some measurable function @xmath95 .
then the following statements hold : 1 .
the operator @xmath96 is closed and densely defined , in case @xmath10 has absolutely continuous norm .
2 . the operator @xmath52 has a bounded inverse if and only if @xmath97 . in this case ,
@xmath98 , for some measurable @xmath99 defined by @xmath100 for each @xmath101 .
the spectrum of @xmath52 is the essential range of @xmath61 , i.e. , @xmath102 .
the proof follows on similar lines as in @xmath47-spaces ( see @xcite and @xcite ) .
[ th : algebra ] for @xmath103 , the set of all bounded multiplication operators on a banach function space @xmath10 of @xmath2-valued functions forms a maximal abelian subalgebra of @xmath104 , the space of all bounded linear operators on @xmath10 .
let @xmath105 .
clearly , @xmath106 is an abelian subalgebra of @xmath104 .
we prove that @xmath106 is maximal , i.e. , if @xmath85 commutes with @xmath106 , then @xmath107 .
let @xmath108 be the unity function .
let @xmath109 and @xmath18 .
then @xmath110 we claim @xmath111 .
suppose that the set @xmath112 has a positive measure for each @xmath113 . by the finite subset property of the underlying measure space
, we assume @xmath114 . then , @xmath115 and we have @xmath116 \rightarrow n \| 1_{f_n } \|_{x } & \leq \| m_v 1_{f_n } \|_{x } \\[.2pc ] \rightarrow n \| 1_{f_n } \|_{x } & \leq \| a 1_{f_n } \|_{x}\end{aligned}\ ] ] for each @xmath117 , which contradicts the boundedness of @xmath85 .
therefore @xmath118 .
since the set of @xmath7-simple functions is dense in @xmath10 , we have @xmath119 .
this proves that @xmath107 and so @xmath106 is a maximal abelian subalgebra of @xmath120 .
[ th : closed ] let @xmath121 be a banach function space of @xmath2-valued measurable functions on @xmath6 and @xmath122 for some @xmath123 . then @xmath52 has closed range if and only if there exists some @xmath124 such that @xmath125-almost all @xmath126 .
suppose @xmath127 for @xmath7-almost all @xmath128 .
then using the same technique as in converse part of theorem @xmath129 in p. 163 of @xcite with @xmath130 , we have @xmath131 for @xmath7-a.e . @xmath68 and each @xmath132 . as the norm @xmath133 is increasing on @xmath10
, so we have @xmath134 for each @xmath132 .
this implies that @xmath52 is invertible .
thus , we conclude that @xmath52 has closed range .
conversely , suppose @xmath52 has closed range .
then there exists some @xmath135 such that @xmath136 let @xmath137 be such that @xmath138 .
so , there exists a measurable set @xmath139 such that @xmath140 .
further , @xmath141 which contradicts ( [ eq : delta ] ) and so @xmath142 .
therefore @xmath143 , for @xmath144-almost all @xmath128 .
let @xmath145 be a banach function space of @xmath146-valued measurable functions on @xmath6 and @xmath147 .
then @xmath52 is a compact operator if and only if @xmath148 is finite - dimensional , for each @xmath149 , where @xmath150 and @xmath151 suppose @xmath52 is a compact operator .
then , its restriction to the invariant subspace @xmath152 is compact . also , @xmath153 has closed range in @xmath154 . but @xmath155 is invertible .
therefore , @xmath156 is finite - dimensional for each @xmath157 .
conversely , suppose @xmath156 is finite - dimensional for each @xmath157 .
in particular , @xmath158 is finite - dimensional for each @xmath113 .
take @xmath159 and define @xmath160 , by @xmath161 0 & \mbox { if}\ x \notin a_n , \end{cases}\ ] ] for each @xmath162 , where @xmath163 .
then for each @xmath164 , we have @xmath165 & \quad\ , + \| ( w_n(x ) -w(x))f(x ) 1_{\omega \backslash a_n } ( x)\| _ { \mathbf{c}^n}\\[.2pc ] & = \| w(x ) f(x ) 1_{\omega \backslash a_n } ( x)\| _ { \mathbf{c}^n } \leq \frac{1}{n } \| f(x ) \| _ { \mathbf{c}^n},\end{aligned}\ ] ] for @xmath7-a.e .
, @xmath68 .
thus , for each @xmath164 , we have @xmath166 this implies that @xmath167 as @xmath168 .
but each @xmath169 is of finite rank and thus @xmath170 is a compact operator .
the next result gives the necessary and sufficient conditions for a multiplication operator @xmath52 on a banach function space @xmath121 to be a fredholm operator thereby generalising the results in @xcite for orlicz spaces and @xcite for @xmath47-spaces .
here , we take @xmath103 .
suppose @xmath171 is a non - atomic measure space and @xmath172 , where @xmath173 is a banach function space having absolutely continuous norm .
then , the following are equivalent . 1
. @xmath52 is invertible .
2 . @xmath52 is fredholm .
the range of @xmath52 , that is , @xmath174 is closed and @xmath175 .
4 . @xmath127 for @xmath7-a.e .
@xmath68 for some @xmath176 . we prove ( iii ) @xmath177 ( iv ) only , as the other implications are obvious by using the previous results .
suppose @xmath174 is closed and @xmath175 .
we claim @xmath52 is onto .
suppose the contrary .
then there exists @xmath178 .
since @xmath174 is closed , there exists @xmath179 the dual ( associate ) space of @xmath10 such that @xmath180 and @xmath181 now ( [ eq : integer ] ) yields that the set @xmath182 has positive @xmath7-measure for some @xmath183 . as @xmath184 is non - atomic
, we can choose a sequence @xmath185 of subsets of @xmath186 with @xmath187 for @xmath188 .
put @xmath189 .
clearly , @xmath190 , as @xmath191 for each @xmath192 . also , for each @xmath193 and so ( [ eq : inte ] ) implies that @xmath194 & = \int_{\omega } \prec\ !
m_u 1_{e_n } f , g_{\circ } \!\succ \ , \d \mu = 0,\end{aligned}\ ] ] where @xmath195 is the conjugate operator of @xmath196 , which implies that @xmath197 a.e . and so @xmath198 . since all the sets in @xmath199 are disjoint , the sequence @xmath200 forms a linearly - independent subset of @xmath201 .
this contradicts the fact that @xmath202 so @xmath52 is onto .
let @xmath203 . then @xmath204 .
since @xmath205 , there is an @xmath206 .
thus , @xmath207 which contradicts the surjectiveness of @xmath52 . for @xmath208 , put @xmath209 and @xmath210 clearly @xmath211 and @xmath212 , for each @xmath208 .
take @xmath213 then @xmath214 & \leq \sigma_{n\in t } \frac { | u(x ) 1_{g_n}(x ) | } { \| 1_{g_n } \|_{x } } \\[.3pc ] & \leq \sigma _ { n \in t } \frac { \| u \| _ { \infty } } { n^2 \| 1_{g_n } \cdot z \|_{x } } , \end{aligned}\ ] ] for each @xmath215 , this proves that @xmath216 therefore @xmath132 and so there exists some @xmath217 such that @xmath218 with @xmath219 since the sets @xmath220 are disjoint , using the absolute continuity of @xmath10 ( theorem 4.1 , p. 20 of @xcite ) , we have @xmath221 & = \left ( \sum_{n \in t } \frac{1 } { \| 1_{g_n } \|_{x } } \right ) \sup_{h \in x^ * , \|h \|_{x^ * } \leq 1}\left\vert \int_{\omega } \prec\ !
1_{g_n } , h \!\succ \ , \d \mu \right\vert\\[.3pc ] & = \sum_{n \in t } 1 .\end{aligned}\ ] ] then , @xmath222 if and only if @xmath223 is finite .
so there is some @xmath224 such that for @xmath225 and @xmath226 , which further implies that @xmath227 & \leq \mu((\cup_{n=1}^{\infty } g_n ) \cup \ker(u ) ) = 0,\end{aligned}\ ] ] that is , @xmath228 , a.e . on @xmath10
this proves ( iv ) .
in this section , we study some applications of the multiplication operators on @xmath10 in semigroup theory . using proposition [ th :
one](i ) , we see that @xmath52 is a closed and densely defined multiplication operator defined on an absolutely continuous banach function space @xmath10 .
the essential spectrum of a multiplication operator is defined as @xmath229 } \overline { \cup _ { x \in \omega } \sigma ( p(x ) ) } , \ ] ] where @xmath230 denotes the spectrum of the matrix @xmath231 and @xmath60 $ ] is the equivalence class of all measurable functions that are @xmath7-a.e .
equal to @xmath61 . for an open @xmath232-disk @xmath233 with center @xmath234 , we have @xmath235 & = \ { z \in \mathbf{c}\hbox{:}\ \mu ( \ { x \in \omega\hbox{:}\ \sigma ( u(x ) ) \cap z+u_{\epsilon } \ } ) > 0,\quad \forall \epsilon > 0 \}.\end{aligned}\ ] ] using propositions 4.11 and 4.12 , p. 32 of @xcite , it is easy to prove the result .
@xmath27 suppose @xmath236 is a matrix multiplication operator on the banach function space @xmath10 with non - void resolvent set @xmath237
. then its spectrum is given by @xmath238 if @xmath239 with the lebesgue measure @xmath7 and @xmath61 is a continuous function with non - void resolvent , then we have @xmath240 the stability of the solutions of the abstract cauchy problem are determined by the spectral bound of the corresponding operator , @xcite .
@xmath27 a matrix multiplication operator @xmath52 is bounded if and only if its spectrum @xmath241 is bounded .
there are a number of functional analytic approaches to the well - posedness of solutions of abstract cauchy problems ( see , for example , @xcite ) .
the proofs of the following results are on the similar lines as in @xcite for @xmath47-spaces , so we only state the results . a matrix multiplication operator @xmath52 generates a semigroup on @xmath10 having absolutely continuous norm if and only if @xmath242 in case @xmath52 generates the semigroup @xmath243 , then we have @xmath244 . on the other hand ,
if @xmath245 defines a semigroup on @xmath10 , then its generator is given by @xmath196 .
if the matrix multiplication operator @xmath65 generates the semigroup @xmath243 , then @xmath246 @xcite@xmath27 let @xmath247 be a linear operator on a banach space @xmath248 such that there exists some constants @xmath249 , the real line and a strongly continuous family @xmath250 in @xmath251 , the space of all the bounded linear operators on @xmath248 with @xmath252 such that its resolvent operator @xmath253 exists and is given by @xmath254 for @xmath255 and @xmath256 .
then @xmath257 is the generator of an @xmath258-times integrated semigroup @xmath250 .
the generator of an @xmath258-times integrated semigroups yields the well - posedness of the abstract cauchy problem , since there exists some constants @xmath259 and @xmath260 as above such that for all initial values @xmath261 , we get a unique solution @xmath262 with theproperty @xmath263 for each @xmath84 , where @xmath264 denotes the @xmath258th - graph norm of @xmath257 ( see @xcite ) .
let @xmath65 be a matrix multiplication operator on a banach function space @xmath265 .
then the following are equivalent . 1 .
the operator @xmath65 is the generator of an integrated semigroup .
there exists a constant @xmath266 such that @xmath267 3 .
the resolvent set of @xmath52 is non - empty such that @xmath268 for some @xmath266 , where @xmath269 denotes the spectral bound of the matrix @xmath270 . as in corollary 4.9 of @xcite
, we see that the multiplication operator @xmath65 satisfying one of the conditions in the statement of the above - stated theorem is a generator of a @xmath271-times integrated semigroup . for a multiplication operator @xmath65 on @xmath272
, we see that it is a generator of a strongly continuous semigroup if and only if there is some @xmath273 such that @xmath274.\ ] ] theorem 1 on p. 16 of @xcite is also true for banach function spaces @xmath272 .
further , if @xmath10 has absolutely continuous norm , then p. 163
, theorem 2 of @xcite , p. 164 , proposition 1 of @xcite and p. 165 , proposition 2 of @xcite can be proved for such banach function spaces .
the matrix multiplication operator @xmath65 on @xmath10 also generates some subclasses of the strongly continuous semigroups on @xmath10 such as analytic , differentiable , norm - continuous semigroups , etc .
see @xcite for details on these types of semigroups . also , we see that the analyticity of semigroups on @xmath10 depends on the spectrum @xmath241 of the matrix multiplication operator @xmath65 on @xmath10 .
the authors are thankful to the referee for valuable suggestions and comments .
one of the authors ( rajeev kumar ) is supported by csir - grant ( sanction no .
9(96)/2002-emr - i , dated 13 - 5 - 2002 ) . | in this paper , we study the matrix multiplication operators on banach function spaces and discuss their applications in semigroups for solving the abstract cauchy problem .
[ ssection ] [ ssection ] |
we are indebted to a. fabrocini for providing us the results of the second order perturbative corrections to the energy per particle of neutron matter for the au6 interaction and to c.j .
pethick for insightful discussions .
portions of this work were supported by murst - national research projects and cineca computing center . .
spin susceptibility ratio @xmath58 of neutron matter .
our afdmc results for the interactions au6 , au8 and reid6 are compared with those obtained from refs . [
8,9 ] by using eq([susc_ratio ] ) .
the statistical error is given in parentheses . [ cols="^,^,^,^,^,^ " , ] | the auxiliary field diffusion monte carlo method is applied to compute the spin susceptibility and the compressibility of neutron matter at zero temperature .
results are given for realistic interactions which include both a two body potential of the argonne type and the urbana ix three body potential .
simulations have been carried out for about 60 neutrons .
we find an overall reduction of the spin susceptibilty by about a factor of 3 with respect to the pauli susceptibility for a wide range of densities .
results for the compressibility of neutron matter are also presented and compared with other available estimates obtained for semirealistic nucleon nucleon interactions and with more traditional many
body techniques , like brueckner s or correlated basis function theories . in this paper
we show that the strong correlations induced by realistic nucleon nucleon interactions reduce , by up to about a factor of 3 , the spin susceptibility , @xmath0 , of degenerate neutron matter .
this reduction may have important implications for problems of astrophysical interest , like for instance , neutrino scattering rates in dense matter and , more generally , the study of supernovae and proto neutron stars@xcite .
starting from the pioneering work by sawyer@xcite , several calculations of the neutrino mean free path in uniform nuclear matter have been performed@xcite which show that the effects due to strong interactions are relevant , particularly in the spin - density channel which couples with the axial vector current . a sizable reduction of @xmath0 , or , equivalently , a large value of the @xmath1 landau parameter leads to an appreciable suppression of the gamow teller transitions .
the neutrino momenta and the momentum transfers in many applications are small compared with the neutron fermi momentum , and both energy transfers and the temperature are small compared with the neutron fermi energy .
therefore , the landau parameters of neutron matter at zero temperature are the main quantities to compute in order to evaluate the mean free path of a neutrino in dense matter . in this respect , _ ab initio _ calculations of the landau parameters or related quantities , such as the compressibility @xmath2 , the effective mass @xmath3 or the spin susceptibility @xmath0 , for degenerate neutron matter are extremely important .
such calculations can now be performed because of recent advances in many
body methods , particularly those based on quantum simulations , and because of the much improved knowledge of the nucleon nucleon interaction .
previous evaluations of the landau parameters @xmath4 and @xmath5 ( with @xmath6 ) for neutron matter were based either on skyrme type potential models@xcite or on microscopic calculations performed with semirealistic bare interactions@xcite .
the qualitative behaviour of the compressibility ratio is @xmath7 where @xmath8 the compressibility of the non interacting fermi gas , is similar in most of the various calculations . on the contrary ,
the spin susceptibility ratio , which is approximately related to @xmath1 by @xmath9 may differ up to a factor of 3 at equilibrium density of nuclear matter @xmath10 , and even more at higher densities .
the skyrme models generally predict much smaller values of @xmath1 with respect to microscopic calculations , and become unstable to spin oscillations , driving toward ferromagnetic ground states@xcite , since @xmath11 becomes negative for densities in the range @xmath12 . at small momentum transfer ,
n n correlations enhance the spin
response approximately by a factor @xmath13 .
therefore , skyrme models predict neutrino mean free paths smaller than microscopic models .
on the other hand , the existing microscopic calculations have been performed with old semirealistic interactions , like the reid or the bethe johnston potentials , and no three body force .
in addition , the many body methods used until now may be questionable for the convergence of the underlying perturbation theory as well as for the treatment of spin dependent correlations .
we report the results of quantum simulations of neutron matter for the old reid@xmath14 potential @xcite and modern realistic interactions , based upon the argonne @xmath15 two body potential plus the urbana three body potential , uix , @xcite , denoted hereafter as au18 .
it is well known that such interactions provide a realistic description of light nuclei and nuclear matter @xcite .
the quantum simulations have been carried out by using a new diffusion monte carlo method , the so called afdmc method@xcite .
it introduces auxiliary field variables in the simulation , which makes it the only existing quantum monte carlo method that can handle spin
dependent nuclear hamiltonians and a relatively large number of nucleons . in this approach
the scalar parts of the hamiltonian are propagated as in standard diffusion monte carlo ( dmc ) @xcite .
auxiliary fields are introduced to replace the spin isospin dependent interactions between pairs of particles with interactions between particles and auxiliary fields .
integrating over the auxiliary fields reproduces the original spin isospin dependent interaction .
the method consists of a monte carlo sampling of the auxiliary fields and then propagating the spin
isospin variables at the sampled values of the auxiliary fields .
this propagation results in a rotation of each particle s spin
isospin spinor .
the guiding function , @xmath16 , in our afdmc calculation is a simple trial function given by a slater determinant of one body space
spin orbitals multiplied by a central jastrow correlation .
the orbitals are plane waves that fit in the box times two component spinors , corresponding to neutron up and neutron down states .
the overlap of a walker with this wave function is the determinant of the space
spin orbitals , evaluated at the walker position and spinor for each particle , and multiplied by the scalar jastrow product .
such an overlap is complex , so the usual fermion sign problem becomes a phase problem .
we constrain the path of the walkers to a region where the real part of the overlap with our trial function is positive . for spin
independent potentials this reduces to the fixed node approximation .
this method has already been applied to unpolarized neutron matter and neutron drops ( @xmath17 ) with fairly realistic interactions that include tensor , spin orbit and three body terms .
the neutron matter calculations have been done with up to 66 neutrons in a periodic box with a low variance ( @xmath18 mev per nucleon ) .
the calculation scales in particle number roughly like fermion monte carlo with central forces @xcite .
we compute the spin susceptibility by applying a magnetic field to the system . ignoring any orbital effects ,
the hamiltonian is given by @xmath19 where @xmath20 and @xmath21 , and the susceptibility is defined as @xmath22 where @xmath23 is the number density and @xmath24 is the ground energy in field @xmath25 .
let us use the pauli expansion of the energy per particle as a function of the spin polarization @xmath26 @xmath27 where the derivatives are with respect to the polarization .
minimizing @xmath28 with respect to @xmath29 one gets the following result for the spin susceptibility @xmath30 for a noninteracting fermi gas the spin susceptibility is @xmath31 .
afdmc allows us to get the energy eigenvalue , @xmath32 , for the interacting system in a field @xmath25 for a state of a given spin asymmetry @xmath33 . assuming that the energy and polarization are known in terms of @xmath34 , @xmath35 in eq .
( [ suspauli ] ) can be obtained as a straightforward application of the chain rule @xmath36^{-2 } \left \ { \frac{\partial^2 e_0}{\partial j_z^2 } - \frac{\partial e_0}{\partial j_z } \left [ \frac{\partial p}{\partial j_z } \right ] ^{-1 } \frac{\partial^2 p}{\partial j_z^2 } \right \ } \ .\ ] ] since we are calculating the lowest energy state , the derivative of the energy with respect to @xmath34 vanishes .
therefore this result reduces to @xmath36^{-2 } \frac{\partial^2 e_0}{\partial j_z^2 } \ . \label{esecond}\ ] ] let us consider the non interacting finite systems as a guide for the quantum simulations of the interacting ones .
for such systems the energy is not a quadratic function of the external field @xmath25 . in figure [ f1 ]
we plot @xmath32 , as a function of @xmath25 , for four different systems with a finite number @xmath37 of noninteracting neutrons in a periodic box at @xmath38 .
the cases , for which we have done simulations , are shown in the figure , namely @xmath39=@xmath40 , @xmath41 , @xmath42 and @xmath43 .
one can see that the various @xmath44 are linear in @xmath25 and each of them is tangent to the pauli parabola ( which refers to infinite fermi gas case ) at some value @xmath45 of the field @xmath25 .
we also compare in the figure the pauli parabola with the exact result for the fermi gas at that density .
they are very close up to @xmath46 mev .
using eq .
( [ esecond ] ) , one gets @xmath47 for @xmath48 and @xmath49 . in the interacting case , the derivatives in eq .
( [ esecond ] ) can be easily estimated by computing @xmath32 with afdmc , and using the following equations @xmath50 whose validity relies on the following reasonable assumptions : ( i ) for @xmath51 , @xmath32 is quadratic in @xmath34 ; ( ii ) for a fixed @xmath34 , @xmath32 is linear in @xmath25 ; ( iii ) the polarization is linear in @xmath34 .
these assumptions become exact in the limit of an infinite system with @xmath34 and @xmath25 small .
the non interacting case indicates the use of @xmath52 and the value for @xmath45 at which @xmath53 is tangent to the pauli parabola ( for the density @xmath54 of fig .
[ f1 ] , @xmath55 mev ) .
most of our calculations have been carried out with @xmath52 . at @xmath56
we have verified the linearity of @xmath53 on @xmath25 beyond @xmath57mev finding that the result for @xmath58 is largely independent on the value of @xmath45 .
we have also verified the dependence of @xmath58 on @xmath59 , performing simulations with @xmath60 and found that it is very weak .
a time step @xmath61 was sufficient in most of the quantum simulations to obtain agreement between the mixed and the growth energies @xcite within the statistical accuracy .
we have made simulations with the reid @xmath14 interaction ( reid6 ) , the same used in the correlated basis function calculation of ref.@xcite .
our result for @xmath58 , at @xmath62 , is about @xmath63 smaller , which indicates that the cbf perturbative calculations of ref.@xcite , based on jastrow correlated basis functions , either have not reached a satisfactory convergence for the landau parameter @xmath1 or eq ( [ susc_ratio ] ) may not be sufficiently adequate for the reid6 interaction .
we have also considered realistic interactions , characterized by the so called @xmath64 or @xmath65 two body potentials plus uix three body potential , and hereafter denoted as au6 and au8 respectively .
the @xmath65 is a two body potential of the argonne type which includes the four central spin isospin components , plus the four tensor and spin orbit ones .
it fits the nucleon
nucleon experimental data and embodies the main features of the argonne @xmath15@xcite .
the @xmath64 potential is the @xmath65 with its spin orbit components removed .
our results for @xmath58 , obtained with the reid6 , au6 and au8 interactions are given in table [ t1 ] and compared with previous microscopic calcuations of the same quantity .
the brueckner theory calculations of ref.@xcite are performed with the full reid potential , and , therefore are not directly comparable with our reid6 calculations .
however , we do not find a sizable contribution to @xmath58 coming from the spin
orbit component of the two body potential .
our results for au6 and au8 coincide within the statistical error . to address the problem of the influence of the three - body force on the results
we have calculated the spin susceptibility at @xmath54 using the argonne @xmath64 potential and no three body interactions .
we have found that while the energy is reduced by roughly 25 mev per particle , the spin susceptibility ratio is practically unchanged ( 0.31(1 ) ) .
we have also calculated the compressibility @xmath2 , given by @xmath66 where @xmath67 is a polynomial fit to the afdmc energies @xmath68 . for a fermi gas
the compressibility is @xmath69 .
the afdmc results for @xmath70 , obtained with the au6 interaction are shown in table [ t2 ] , where they are also compared with the corresponding cbf estimates ( au6-cbf ) and other existing microscopic calculations@xcite .
for the sake of completeness we show in fig .
[ f2 ] the equations of state of neutron matter , which have been used to compute the compressibility ratio of table [ t2 ] .
the au18 results are taken from ref.@xcite , and have been obtained with the full au18 model interaction , by using variational fhnc / soc methods .
the au6-cbf@xcite results have been obtained by using the au6 interaction , as in the afdmc simulations , and essentially the same many
body technique as in ref.@xcite .
they also include the corrections coming from the lowest order elementary diagram , as discussed in refs .
@xcite .
one can see that the afdmc results for both the equation of state and the compressibility of the au6 model of neutron matter are in reasonably good agreement with those obtained by using cbf theory .
we have estimated the finite size effects in the afdmc simulations by performing variational calculations with the periodic box fhnc method of ref.@xcite .
they indicate a correction which is at most @xmath71 of the mixed energy per particle .
an error of the same size is expected for the compressibilty ratio and of an order of magnitude smaller for the spin susceptibility .
a more detailed discussion of the afdmc calculation and the equation of state of neutron matter will be given elsewhere@xcite .
in conclusion , we have presented new results on the spin susceptibility and the compressibility of neutron matter at zero temperature , which show a strong reduction of these quantities with respect to their fermi gas values .
the calculations have been performed for realistic model interactions , which include tensor , spin orbit and three body terms , and by using the afdmc method , a newly developed quantum simulation technique , capable of dealing with strongly spin dependent interactions .
of particular relevance is the quenching by about a factor of 3 of the spin susceptibility with respect to the commonly used pauli value , found at all the various densities considered .
such a reduction has strong effects on the mean free path of a neutrino in dense matter and should be seriously taken into account in the studies of supernovae and proto neutron stars . |
[ sec : intro ] the set of four - dimensional n=2 superconformal @xmath7 '' refers to those of the theories that posess a lagrangian . ]
yang - mills gauge theories with coupling @xmath8 follows kodaira s classification @xcite of toroidal singularities ( @xmath9 being the torus modulus ) .
( we shall refer to these theories as seiberg - witten ( sw ) theories ) .
this classification falls into an ade pattern , and the type of singularity gives the global symmetry of the corresponding theory .
there are @xmath10 and @xmath11 singularities at i m @xmath12 , as well as a @xmath4 singularity that can occur at all values of @xmath9 .
furthermore , there are six singularities at finite values of @xmath9 , of types @xmath1 , @xmath2 , @xmath3 , @xmath5 , @xmath6 and @xmath0 .
hence there are precisely seven strongly coupled superconformal theories , with global symmetries equal to the corresponding singularity type ( see also @xcite ) . of these theories
the latter three have no lagrangian description , which makes them difficult to study . however , a possible way to sidestep this difficulty appeared when @xcite found the remarkable fact that the generalised coulomb branches of the @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 theories are identical to the higgs branches of other , a priori completely different gauge theories .
these gauge theories , which we will call _ quiver _ theories , are four - dimensional n=2 supersymmetric yang - mills theories with gauge and matter content that may be described by an ade dynkin diagram ( or , rather , a _ quiver _ diagram ) .
this dynkin diagram is precisely that of the ade global symmetry of the corresponding sw theory .
the authors of @xcite extended the analysis to the @xmath6 case and , encouraged by these results , proposed the existence of a mirror symmetry between four - dimensional sw theories and quiver gauge theories , analogous to the mirror symmetry acting on three - dimensional gauge theories @xcite .
the mirror symmetry would exchange the generalised coulomb branch of one theory with the higgs branch of the other , and would provide a map between mass parameters of one theory and fayet - iliopoulos parameters of the other .
since the coulomb branch receives quantum corrections but the higgs branch does not , one consequence of the mirror symmetry is that quantum effects in one theory arise classically in the dual theory , and vice versa . in this paper , we show the remaining moduli space identity , i.e. for the @xmath0 case , thus exhausting the set of sw theories . we also derive , at least implictly , the map between mass and fayet - iliopoulos parameters for the @xmath0 case .
the quiver gauge theory can be realised as the worldvolume theory of a d3-brane probing an orbifold singularity , and in section [ sec : quiver ] we derive the gauge and matter content from such a setup , showing how to describe it by means of a quiver diagram . in section [ sec : e8curve ] we construct the curve describing the higgs branch , expressed in gauge group invariants and fayet - iliopoulos parameters . in section [ sec : hwpicture ] we attempt a deeper understanding of the quiver theory in terms of the hanany - witten picture @xcite , and finally , section [ sec : conclusions ] contains a summary and discussion .
[ sec : quiver ] to construct the quiver gauge theory we start from type iib string theory in ten flat dimensions ( labelled @xmath13 ) and make an orbifold @xmath14 out of the 6789 directions @xcite . if @xmath15 is a discrete subgroup of @xmath7 , then ( the non - compact ) @xmath14 , with its single fixed point at the origin , may be viewed as a local description of a k3 orbifold near one of its several fixed points .
then @xmath15 may be chosen from one of the following groups : the cyclic groups @xmath16 , the dihedral groups @xmath17 , the trihedral group @xmath18 , the octahedral group @xmath19 , and the icosahedral group @xmath20 .
depending on the choice of @xmath15 , the resulting quiver theory will be associated with a dynkin diagram of type @xmath10 , @xmath11 , @xmath5 , @xmath6 or @xmath0 , respectively .
as we are interested in the last case here , we take @xmath21 .
we now probe the singularity at the origin of @xmath14 by putting on it a d3-brane and its @xmath22 images ( @xmath23 the order of @xmath15 ) images to make a full representation of @xmath15 . ]
living in the 0123 dimensions .
open strings stretching between the branes provide a massless neveu - schwarz sector that consists of the states @xmath24 where @xmath25 are the lowest ns raising modes appearing in the laurent expansion of the string worldsheet fermions , and the ( tachyonic ) ns ground state @xmath26 has odd fermion number , @xmath27 .
the @xmath28 hermitian matrices @xmath29 , @xmath30 and @xmath31 are chan - paton matrices , required to obey invariance under @xmath15 , @xmath32 here the matrices @xmath33 make up the regular representation of the action of @xmath15 on the chan - paton indices , and @xmath34 is some matrix in the @xmath35 representation of @xmath15 , acting on the two - vector @xmath36 . the invariance conditions ( [ eq : lambdav])([eq : lambdaii ] ) break the original @xmath37 gauge group to a product of unitary subgroups , @xmath38 , which acts trivially on vectors and hypermultiplets , implying that the actual nontrivial gauge group is @xmath39 . ] and imply that the gauge field on the brane , @xmath40 , transforms in the adjoint of @xmath41 .
moreover , the @xmath42 make up a hypermultiplet in the adjoint of @xmath41 , and the @xmath43 make up a hypermultiplet transforming in the fundamentals of subgroups @xmath44 , as ( @xmath45 ) .
hypermultiplets are sometimes referred to as bifundamentals . ]
the @xmath42 may be interpreted as the position of a brane in the 45 directions , and the @xmath43 similarly parameterise motions of the brane in the 6789 directions .
for @xmath46 , the gauge group is @xmath47 , and the @xmath43 hypermultiplet of a chiral and an antichiral superfield .
the chiral field transforms as described in the text and the antichiral transforms in the conjugate representation . ]
transforms as @xmath48 . comparing this information with an @xmath0 dynkin diagram
, we see that the matter and gauge content can be summarised by an extended dynkin diagram , as in fig .
[ fig : quiver ] @xcite .
we associate each subgroup @xmath49 with a node , letting the edges between them represent the @xmath43 hypermultiplets .
in addition , the edges are equipped with arrows to indicate the way in which the hypermultiplets transform , the direction being from the fundamental towards the antifundamental representation .
the result is a _ quiver _
diagram @xcite , which , besides being a convenient gauge theory summary in general , will be useful to us when computing the curve that describes the higgs branch ( in section [ sec : e8curve ] ) .
due to the orbifolding of the 6789 directions one needs to take into account twisted states .
we denote by a 3-vector @xmath50 the triplet of ns - ns twisted sector scalar fields associated with the @xmath51:th @xmath52 generator of @xmath41 , i.e. @xmath53 labels the nodes of the quiver diagram .
the low energy effective
action on a d3-brane then includes the potential term @xmath54 ^ 2 ) + \sum_{i , m } { \mathrm{tr}}([x^i , x^m]^2 ) \\ & & + \sum_{m , n } { \mathrm{tr}}([x^m , x^n]^2 ) + h(x^m , x^n,\vec{\phi}_k ) , \label{eq : potential}\end{aligned}\ ] ] where @xmath55 is a function of @xmath50 and the @xmath43 hypermultiplets , involving a term @xmath56 and products of the form @xmath57 ( @xmath58 labels the three components of @xmath50 ) .
the last two terms of ( [ eq : potential ] ) may be rewritten using the definition @xcite @xmath59 , \label{eq : momentmap}\ ] ] where @xmath60 and @xmath61 , with @xmath62 and @xmath63 .
the @xmath64 are generators of the @xmath51:th @xmath52 subgroup of @xmath41 , and @xmath65 are the pauli matrices . labelling the rest of the generators of @xmath41 by the index @xmath66 , we obtain @xmath67 for the last two terms of eq .
( [ eq : potential ] ) . the vacuum condition @xmath68 yields the vacuum moduli space as two branches , the coulomb branch and the higgs branch . on the coulomb branch
we have @xmath69 , @xmath70 and @xmath71 , i.e. the d3-brane is stuck at the orbifold singularity in the 6789 directions but is free to move in the 45 directions
the higgs branch , on the other hand , requires @xmath72 , implying that the d3-brane is stationary in the 45 directions , while the @xmath43 may be nonzero .
the kind of space that the @xmath43 describe depends on the value of @xmath50 . if @xmath69 the higgs branch is just the orbifold @xmath14 , whereas for non - vanishing @xmath50 this hypermultiplet space is a resolved version of @xmath14 , with @xmath50 controlling the size of the singularity blow - up .
the latter space is an asymptotically locally euclidean ( ale ) space . that the higgs branch takes this form
may be seen by noting that the functions ( [ eq : momentmap ] ) are precisely the moment maps arising in the mathematical construction of hyperkhler quotients @xcite .
it turns out that the twisted sector moduli @xmath50 discussed above are just the fayet - iliopoulos ( fi ) parameters of the quiver theory .
the classical vacua in gauge theories are determined by integrating out the auxiliary fields and requiring that the resulting potential vanish . in our case one
obtains , for each @xmath52 generator of the gauge theory , a real equation from the d - terms and a complex equation from the f - terms .
these together constitute the three equations @xmath73 where @xmath74 is defined as in eq .
( [ eq : momentmap ] ) and the 3-vector @xmath75 denotes the triplet of fi terms associated with the @xmath51:th @xmath52 generator ( r stands for `` real '' and c for `` complex '' ) .
the condition ( [ eq : susycnd ] ) is precisely the vacuum requirement that the first term in eq .
( [ eq : hmpot ] ) vanish , if we identify @xmath50 with @xmath76 , an identification which is corroborated in section [ sec : hwpicture ] .
thus we see that the fi parameters are associated with resolution of the orbifold singularity , and that there is one fi term for every node of the quiver diagram .
in fact , the fi parameters may be defined as the period of the hyperkhler triplet @xmath77 of complex forms , @xmath78 where @xmath79 is the @xmath51:th of the eight acts trivially leads to a relation among the fi parameters , see section [ sec : e8curve ] . ]
2-cycles required to blow up the @xmath80 singularity .
singularities fall into an ade classification according to the pattern of intersecting 2-cycles required to resolve them @xcite .
these 2-cycles behave exactly like simple roots of the corresponding dynkin diagram , and the classification matches the quiver diagrams so that the 2-cycles may be identified with the fi parameters through eq .
( [ eq : period ] ) . ]
[ sec : e8curve ] the algebraic variety describing the higgs branch of the quiver theory is defined by an equation involving three polynomials in the bifundamentals , that are invariant under the gauge group @xmath41 @xcite .
our goal here is to show that this curve is identical to the seiberg - witten ( sw ) curve describing the generalised coulomb branch of the sw theory with @xmath0 global symmetry .
the generalised coulomb branch was defined in @xcite as the fibration of the seiberg - witten torus over the ordinary coulomb branch .
the sw curve was found in @xcite , but we will use the results of @xcite , as their form of the curve is more useful for our purposes . to find the algebraic curve we use the graphical method described in @xcite , which is based on the quiver diagram .
traces over the bifundamentals are represented by loops in the diagram , and the f - flatness conditions ( [ eq : susycnd ] ) are imposed as graphical rules for manipulating these traces .
the rules are summarised in fig .
[ fig : rules ] .
note that , by combining traces of these rules , we obtain a relation between the fi terms , @xmath82 which allows us to eliminate one of the @xmath83 s . to ultimately obtain a form of the curve that allows immediate comparison with the sw curve as given in @xcite ,
our approach is to express the square of the highest - order invariant in terms of lower - order invariants . for this
we need the three @xmath41-invariants @xmath84 , @xmath85 and @xmath86 , as well as some simpler matrices @xmath87 , @xmath88 and @xmath89 that we define to simplify calculations ; they are all given in fig .
[ fig : e8defs ] .
using the schouten identity @xmath90 @xmath91 and the cyclic property of the trace we can reduce the square of the highest invariant , which , as can be seen from the diagram ( using a @xmath52 schouten identity ) , may be written as @xmath92 , to products of ( @xmath84 and ) traces of at most two of the matrices @xmath87 , @xmath88 , @xmath89 , @xmath93\right ) \\\nonumber & & + { \mathrm{tr}}(b^2)\left ( { \mathrm{tr}}(ac)^2+{\mathrm{tr}}(a){\mathrm{tr}}(c ) \left[{\frac{1}{2}}{\mathrm{tr}}(a){\mathrm{tr}}(c)-{\mathrm{tr}}(ac ) \right]\right ) \\\nonumber & & + { \mathrm{tr}}(c^2)\left ( { \mathrm{tr}}(ab)^2+{\mathrm{tr}}(a){\mathrm{tr}}(b ) \left[{\frac{1}{2}}{\mathrm{tr}}(a){\mathrm{tr}}(b)-{\mathrm{tr}}(ab ) \right]\right ) \\\nonumber & & -{\frac{1}{2}}{\mathrm{tr}}(a^2){\mathrm{tr}}(b^2){\mathrm{tr}}(c^2)+\\\nonumber & & + { \mathrm{tr}}(a){\mathrm{tr}}(b){\mathrm{tr}}(c)\left ( { \rule{0pt}{0.5 cm } } { \mathrm{tr}}(a){\mathrm{tr}}(bc)+{\mathrm{tr}}(b){\mathrm{tr}}(ac)+{\mathrm{tr}}(c){\mathrm{tr}}(ab ) \right ) \\\nonumber & & { \rule{0pt}{0.7 cm } } -{\mathrm{tr}}(a)^2{\mathrm{tr}}(bc)^2-{\mathrm{tr}}(b)^2{\mathrm{tr}}(ac)^2-{\mathrm{tr}}(c)^2 { \mathrm{tr}}(ab)^2 \\ & & { \rule{0pt}{0.7 cm } } -{\mathrm{tr}}(a)^2{\mathrm{tr}}(b)^2{\mathrm{tr}}(c)^2 - 2{\mathrm{tr}}(ab){\mathrm{tr}}(ac){\mathrm{tr}}(bc ) \label{eq : tracecurve}\end{aligned}\ ] ] having done this , we realize that in fact we need only to calculate a few traces .
however , these calculations are nontrivial and the results are much too lengthy to fit in this paper , so we only give the general form of the traces , as polynomials in @xmath84 , @xmath85 and @xmath86 , @xmath94 the coefficients @xmath95 , @xmath96 , etc .
, are polynomials in the fayet - iliopoulos terms @xmath83 , @xmath97 .
was eliminated using eq .
( [ eq : firel ] ) . ]
we note here that the invariants @xmath84 , @xmath85 and @xmath86 are of order 15 , 10 and 6 in the @xmath83 s , and the traces of @xmath87 , @xmath88 and @xmath89 are of order 3 , 5 and 7 , respectively . from this information it is easy to find the orders of the coefficients defined in ( [ eq : traces ] ) ; for instance , the ones that are most difficult to compute , , we found it was necessary to use the schouten identity to express @xmath98 . ]
@xmath99 and @xmath100 , are polynomials of order 10 and 12 , respectively , in eight variables . to compare our curve ( [ eq : tracecurve ] ) with the sw curve , we need to put it on its canonical form , @xmath101 where @xmath102 and the @xmath103 are polynomials of order @xmath104 in the @xmath83 s .
we accomplish this by plugging in the traces ( [ eq : traces ] ) into eq .
( [ eq : tracecurve ] ) and shifting our variables @xmath84 , @xmath85 , @xmath86 .
the substitution of traces puts our curve on the form @xmath105 where @xmath106 and the @xmath58 s and @xmath107 s are independent of the invariants @xmath84 , @xmath85 and @xmath86 .
to get rid of the @xmath84-term , we shift @xmath84 by @xmath108 . expanding the result
, we obtain a coefficient of the @xmath109-term such that we must shift @xmath85 by @xmath110 in order to eliminate @xmath109 .
finally , the @xmath111-term is substituted away by @xmath112 , and we end up with unwieldy expressions for the @xmath103 s , which should be compared to the seiberg - witten coefficients @xmath113 of @xcite . finding the explicit relation between our fi parameters and the mass parameters @xmath113 would be a next to impossible task unless we had a good idea of what it should be . guided by our conjecture that the higgs branch algebraic curve and the seiberg - witten curve are identical , and the fact that noguchi et al .
@xcite used casimir invariants to express their sw curve , we use @xmath0 casimirs written in terms of the fi parameters .
it is then straightforward , although very time consuming , to compare our coefficients with those of @xcite . to find the casimirs , first note that the simple roots of @xmath0 may , as we saw in section [ sec : quiver ] , be identified with the fi parameters via eq .
( [ eq : period ] ) , @xmath114 where @xmath115 is a 2-cycle behaving as a simple root , and @xmath116 is the khler form on the orbifold .
the casimir invariants @xmath117 may be found as @xcite @xmath118 where @xmath119 is the coefficient of @xmath120 in the characteristic polynomial @xmath121 where @xmath122 are the weights of the fundamental representation , and @xmath123 . factoring out @xmath124 and defining @xmath125 = \sum_{k=1}^{248 } v_k^n , \label{eq : chi}\ ] ]
we may rewrite ( [ eq : charpoly ] ) as @xmath126 a form which facilitates the extraction of coefficients .
writing the weights @xmath122 in the usual way as linear combinations of simple roots ( see appendix [ sec : weights ] ) , we thus obtain the casimirs @xmath117 expressed in fi parameters @xmath83 .
were useful in this calculation . ]
to actually compare @xmath103 and @xmath113 turned out to be quite demanding .
the highest - order coefficient @xmath127 is a polynomial of order 30 in eight variables , which means that it has in principle 10295472 terms .
we did our calculations in maple v software on a digital alpha workstation , but since it would not store a polynomial of that size in its memory , much less compare two of them , we tried to find other software that would , without success .
below we describe the partly numerical check we performed , which , short of writing a dedicated program , seems the best one possible , as well as totally convincing .
we explicitly checked up to @xmath128 by comparing the full expressions for the polynomials and got complete agreement .
the higher order @xmath129 s we could only check numerically and they agree when all the @xmath83 s are set to random and different prime numbers . we have thus shown that the @xmath0 quiver higgs branch is equal to the generalised coulomb branch of the sw theory with @xmath0 global symmetry .
the iib picture of d3-branes on a @xmath130 orbifold singularity ( which is of type @xmath131 ) is t - dual to a picture of type iia string theory in a background of d4-branes stretching between ns5-branes @xcite .
this dual picture , the hanany - witten ( hw ) picture @xcite , provides an intuitive geometric interpretation of blow - ups of @xmath131 type singularities .
an analogous picture exists for @xmath11 type singularities @xcite , and it seems plausible that there are generalisations also to @xmath5 , @xmath6 and @xmath0 . in this section , we analyse the hw picture for the @xmath130 case along the lines of @xcite ( see also @xcite ) ; in particular we clarify the role of the fayet - iliopoulos terms .
starting from the type iib string theory configuration ( @xmath132 means the object is extended in that direction , and @xmath133 means it is pointlike ) @xmath134 we t - dualise along the 6-direction to get @xmath135 in type iia string theory in flat spacetime .
there are @xmath104 ns5-branes , which all coincide in the 789 directions , but not necessarily in the 6-direction . between them
d4-branes are suspended , which are the t - duals of the iib d3-branes .
the rotational symmetry @xmath136 of the 789 coordinates translates into the @xmath137 symmetry of the gauge theory living on the d4-branes .
the hypermultiplets arise from fundamental strings stretching across the ns5-branes , between neighbouring d4-branes . resolving singularities in the iib picture
corresponds to separating ns5-branes along the 789 directions in the iia picture . by an @xmath7 rotation
we can always pick the direction of displacement to be @xmath138 .
note that such a displacement breaks the 789 rotational symmetry ; that is , blowing up a singularity breaks the @xmath137 symmetry .
if we move some of the ns5-branes in this way , with the d4-branes still stuck to them , and then t - dualise along @xmath139 again , we do not regain the d3-brane picture .
rather , the now tilted d4-branes dualise to a set of d5-branes ( with nonzero b - field ) with their 67 worldvolume coordinates wrapped on 2-cycles . shrinking these 2-cycles to zero size ,
each of the wrapped d5-branes is a fractional d3-brane , which can not move away from the singularity .
thus a fractional d3-brane corresponds to a d4-brane whose ends are stuck on ns5-branes . to move a fractional d3-brane , or , equivalently , a wrapped d5-brane , along the 6789 directions , we need to add @xmath140 images ( under @xmath16 ) , all associated with a 2-cycle each .
the sum of the full set of 2-cycles is homologically trivial and can be shrunk to zero size .
then the collection of wrapped d5-branes will look like a single d3-brane that can move around freely in the orbifold .
this procedure corresponds in the hw picture to starting out with a single d4-brane stretching between two of the @xmath104 ns5-branes , and wanting to move the d4-brane ( in the 7-direction , say ) away from the ns5-branes , detaching its ends . in order not to violate the boundary conditions of the d4-brane
, we then need to put one d4-brane between each unconnected pair of ns5-branes and join them at the ends .
we then get a total of @xmath104 d4-branes forming a single brane winding once around the periodic 6-direction .
the d4-brane may now be lifted off the ns5-branes and move freely , corresponding to the free d3-brane in the t - dual picture .
we may also gain some insight concerning the role played by the fi parameters in the hw picture , from the worldvolume theory of a wrapped d5-brane on the orbifold singularity .
consider such a brane living in the 012367 directions , with its 67 worldvolume coordinates wrapped on a 2-cycle @xmath79 .
the born - infeld and chern - simons terms in the worldvolume action are , schematically , @xcite @xmath141 where @xmath142 is the metric on the worldvolume , @xmath143 is the r - r @xmath144-form , @xmath145 is a constant , and @xmath146 where the 2-form @xmath147 is the field strength of the gauge field on the brane and @xmath148 is the ns - ns 2-form on the brane .
dimensional reduction to the 0123 directions , by integrating over the 2-cycle , puts the first term of ( [ eq : d5action ] ) on the form @xmath149 where @xmath150 , @xmath151 is the metric on the 67 directions , and @xmath152 is the metric on the 0123 directions . expanding ( [ eq : d5bired ] ) we obtain the coupling constant @xmath153 in four dimensions as the coefficient of @xmath154 .
it is just the factor on the left in ( [ eq : d5bired ] ) , which we can write as @xmath155 in the hw picture the coupling constant of the four dimensional theory is proportional to the length of the d4-brane in the additional fifth direction of the brane .
hence ( [ eq : gk ] ) measures the total distance between two ns5-branes between which the d4-brane is suspended . furthermore , since the distance between the ns5-branes in the isometry direction ( in our case @xmath156 ) is given by the flux of the @xmath148 field on the corresponding cycle , we have to interpret @xmath157 as the position of the ns5-branes in a direction orthogonal to that , let us choose @xmath158 .
movement of the ns5-branes in the remaining directions @xmath159 and @xmath160 now corresponds to turning on the @xmath137 partners of the khler form .
the integral of @xmath116 over a 2-cycle is also , by definition , a fayet - iliopoulos term .
a hyperkhler manifold has an su(2 ) manifold of possible complex structures . choosing a complex structure we can define the khler form @xmath116 as @xmath161 , and the holomorphic 2-form as @xmath162 .
these three 2-forms rotate into each other under @xmath137 transformations , corresponding to choosing a different complex structure .
the @xmath51:th triplet of fi terms is defined by the period of @xmath163 ( and hence also transforms as a triplet under @xmath137 ) , as @xmath164 hence @xmath165 where @xmath166 is the real component of the triplet of fi terms @xmath167 .
another way to obtain the fi terms of the four - dimensional yang - mills theory is via dimensional reduction and supersymmetrisation of the d5-brane worldvolume theory @xcite . the third term of ( [ eq : d5action ] )
can be rewritten as @xmath168 where @xmath169 is the hodge dual potential of @xmath170 in six dimensions .
after integration over the @xmath51:th 2-cycle we supersymmetrise this to @xmath171 where @xmath172 is a chiral superfield whose complex scalar component is @xmath173 , and @xmath174 is the vector superfield containing @xmath40 . here
the imaginary part @xmath166 of the scalar component is the real fi term in four dimensions , and we see that it arises as the superpartner of @xmath169 .
in this paper we have discussed the conjectured mirror symmetry between the higgs branch of quiver gauge theories and the generalised coulomb branch of certain four - dimensional strongly coupled , globally symmetric gauge theories .
this symmetry was hinted at in @xcite , where equivalence of the @xmath5 curves was established , and it was substantiated in @xcite , where equivalence of the @xmath6 curves was presented and the conjecture was made precise .
here we have shown the final equivalence , that of the @xmath0 curves for the two branches .
the proof we have given is by direct comparison
. it would be nice to find a proof such as a chain of dualities leading from one model to the other , as indicated in @xcite .
there are obstacles to this in that , e.g. , many of the dualities expected to enter that chain are not known . in this context
we emphasize an important aspect of our results , namely as a guide for finding hanany - witten type constructions .
assume that one is contemplating a hw picture of ns5-branes with d4-branes ending on them and that this is supposed to describe the iia t - dual of d3-branes on an ( @xmath5 , @xmath6 or ) @xmath0 singularity in iib .
moving the ns5-branes in the hw picture corresponds to blowing up the singularity in the dual picture . as discussed in section [ sec : hwpicture ] , the fi parameters give the position of the ns5-branes in the hw picture when they are moved . since we have given the relation of the fi parameters to the parameters governing the deformation of the algebraic variety , one may now check that the possible motions on the hw side ( allowed by the particular geometry suggested ) correspond to the known allowed deformations .
we have performed the above check for some of the known dualities and hope to use it in future efforts to find hw pictures of the @xmath175-theories .
finally we mention that all is set up for finding the remaining trace identities in appendix [ sec : traceids ] .
the calculations will be carried out as soon as the computer capacity becomes available to us .
the work of ul was supported in part by nfr grant 650 - 1998368 and by eu contract hprn - ct-2000 - 0122 .
rvu was supported by the czech ministry of education under contract no 143100006 .
he would also like to thank stockholm university for hospitality during part of this work .
the @xmath0 casimir invariants may be expressed in terms of fundamental weights as shown in section [ sec : e8curve ] .
the weights may in turn be expressed in terms of the @xmath0 simple roots . here
, the procedure is quite general . ]
we want to find the 248 weights of the fundamental representation of @xmath0 .
this is the same as the adjoint representation , and it is real , whence half of the weights are just minus the first half .
in addition one finds that eight of them are zero , so we end up with only 120 nonvanishing independent weights . viewing the simple roots @xmath83 as 8-vectors , the cartan matrix for @xmath0 , @xmath176 may be viewed as the matrix of scalar products @xmath177 .
if we start with a highest weight vector @xmath178 , lower weights are obtained by subtracting simple roots whose scalar products with @xmath178 are positive @xcite . to be explicit ,
choose @xmath178 such that its scalar product with @xmath179 is 1 and otherwise zero , i.e. @xmath178 would be represented by a row @xmath180 $ ] in the cartan matrix . then the only possible next weight is @xmath181 , represented by @xmath182 $ ] in the cartan matrix .
iterating this procedure , subtracting simple roots whose scalar products with the previous weight are positive , we obtain 120 nonvanishing weights @xmath183 in `` cartan representation '' . to express these in terms of simple roots , we invert the cartan matrix and compute @xmath122 as @xmath184 where @xmath185 denotes the @xmath186:th component of the vector @xmath183 , @xmath89 is the inverse of the cartan matrix , and the factor 2 is a normalisation . note the slight change in convention ; all the @xmath83 s here have the same sign , whereas two of our fi terms in section [ sec : e8curve ] ( @xmath187 and @xmath188 ) have opposite sign relative to the others . the traces
@xmath189 $ ] ( eq .
( [ eq : chi ] ) ) used in writing the casimir invariants in section [ sec : e8curve ] satisfy identities that simplify the calculations slightly .
such trace identities were derived by @xcite for the simple lie algebras up to @xmath5 , whereas the @xmath6 identities were calculated by the authors of @xcite .
for @xmath0 we found the following identities , @xmath190 although straightforward in principle , due to limitations of maple , we were not able to derive the rest of the trace identities , i.e. those expressing @xmath191 , @xmath192 , @xmath193 and @xmath194 in terms of the independent @xmath195 s .
77 k. kodaira , _ on compact analytic surfaces , ii - iii _ , ann . of math .
* 77 * ( 1963 ) 563 ; * 78 * ( 1963 ) 1 k. dasgupta and s. mukhi , _ f - theory at constant coupling _ , phys . lett . * b385 * ( 1996 ) 125 - 131 , hep - th/9606044 n. seiberg , _ five dimensional susy field theories , non - trivial fixed points and string dynamics _ ,
lett . * b388 * ( 1996 ) 753 - 760 , hep - th/9608111 u. lindstrm , m. roek and r. von unge , _ hyperkhler quotients and algebraic curves _ ,
jhep * 0001 * ( 2000 ) 022 , hep - th/9908082 i.y .
park and r. von unge , _ hyperkhler quotients , mirror symmetry , and f - theory _ , jhep * 0003 * ( 2000 ) 037 , hep - th/0001051 k. intriligator and n. seiberg , _ mirror symmetry in three dimensional gauge theories _ , phys . lett .
* b387 * ( 1996 ) 513 - 519 , hep - th/9607207 a. hanany and e. witten , _ type iib superstrings , bps monopoles , and three - dimensional gauge dynamics _ , nucl .
* b492 * ( 1997 ) 152 - 190 , hep - th/9611230 c.v .
johnson and r.c .
myers , _ aspects of type iib theory on ale spaces _ , phys .
* d55 * ( 1997 ) 6382 - 6393 , hep - th/9610140 m.r .
douglas and g. moore , _ d - branes , quivers , and ale instantons _ , hep - th/9603167 u. lindstrm and m. roek , _ scalar tensor duality and n=1,2 nonlinear @xmath196-models _ , nucl .
* b222 * ( 1983 ) 285 n.j .
hitchin , a. karlhede , u. lindstrm and m. roek , _ hyperkhler metrics and supersymmetry _ , commun . math
. phys . * 108 * ( 1987 ) 535 - 589 p.s .
aspinwall , _
k3 surfaces and string duality _ , lectures given at the tasi-96 summer school on strings , fields and duality , hep - th/9611137 j. minahan and d. nemeschansky , _ superconformal fixed points with e@xmath197 global symmetry _ , nucl
. phys .
* b489 * ( 1997 ) 24 - 46 , hep - th/9610076 m. noguchi , s. terashima and s - k .
yang , _ n=2 superconformal field theory with ade global symmetry on a d3-brane probe _ , nucl
. phys .
* b556 * ( 1999 ) 115 - 151 , hep - th/9903215 a. karch , d. lst and d.j .
smith , _ equivalence of geometric engineering and hanany - witten via fractional branes _ ,
* b533 * ( 1998 ) 348 - 372 , hep - th/9803232 a. kapustin , _
d@xmath197 quivers from branes _ ,
jhep * 9812 * ( 1998 ) 015 , hep - th/9806238 a. hanany and a. zaffaroni , _ issues on orientifolds : on the brane construction of gauge theories with @xmath198 global symmetry _ ,
jhep * 9907 * ( 1999 ) 009 , hep - th/9903242 k. dasgupta and s. mukhi , _ brane constructions , fractional branes and anti - desitter domain walls _ ,
jhep * 9907 * ( 1999 ) 008 , hep - th/9904131 c.v .
johnson , _ d - brane primer _ , lectures given at ictp , tasi , and busstepp , sec . 9 , hep - th/0007170 t. banks , m.r .
douglas and n. seiberg , _ probing f - theory with branes _ ,
b390 * ( 1997 ) 205 - 209 , hep - th/9606199 a. sen , _ f - theory and orientifolds _ , nucl .
* b475 * ( 1996 ) 562 - 578 , hep - th/9605150 o. dewolfe , t. hauer , a. iqbal and b. zwiebach , _ uncovering the symmetries on [ p , q ] 7-branes : beyond the kodaira classification _ , hep - th/9812028 h. georgi , _ lie algebras in particle physics _
, 2nd ed . ,
perseus books group ( 1999 ) a.j . macfarlane and h.
pfeiffer , _ on characteristic equations , trace identities and casimir operators of simple lie algebras _
, j. math .
phys . * 41 * no . 5 ( 2000 ) 3192 - 3225 , math - ph/9907024 | we show that the higgs branch of a four - dimensional yang - mills theory , with gauge and matter content summarised by an @xmath0 quiver diagram , is identical to the generalised coulomb branch of a four - dimensional superconformal strongly coupled gauge theory with @xmath0 global symmetry .
this is the final step in showing that there is a higgs - coulomb identity of this kind for each of the cases @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 and @xmath0 .
this series of equivalences suggests the existence of a mirror symmetry between the quiver theories and the strongly coupled theories .
we also discuss how to interpret the parameters of the quiver gauge theory in terms of the hanany - witten picture .
usitp-0102 + february 2001 + hep - th/0102038 + |
organizations including the census bureau , medical establishments , and internet companies collect and publish statistical information @xcite .
the census bureau may , for instance , publish the result of a query such as : `` how many individuals have incomes that exceed $ 100,000 ? '' . an implicit hope in this approach
is that aggregate information is sufficiently anonymous so as not to breach the privacy of any individual .
unfortunately , publication schemes initially thought to be `` private '' have succumbed to privacy attacks @xcite , highlighting the urgent need for mechanisms that are _ provably _ private .
the differential privacy literature @xcite has proposed a rigorous and quantifiable definition of privacy , as well as provably privacy - preserving mechanisms for diverse applications including statistical queries , machine learning , and pricing .
informally , for @xmath3 $ ] , a randomized mechanism is @xmath4-differentially private if changing a row of the underlying database the data of a single individual changes the probability of every mechanism output by at most an @xmath4 factor .
larger values of @xmath4 correspond to greater levels of privacy .
differential privacy is typically achieved by adding noise that scales with @xmath4 .
while it is trivially possible to achieve any level of differential privacy , for instance by always returning random noise , this completely defeats the original purpose of providing useful information . on the other hand ,
returning fully accurate results can lead to privacy disclosures @xcite . _
the goal of this paper is to identify , for each @xmath3 $ ] , the optimal ( i.e. , utility - maximizing ) @xmath4-differentially private mechanism . _
we consider databases with @xmath5 rows drawn from a finite domain @xmath6 .
every row corresponds to an individual .
two databases are _ neighbors _ if they coincide in @xmath7 rows .
a _ count query _
@xmath8 takes a database @xmath9 as input and returns the result @xmath10 that is the number of rows that satisfy a fixed , non - trivial predicate on the domain @xmath6 .
such queries are also called predicate or subset - sum queries ; they have been extensively studied in their own right @xcite , and form a basic primitive from which more complex queries can be constructed @xcite . a randomized mechanism with a ( countable ) range @xmath11 is a function @xmath12 from @xmath13 to @xmath11 , where @xmath14 is the probability of outputting the response @xmath15 when the underlying database is @xmath16 . for @xmath17 $ ]
, a mechanism @xmath12 is _ @xmath4-differentially private _ if the ratio @xmath18 lies in the interval @xmath19 $ ] for every possible output @xmath20 and pair @xmath21 of neighboring databases .
( we interpret @xmath22 as @xmath23 . ) intuitively , the probability of every response of the privacy mechanism and hence the probability of a successful privacy attack following an interaction with the mechanism is , up to a controllable @xmath4 factor , independent of whether a given user `` opts in '' or `` opts out '' of the database .
a mechanism is _ oblivious _ if , for all @xmath20 , @xmath24 whenever @xmath25 if the output distribution depends only on the query result .
most of this paper considers only oblivious mechanisms ; for optimal privacy mechanism design , this is without loss of generality in a precise sense ( see section [ sec : oblivious ] ) .
the notation and definitions above simplify for oblivious mechanisms and count queries .
we can specify an oblivious mechanism via the probabilities @xmath26 of outputting a response @xmath27 for each query result @xmath28 ; @xmath4-differential privacy is then equivalent to the constraint that the ratios @xmath29 lie in the interval @xmath19 $ ] for every possible output @xmath20 and query result @xmath30 .
the _ @xmath4- geometric mechanism _ is defined as follows .
when the true query result is @xmath31 , the mechanism outputs @xmath32 .
@xmath33 is a random variable distributed as a two - sided geometric distribution : @xmath34=\frac{1-\alpha}{1+\alpha } \alpha^{|z|}$ ] for every integer @xmath35 .
this ( oblivious ) mechanism is @xmath4-differentially private because the probabilities of adjacent points in its range differ by an @xmath4 factor and because the true answer to a count query differs by at most one on neighboring databases .
this paper pursues strong and general utility guarantees . just as differential privacy guarantees protection against every potential attacker , independent of its side information , we seek mechanisms that guarantee near - optimal utility to _ every _ potential user , independent of its side information and preferences .
we now formally define preferences and side information .
we model the preferences of a user via a _ loss function _ @xmath36 ; @xmath37 denotes the user s loss when the query result is @xmath38 and the mechanism s ( perturbed ) output is @xmath15 .
we allow @xmath36 to be arbitrary , subject only to being nonnegative , and nondecreasing in @xmath39 for each fixed @xmath38 .
for example , the loss function @xmath39 measures mean error , the implicit measure of ( dis)utility in most previous literature on differential privacy .
two among the many other natural possibilities are @xmath40 , which essentially measures variance of the error ; and the binary loss function @xmath41 , defined as @xmath42 if @xmath43 and @xmath23 otherwise .
we model the side information of a user as a prior probability distribution @xmath44 over the query results @xmath28 .
this prior represents the beliefs of the user , which might stem from other information sources , previous interactions with the mechanism , introspection , or common sense .
we emphasize that we are _ not _ introducing priors to weaken the definition of differential privacy ; we use the standard definition of differential privacy ( which makes no assumptions about the side information of an attacker ) and use a prior only to discuss the _ utility _ of a ( differentially private ) mechanism to a potential user . now consider a user with a prior @xmath44 and loss function @xmath36 and an oblivious mechanism @xmath12 with range @xmath11 . for a given input @xmath16 with query result @xmath45 ,
the user s expected loss is @xmath46 , where the expectation is over the coin flips internal to the mechanism . the user s prior then yields a measure of the mechanism s overall ( dis)utility to the the user : @xmath47 this is simply the expected loss over the coin tosses of the mechanism and the prior .
we can then define the _ optimal _ @xmath4-differentially private mechanism for a user as one that minimizes the user - specific objective function ( [ eq : utility ] ) .
could a single mechanism be good simultaneously for all users ? a crucial observation for an affirmative answer is that a user has the power to post - process the output of a privacy mechanism , and that such post - processing can decrease the user s expected loss .
fix a database size @xmath5 that is odd .
consider a user with the binary loss function @xmath48 and prior @xmath49 , @xmath50 for all @xmath51 , that interacts with the @xmath4-geometric mechanism . without post - processing ,
i.e. , when the user accepts the mechanism s outputs at face value , the user s expected loss is @xmath52 .
if the user maps outputs of the geometric mechanism that are at least @xmath53 to @xmath5 and all other outputs to @xmath42 , it effectively induces a new mechanism with the much smaller expected loss of @xmath54 . in general , a ( randomized ) _ remap _ of a mechanism @xmath12 with range @xmath11 is a probabilistic function @xmath55 , with @xmath56 denoting the probability that the user reinterprets the mechanism s
response @xmath57 as the response @xmath20 .
a mechanism @xmath12 and a remap @xmath55 together induce a new ( @xmath4-differentially private ) mechanism @xmath58 with @xmath59 .
we assume that a ( rational ) user with prior @xmath60 and loss function @xmath36 , interacting with a publicly known mechanism @xmath12 , employs a remap @xmath55 that induces the mechanism @xmath58 that minimizes its expected loss over all such remaps .
it is well known ( e.g. ( * ? ? ?
* chapter 9 ) ) and easy to see that , among all possible ( randomized ) remappings , the optimal one follows from applying bayes rule and then minimizing expected loss .
precisely , for each response @xmath15 of @xmath12 , compute the posterior probability over query results : for every @xmath61 , @xmath62 . then , choose the query result @xmath63 that minimizes expected loss subject to the posterior and set @xmath64 and @xmath65 for @xmath66 .
this remap is simple , deterministic , and can be computed efficiently .
1 in [ fig : factor ] our main result is that for every @xmath3 $ ] , _ the @xmath4-geometric mechanism is simultaneously optimal for every rational user . _
[ thm : main ] let @xmath67 denote the @xmath4- geometric mechanism for some database size @xmath68 and privacy level @xmath3 $ ] , and let @xmath69 denote an optimal remap of @xmath67 for the user @xmath1 with prior @xmath60 and ( monotone ) loss function @xmath36
. then @xmath70 minimizes @xmath1 s expected loss over all oblivious , @xmath4-differentially private mechanisms with range @xmath71 .
this is an extremely strong utility - maximization guarantee : _ every _ potential user @xmath1 , no matter what its side information and preferences , derives as much utility from the geometric mechanism as it does from interacting with a differentially private mechanism @xmath2 that is optimally tailored to @xmath1 .
we reiterate that the prior from the utility model plays no role in the definition of privacy , which is the standard , worst - case ( over adversaries with arbitrary side - information and intent ) guarantee provided by differential privacy .
we emphasize that while the geometric mechanism is user - independent ( all users see the same distribution over responses ) , different users remap its responses in different ways , as informed by their individual prior distributions and loss functions .
rephrasing theorem [ thm : main ] , for every user there is an optimal mechanism for it that factors into a user - independent part the @xmath4-geometric mechanism and a user - specific computation that can be delegated to the user .
( see figure [ fig : factor ] . )
theorem [ thm : main ] shows how to achieve the same utility as a user - specific optimal mechanism without directly implementing one .
direct user - specific optimization would clearly involve several challenges .
first , it would require advance knowledge or elicitation of user preferences , which we expect is impractical in most applications . and
even if a mechanism was privy to the various preferences of its users , it would effectively need to answer the same query in different ways for different users , which in turn degrades its differential privacy guarantee . in theorem [ thm : main ]
, the restriction to oblivious mechanisms is , in a precise sense , without loss of generality .
( see section [ sec : oblivious ] . )
the restriction to the range @xmath71 effectively requires that the mechanism output is a legitimate query result for some database ; this type of property is called `` consistency '' in the literature ( e.g. @xcite ) .
differential privacy is motivated in part by the provable impossibility of absolute privacy against attackers with arbitrary side information @xcite .
one interpretation of differential privacy is : no matter what prior distribution over databases a potential attacker has , its posterior after interacting with a differentially private mechanism is almost independent of whether a given user `` opted in '' or `` opted out '' of the database @xcite .
below we discuss the papers in the differential privacy literature closest to the present work ; see @xcite for a recent , thorough survey of the state of the field .
dinur and nissim @xcite showed that for a database with @xmath5 rows , answering @xmath72 randomly chosen subset count queries with @xmath73 error allows an adversary to reconstruct most of the rows of the database ( a blatant privacy breach ) ; see dwork et al .
@xcite for a more robust impossibility result of the same type .
most of the differential privacy literature circumvents these impossibility results by focusing on interactive models where a mechanism supplies answers to only a sub - linear ( in @xmath5 ) number of queries .
count queries ( e.g. @xcite ) and more general queries ( e.g. @xcite ) have been studied from this perspective .
blum et al . @xcite take a different approach by restricting attention to count queries that lie in a restricted class ; they obtain non - interactive mechanisms that provide simultaneous good accuracy ( in terms of worst - case error ) for all count queries from a class with polynomial vc dimension .
kasiviswanathan et al .
@xcite give further results for privately learning hypotheses from a given class .
the use of abstract `` utility functions '' in mcsherry and talwar @xcite has a similar flavor to our use of loss functions , though the motivations and goals of their work and ours are unrelated .
motivated by pricing problems , mcsherry and talwar @xcite design differentially private mechanisms for queries that can have very different values on neighboring databases ( unlike count queries ) ; they do not consider users with side information ( i.e. , priors ) and do not formulate a notion of mechanism optimality ( simultaneous or otherwise ) .
finally , in recent and independent work , mcsherry and talwar ( personal communication , october 2008 ) also apply linear programming theory in the analysis of privacy mechanisms . again
, their goal is different : they do not consider a general utility model , but instead ask how expected error must scale with the number of queries answered by a differentially private mechanism .
this section proves theorem [ thm : main ] .
the proof has three high - level steps . 1 . for a given user @xmath1
, we formulate the problem of determining the differentially private mechanism that minimizes expected loss as a solution to a linear program ( lp ) .
the objective function of this lp is user - specific , but the feasible region is not .
2 . we identify several necessary conditions met by every privacy mechanism that is optimal for some user .
3 . for every privacy mechanism that satisfies these conditions ,
we construct a remap @xmath55 such that @xmath74 . by assumption
, a rational user employs an `` optimal remap '' of @xmath67 , so the mechanism induced by this map must be optimal for the user @xmath1 . fix a database size @xmath5 and a privacy level @xmath4 .
theorem [ thm : main ] is trivially true for the degenerate cases of @xmath75 .
so , we assume that @xmath76 . for every fixed user with loss function @xmath36 and prior @xmath60 ,
the formulation of privacy constraints in section [ sec : dp ] together with the objective function yields the following lp whose solution is an optimal mechanism for this user .
@xmath77 since the lp is bounded and feasible , we have the following ( e.g. @xcite ) .
[ lem : lp ] every user - specific lp has an optimal solution that is a vertex .
for the rest of this section , fix a user with prior @xmath44 and a loss function @xmath37 that is monotone in @xmath39 for every @xmath78 .
fix a mechanism @xmath12 that is optimal for this user , and also a vertex of the polytope of the user - specific lp .
vertices can be uniquely identified by the set of all constraints that are tight at the vertex .
this motivates us to characterize the _ state of constraints _ ( slack or tight ) of mechanisms that are optimal for some user .
we now begin the second step of the proof .
we will view @xmath12 as a @xmath79-matrix where rows correspond to query results ( inputs ) and columns correspond to query responses ( outputs ) .
we state our necessary conditions in terms of an @xmath80 _ constraint _ matrix @xmath81 associated with the mechanism @xmath12 .
row @xmath38 of the constraint matrix corresponds to rows @xmath38 and @xmath82 of the corresponding mechanism .
every entry of @xmath83 , for @xmath84 , @xmath85 takes on exactly one of four values . if @xmath86 then @xmath87 . if @xmath88 , then there are three possibilities . if @xmath89 then @xmath90 . if @xmath91 then @xmath92 .
otherwise @xmath93 . [ cols="^,^,^,^,^,^,^",options="header " , ] [ f : simmech ] we are now ready to prove theorem [ thm : imposs ] .
theoremthm : imposs by the previous lemma and the definitions of @xmath94 and @xmath95 , the mechanism @xmath12 must have the form in figure [ f : simmech ] . because @xmath12 is @xmath96-differentially private and as @xmath97 and @xmath98 differ in exactly two rows , columns @xmath5 and @xmath99 yield the following inequalities : @xmath100 , and @xmath101 . adding the two inequalities
, we have : @xmath102 . because all the entries of @xmath12 are probabilities , we have @xmath103 .
so it must be that @xmath104 . by a similar argument applied to the databases @xmath105 } and @xmath106
, we can show that @xmath107 .
thus , the probability masses on @xmath36 when the underlying databases are @xmath97 and @xmath108 are @xmath109 and @xmath110 respectively . but
this violates privacy because , the two databases differ in exactly one row but the two probabilities are not within a factor @xmath111 of each other .
we proposed a model of user utility , where users are parametrized by a prior ( modeling side information ) and a loss function ( modeling preferences ) .
theorem [ thm : main ] shows that for every fixed count query , database size , and level of privacy , there is a single simple mechanism that is simultaneously optimal for all rational users .
are analogous results possible for other definitions of privacy , such as the additive variant of differential privacy ( see @xcite ) ?
is an analogous result possible for other types of queries or for multiple queries at once ? when users have priors over databases ( theorem [ thm : imposs ] ) , are any positive results ( such as simultaneous _ approximation _ ) achievable via a single mechanism ?
we thank preston mcafee , john c. mitchell , rajeev motwani , david pennock and the anonymous referees .
l. backstrom , c. dwork , and j. kleinberg .
wherefore art thou r3579x ? : anonymized social networks , hidden patterns , and structural steganography . in
_ proceedings of the 16th international conference on world wide web ( www ) _ , pages 181190 , 2007 .
b. barak , k. chaudhuri , c. dwork , s. kale , f. mcsherry , and k. talwar .
privacy , accuracy , and consistency too : a holistic solution to contingency table release . in _ proceedings of the 26th acm sigact - sigmod - sigart symposium on principles of database systems ( pods ) _ , pages 273282 , 2007 .
a. blum , c. dwork , f. mcsherry , and k. nissim .
practical privacy : the sulq framework . in _ proceedings of the 24th acm sigact - sigmod - sigart symposium on principles of database systems ( pods ) _ , pages 128138 , 2005 . c. dwork .
differential privacy . in _ proceedings of the 33rd annual international colloquium on automata , languages , and programming ( icalp )
_ , volume 4051 of _ lecture notes in computer science _ ,
pages 112 , 2006 . c. dwork .
differential privacy : a survey of results . in _
5th international conference on theory and applications of models of computation ( tamc ) _ , volume 4978 of _ lecture notes in computer science _ , pages 119 , 2008 . c. dwork , f. mcsherry , k. nissim , and a. smith . calibrating noise to sensitivity in private data analysis . in _
third theory of cryptography conference ( tcc ) _ , volume 3876 of _ lecture notes in computer science _ , pages 265284 , 2006 . c. dwork and k. nissim .
privacy - preserving datamining on vertically partitioned databases . in _
24th annual international cryptology conference ( crypto ) _ , volume 3152 of _ lecture notes in computer science _ , pages 528544 , 2004 .
s. p. kasiviswanathan , h. k. lee , k. nissim , s. raskhodnikova , and a. smith .
what can we learn privately ?
in _ proceedings of the 49th annual ieee symposium on foundations of computer science ( focs ) _ , pages 531540 , 2008 . | a mechanism for releasing information about a statistical database with sensitive data must resolve a trade - off between utility and privacy .
publishing fully accurate information maximizes utility while minimizing privacy , while publishing random noise accomplishes the opposite .
privacy can be rigorously quantified using the framework of _ differential privacy _ , which requires that a mechanism s output distribution is nearly the same whether or not a given database row is included or excluded .
the goal of this paper is strong and general utility guarantees , subject to differential privacy .
we pursue mechanisms that guarantee near - optimal utility to every potential user , independent of its side information ( modeled as a prior distribution over query results ) and preferences ( modeled via a loss function ) .
our main result is : for each fixed count query and differential privacy level , there is a _ geometric mechanism _
@xmath0 a discrete variant of the simple and well - studied laplace mechanism that is _ simultaneously expected loss - minimizing _ for every possible user , subject to the differential privacy constraint .
this is an extremely strong utility guarantee : _ every _ potential user @xmath1 , no matter what its side information and preferences , derives as much utility from @xmath0 as from interacting with a differentially private mechanism @xmath2 that is optimally tailored to @xmath1 .
more precisely , for every user @xmath1 there is an optimal mechanism @xmath2 for it that factors into a user - independent part ( the geometric mechanism @xmath0 ) followed by user - specific post - processing that can be delegated to the user itself .
the first part of our proof of this result characterizes the optimal differentially private mechanism for a fixed but arbitrary user in terms of a certain basic feasible solution to a linear program with constraints that encode differential privacy .
the second part shows that all of the relevant vertices of this polytope ( ranging over all possible users ) are derivable from the geometric mechanism via suitable remappings of its range . |
let @xmath6 be a bounded , positive , continuous , symmetric function defined for @xmath7 . here
we address nonlocal cauchy problems of the form @xmath8 dy \quad \mbox{in } \
{ \mathbb{r}}_+ \times { \mathbb{r}}^n\end{aligned}\ ] ] with the initial condition @xmath9 satisfying @xmath10 .
on @xmath1 we also assume the basic condition @xmath11 by the symmetry of @xmath1 we have the same property exchanging @xmath12 and @xmath13 .
our main interest here is the asymptotic behavior of the solution of as @xmath14 .
it is well known that the diffusive nature of the equation implies that the solution goes to zero when @xmath14 . for smooth kernels
@xmath1 with compact support , it is proven in @xcite that the solution @xmath15 of the equation has the decay estimate @xmath16 for any @xmath17 .
note that this decay rate is the same as the one that holds for solutions of the classical heat equation . in the case of an equation in convolution form , that is when @xmath18 with @xmath19 a nonnegative radial function , not necessarily compactly supported , it is proven in @xcite that the solutions of equations with the form have the decay estimate @xmath20 provided the function @xmath19 has a fourier
transform satisfying the expansion @xmath21 , where @xmath22 is a constant .
it is remarkable that in this case the decay estimate is analogous to the one for the @xmath23order fractional heat equation , @xmath24 , with @xmath25 .
we also note that the convolution form of the equation allows the use of fourier analysis to obtain this result . in this work
we consider kernels @xmath1 not in convolution form and not compactly supported in order to investigate the diffusive role of the asymptotic tails of the kernels when doing decay estimates .
we illustrate this feature asking for an asymptotic profile for @xmath1 with the following lower bound @xmath26 for certain constants @xmath27 and @xmath25 . for simplicity , to clarify the arguments we will assume @xmath28 .
we remark that the use of fourier analysis is not helpful here due to fact that our operator is not in convolution form . despite of this difficulty ,
energy methods can be carried out in order to obtain the main result of this paper , which reads as follows : [ teo ] let @xmath29 and @xmath1 be a bounded , nonnegative , continuous , symmetric function defined in @xmath30 satisfying and .
then , for each @xmath31 , the solution of associated to an initial condition @xmath10 goes to zero in @xmath32 as @xmath14 .
moreover , the solution has an asymptotic decay that can be bounded by @xmath33 and , for all @xmath34 , it holds @xmath35 where the constant @xmath36 depends on @xmath37 and @xmath38 . for the case
@xmath39 $ ] we have the following [ cor ] let @xmath29 and @xmath1 be a bounded , nonnegative , continuous , symmetric function defined in @xmath30 satisfying and .
let @xmath15 be the solution of associated to an initial condition @xmath10 . if @xmath39 $ ] , then the asymptotic decay given by holds .
moreover , for all @xmath40 and for all @xmath41 we have @xmath42 with @xmath36 independent of @xmath43 . note that the kernel @xmath1 defining the nonlocal operator in shares , at one hand , the `` fat '' tails of the fractional laplacian of order @xmath44 , and on the other , the integrability property of zero order operators .
however , the decay rate of the solution of problem is as in the case of the fractional laplacian ( see ) and not as in the classical zero order case defined by compactly supported kernels .
this shows the diffusive effect of the fractional laplacian is more related with the tails of the kernel , which allows to transport mass to infinity through long jumps , than the highly intense , but shorter jumps related with the singularity at the origin .
concerning the existence and uniqueness of problem , the symmetry , boundedness and integrability assumptions over @xmath1 , allows us to perform a fixed point argument to obtain the following result whose proof is omitted .
[ existence ] let @xmath45 be a bounded and symmetric function satisfying assumption and , and let @xmath10 .
then , there exists a unique solution @xmath46 of equation .
this solution satisfies @xmath47 and @xmath48 for all @xmath49 .
the symmetry assumption over @xmath1 allows us to use an energy approach in order to get theorem [ teo ] . roughly speaking
, this assumption allows us to integrate by parts " equation .
for @xmath50 we multiply the equation by @xmath51 and integrate , obtaining the inequality @xmath52 where we have avoided the dependence on @xmath5 of the function @xmath15 for simplicity .
now we recall the following inequality ( whose proof is given in the appendix ) : let @xmath50 and @xmath53 . then , there exists a constant @xmath36 depending only on @xmath54 , such that @xmath55 hence , using this inequality into , we conclude @xmath56 our strategy is to get an estimate for the energy @xmath57 in the right - hand side of the above expression in terms of a higher @xmath58norm of @xmath15 .
this can be accomplished due to the asymptotic behavior of the kernel .
being similar to the tails of the kernel of the fractional laplacian of order @xmath44 , we compare @xmath59 in with an ad - hoc fractional seminorm , for which sobolev - type inequalities are available .
we recall that for @xmath25 and @xmath17 , @xmath60 is the fractional sobolev space of all @xmath61 functions with bounded fractional seminorm @xmath62_{\sigma , q}$ ] , given by @xmath63_{\sigma , q}^q = \iint_{{\mathbb{r}}^{2n } } \frac{|v(x + z ) - v(x)|^q}{|z|^{n + q\sigma } } dx\ , dz.\end{aligned}\ ] ] under these definitions , the key step in our approach is to relate the energy in with the corresponding fractional seminorm defined in .
once this is accomplished , we apply the following fractional sobolev - type inequality , which asserts the existence of a constant @xmath64 such that for each @xmath65 with @xmath66 , it holds @xmath67 ( see @xcite ) . by the use of the above inequality we get the desired increment of the @xmath58norm of the solutions , and
after this the proof follows standard arguments .
once we have obtained theorem [ teo ] which is valid for @xmath68 , the corresponding decay estimate for the remaining case @xmath39 $ ] can be obtained by interpolation .
hence , at the very end our arguments rely on estimates for the nonlinear operator in the right hand side of .
first , we consider a positive smooth function @xmath69 with the following properties @xmath70 with the aid of this function , we split a function @xmath15 into two parts
. we will denote the `` smooth '' part of @xmath15 as @xmath71 and the remaining as @xmath72 .
we let @xmath73 as a first property of this decomposition we have that each @xmath74 norm of the functions @xmath71 and @xmath72 is controlled by the corresponding norm of @xmath15 .
[ vw < u ] let @xmath71 and @xmath72 be given by . for each @xmath75 , there exists @xmath76 such that @xmath77 we start with @xmath71 . using its definition , we have @xmath78 \qquad\displaystyle = \int_{{\mathbb{r}}^n } \big{| } \int_{{\mathbb{r}}^n } \psi(x- y)^{1/q ' } \psi(x - y)^{1/q } u(y ) dy \big{|}^q dx \\[10pt ] \qquad\displaystyle \leq \int_{{\mathbb{r}}^n } \big { [ } \big { ( } \int_{{\mathbb{r}}^n } \psi(x - y ) dy\big{)}^{1/q ' } \big{(}\int_{{\mathbb{r}}^n } \psi(x - y)|u(y)|^q dy\big{)}^{1/q } \big{]}^q dx \\[10pt ] \qquad\displaystyle \leq c(q , \psi ) \int_{{\mathbb{r}}^n } \int_{{\mathbb{r}}^n } \psi(x - y ) |u(y)|^q dy dx \\[10pt ] \qquad\displaystyle = c(q , \psi ) \int_{{\mathbb{r}}^n } |u(y)|^q \int_{{\mathbb{r}}^n } \psi(x - y ) dx dy \\[10pt ] \displaystyle \qquad \leq c(q , \psi ) \int_{{\mathbb{r}}^n } |u(y)|^q dy . \end{array}\ ] ] the inequality conrresponding to @xmath72 easily follows from the triangular inequality in @xmath74 .
now we state a key result to get the desired estimate on the decay rate .
[ keyprop ] let @xmath79 a bounded and symmetric function satisfying hypotheses and .
let @xmath80 satisfying and @xmath81 .
then , there exists a constant @xmath64 such that for all @xmath82 and @xmath83 defined in , we have @xmath84_{2\sigma q^{-1 } , q}^q + \|w\|_{l^q({\mathbb{r}}^n)}^q \leq c \iint_{{\mathbb{r}}^{2n } } j(x , y ) |u(x ) - u(y)|^q dx\ , dy.\ ] ] the constant @xmath36 depends on @xmath85 and @xmath38 . for the estimate concerning @xmath72 , we have @xmath86 applying holder s inequality , we get @xmath87\displaystyle \leq \int_{{\mathbb{r}}^n } \big { ( } \int_{{\mathbb{r}}^n } \psi(x - z)dz \big{)}^{q / q ' } \big { ( } \int_{{\mathbb{r}}^n}\psi(x - z ) |u(x ) - u(z)|^q dz \big { ) } dx \\[10pt]\displaystyle \leq c(q , q ' , \psi ) \int_{{\mathbb{r}}^n } \int_{{\mathbb{r}}^n } \psi(x - z ) |u(x ) - u(z)|^q dz\ , dx . \end{array}\ ] ] since @xmath80 is supported in @xmath88 , clearly we have that for all @xmath89 , @xmath90 .
meanwhile , since @xmath1 is positive , when @xmath91 there exists a constant @xmath36 depending only on @xmath92 such that @xmath93 . then @xmath94 now we deal with the term concerning @xmath71 .
we split the fractional seminorm as @xmath95_{2\sigma q^{-1 } , q}^q & = \ \iint_{|x - y| > 1 } \frac{|v(x ) - v(y)|^q}{|x - y|^{n + 2\sigma}}dxdy + \iint_{|x - y| \leq 1 } \frac{|v(x ) - v(y)|^q}{|x - y|^{n + 2\sigma}}dxdy \\ & = : \
i_{ext } + i_{int}. \end{split}\ ] ] we look at these integrals separately . for @xmath96 , using the definition of @xmath71 we have @xmath97 now , we can look at the measure @xmath98 as a probability measure because of . clearly , the function @xmath99 is convex in @xmath100 .
then , we apply jensen s inequality on the @xmath101integral in right - hand side of the last expression , concluding @xmath102 which , after an application of fubini s theorem , gives @xmath103 then , applying the change @xmath104 in the @xmath105 integral and using , we conclude @xmath106 using this last expression , we obtain from the assumption that @xmath107 now we deal with @xmath108 . in this case , using the definition of @xmath80 , we can write @xmath109 note that by using , we have for all @xmath110 @xmath111 and then @xmath112 thus , using this equality into , we get @xmath113 however , note that if @xmath114 in the @xmath115 integral , since @xmath116 necessarily @xmath117 .
then , due to the fact that @xmath80 is supported in the unit ball , the contribution of the integrand when @xmath114 is null in the @xmath115 integral . taking this into account , applying hlder s inequality into the @xmath101integral , we have @xmath118 where @xmath119 is the hlder conjugate to @xmath54 . by fubini s theorem
we can write @xmath120 where latexmath:[\[\psi(x ) = \int_{|x - y| < 1 } \big { ( } \int_{|x - \tilde{z}| < 2 } |\psi(x - \tilde{z } ) - \psi(y - \tilde{z})|^{q ' } d\tilde{z } \big{)}^{q / q ' } using the regularity of @xmath80 , we have @xmath122 and since @xmath68 , we conclude the last integral is convergent , obtaining @xmath123 which leads us to the following estimate for @xmath108 @xmath124 from this , it is easy to get @xmath125 which by the use of , let us conclude that @xmath126 this last estimate together with concludes the proof .
for simplicity , we use the following notation : for @xmath50 , we denote @xmath127 .
hence , the fractional critical exponent of @xmath54 relative to @xmath128 is denoted by @xmath129 and is given by @xmath130 we need an intermediate result .
[ lemanormaulq ] let us assume the hypotheses of theorem [ existence ] and proposition [ keyprop ] .
then , there exists two constants @xmath131 depending on @xmath132 and @xmath133 such that @xmath134 is defined in and @xmath135 satisfies the equality @xmath136 * proof : * using the definition of @xmath71 and @xmath72 in , we have latexmath:[\[\label{ineqlemanormalq1 } by the definition of @xmath71 , it belongs to @xmath138 for all @xmath139 .
hence , we can interpolate , obtaining @xmath140 where @xmath141 is given by . recalling the sobolev - type inequality
, we have @xmath142_{\tilde{\sigma } , q}^\theta ||v||_{l^1({\mathbb{r}}^n)}^{1 - \theta},\ ] ] and then , using this last inequality , , proposition [ keyprop ] and the property @xmath47 given in theorem [ existence ] , we get . * proof of theorem [ teo ] . * along this proof
, we denote by @xmath143 a constant independent of @xmath43 and @xmath144 a constant depending on @xmath43 ( making explicit this dependence ) .
we recall that @xmath68 .
consider the constants @xmath145 of the previous lemma and denote @xmath146 with these definitions , inequality can be written as @xmath147 or equivalently @xmath148 hence , denoting @xmath149(t):= ||u(\cdot , t)||_{{l^q({\mathbb{r}}^n)}}^q$ ] , we use to conclude @xmath150 satisfies the differential inequality @xmath151 direct computations allows us to write @xmath152 where @xmath153 for some constant @xmath154 . using this expression , we obtain from that , for all @xmath41 , @xmath155 satisfies @xmath156 we claim that there exists a @xmath157 such that @xmath158 . otherwise , if @xmath159 , from we have @xmath160 holds for all @xmath41 , implying that @xmath150 has exponential decay , which is a contradiction with @xmath161 .
hence , since there exists such @xmath162 and since implies in particular that @xmath150 is nonincreasing , from we conclude that for all @xmath163 , @xmath164 from this we get @xmath165 again for @xmath36 depending on @xmath132 and @xmath133 .
this proves that the @xmath74-norm of the solution goes to zero with the desired rate , that is , holds .
now we look for a bound valid for all @xmath34 , that is . if @xmath166 , that is when @xmath167 , by and since @xmath150 is nonincreasing , we get @xmath168 for all @xmath41 . integrating from @xmath169 to @xmath5 , using the expression for @xmath141 and recalling @xmath170 , we conclude @xmath171 where the constant @xmath36 does not depend on @xmath43 . in the case
@xmath172 , the corresponding inequality @xmath173 allows us to replace the dependence on @xmath174 of @xmath175 in by the norm @xmath176 .
thus , this replacement makes the analogous inequality to that in this case reads as @xmath177 where , for some @xmath178 , we have denoted @xmath179 since @xmath150 is nondecreasing our interest is to estimate @xmath180 in from below in the interval @xmath181 .
it is easy to see that for all @xmath36 small enough , independent of @xmath43 we can get @xmath182 which means that @xmath183 replacing this inequality into we obtain that @xmath184 holds for all @xmath41 , where @xmath185 is independent of @xmath43 .
this inequality is exactly the same as but with the @xmath186norm of @xmath43 replacing the @xmath187norm .
integrating , we conclude the estimate @xmath188 and therefore inequality holds .
* proof of corollary [ cor ] : * by an interpolation argument , there exists @xmath135 such that @xmath189 where we have used theorem [ teo ] in the last inequality .
* proof of inequality :* clearly the inequality holds if @xmath190 , then , without loss of generality , we assume @xmath191 .
if this is the case , is equivalent to @xmath192 for all @xmath193 .
now , in the case @xmath194 we have the left - hand side is bounded from above by 1 , meanwhile the right - hand side is bounded from above by @xmath195 .
hence , it is sufficient to consider @xmath196 . for the case
@xmath197 , the desired inequality is equivalent to @xmath198 when @xmath202 , the conclusion follows from the fact that the derivative of @xmath203 is strictly positive ( independently of @xmath36 ) at @xmath204 .
then , choosing @xmath64 small enough we obtain the result .
* aknowledgements : * j.d.r .
was partially supported by mec mtm2010 - 18128 and mtm2011 - 27998 ( spain ) and fondecyt 1110291 international coperation .
was partially supported by fondecyt 1110291 and programa basal - cmm u. de chile .
was partially supported by conicyt , grants capital humano avanzado , realizacin de tesis doctoral and cotutela en el extranjero . | in this paper we obtain bounds for the decay rate for solutions to the nonlocal problem @xmath0 dy$ ] . here
we deal with bounded kernels @xmath1 but with polynomial tails , that is , we assume a lower bound of the form @xmath2 , for @xmath3 .
our estimates takes the form @xmath4 for @xmath5 large . |
experiments have raised the issue of the relative stability of various fractional quantum hall states in graphene . the earliest results in suspended graphene in the two - terminal geometry @xcite gave indications of fractional quantum hall effect ( fqhe ) at @xmath0 .
the more recent experiments by dean _
et al _ and ghahari _ et al _ @xcite in four - terminal hall bar geometry for graphene on boron nitride hexagonal substrate have revealed a much richer structure .
specifically , they observe fqhe at filling factors @xmath1 and in the central ( @xmath2 ) landau level , and at @xmath3 in the @xmath4 landau level . in addition
, they also see evidence fqhe at @xmath5 in the @xmath2 ll , but none at @xmath6 .
hansel , lafkioti , and v. krsti have also reported 4/3 fqhe in a four - terminal geometry @xcite .
more recently , feldman , krauss , smet and yacoby @xcite have observed incompressible states in suspended flakes in the sequence @xmath7 for all @xmath8 in the @xmath9 range , but only at @xmath10 and @xmath11 in the @xmath12 range .
the difference of the @xmath13 and @xmath12 interval , and the absence of fqhe at @xmath14 in particular , remains a puzzle .
our treatment below builds upon our understanding of the fqhe in gaas systems . in gaas systems ,
the fqhe is associated with the formation of composite fermions ( cfs ) , bound states of electrons and an even number of quantized vortices @xcite . the electronic ll splits into landau - like levels of composite fermions , called @xmath15 levels ( @xmath15ls ) , and the integer quantum hall effect of composite fermions produces the prominently observed fqhe at @xmath16 , where @xmath17 is the number of completely filled @xmath15 levels and @xmath18 is the number of vortices bound to electrons .
the cf theory has been extended to include spin @xcite : with spin , fqhe occurs at the same fractions as before , namely @xmath16 , but now , in general , fqhe states with different spin polarizations are available at each fraction , analogous to the situation at integer fillings where different spin polarizations would occur depending on the strength of the zeeman splitting relative to the cyclotron energy .
experiments have measured the transitions between differently polarized states , the actual spin polarizations of the states , cf @xmath15 level fan diagram , and the phase boundaries @xcite , which are all found to be in good agreement with the cf theory .
( for the integer quantum hall effect , this physics is not relevant for gaas , because the zeeman energy is very small compared to the cyclotron energy , thus producing the state with the smallest spin polarization .
however , for fqhe , the zeeman energy can be of the order of the effective cyclotron energy of composite fermions , and therefore many different spin polarization occur . )
we note that some of the states were written down in an early work of halperin on multicomponent wave functions for the fqhe @xcite .
the fqhe problem in graphene differs from that in gaas in two respects .
first , in graphene , each electron has four - components , because of two spin projections and two valleys , producing an approximate su(4 ) symmetry when the zeeman energy and the valley splittings are negligible .
second , the linear dispersion leads to an interaction that is in general different from that in gaas @xcite .
our previous theoretical work on fqhe in graphene @xcite has shown that composite fermions are formed in both @xmath2 and @xmath4 lls .
the fqhe is thus still predicted to occur at @xmath19 , where @xmath20 is the fractional part of the filling factor in ll @xmath21 , but , in general , many more kinds of incompressible states are now available at each fraction , depending on whether the system is in the su(2 ) limit or su(4 ) .
the actual ground state is determined by a competition between the cf kinetic energy , which favors occupation of the lowest available @xmath15 levels , the cf exchange energy , which favors occupation of @xmath15 levels with the same su(4 ) index , and the zeeman and valley splittings @xcite .
some of the observations on fqhe in graphene are consistent with theoretical expectations : ( i ) the observed fractions have the form @xmath22 as measured relative to a filled landau level .
( ii ) the strength of fqhe in the @xmath4 ll is comparable to that in the @xmath23 ll , to be contrasted with gaas systems @xcite , where fqhe in the second ll is much weaker than that in the lowest ll .
( iii ) as predicted in @xcite , no fqhe has been observed at even denominator fractions in the @xmath4 ll .
some other observations are unexpected , however .
noteworthy is the absence of fractions such as @xmath14 and @xmath24 , while @xmath25 and @xmath26 are strong .
also puzzling is the appearance of fqhe at 1/3 ; one would have expected @xmath14 to be stronger given that its reference state at @xmath27 does not require any spontaneous symmetry breaking .
in contrast , in the @xmath4 ll , @xmath28 and @xmath29 are both observed strongly .
these fractions do not appear , at first sight , to be consistent with either su(2 ) or su(4 ) symmetry , thus posing a well defined puzzle whose resolution will shed important light on the nature of the fqhe in graphene .
our aim in this paper is to determine the stability of various states by evaluating the gaps using the cf theory .
the gaps in the su(2 ) limit were evaluated previously for states of the form @xmath30 for the lowest ll of gaas ( these results carry over to the @xmath2 landau level of graphene @xcite ) but not for the @xmath31 landau level of graphene .
in addition , gaps for the reverse - flux attached states at @xmath32 have not been calculated for technical reasons .
( lowest landau level projection , which is needed to obtain states appropriate for the high field limit , is prohibitively expensive for large systems for the latter class of states . )
however , the reverse - flux attached states are relevant to the graphene problem because the observed @xmath33 fqhe state is of that form @xcite .
we estimate below the gaps for the states relevant to the recent experiments .
in addition , we obtain the collective mode dispersions for various graphene fqhe states in both the su(2 ) and su(4 ) limits .
the issue of the @xmath34 state in graphene has been addressed recently in exact diagonalization studies @xcite .
another approach for constructing fqhe states in the su(4 ) limit can be found in @xcite .
a chern - simons treatment of composite fermions in the su(4 ) limit is given in @xcite .
our paper is organized as follows . following necessary definitions in section [ model ] , we study the excitation structure of the su(2 ) limit in section [ su2limit ] and the su(4 ) limit in section [ su4limit ] .
we discuss our results in light of the recent experiments @xcite .
our conclusions are presented in section [ conclusion ] .
for electrons confined to a landau level , the kinetic energy is quenched , and we need to consider the hamiltonian that is diagonal in sublattice components ( suppressed in our notation ) , @xmath35 here @xmath36 is the dielectric constant of the background material , @xmath37 and @xmath38 are pauli matrices related to spin and dirac valley indices , and @xmath39 and @xmath40 are zeeman and valley splittings . for @xmath41 and @xmath42 ,
this hamitonian possesses an su(4 ) symmetry , with @xmath43 ( or any appropriate linear combinations thereof ) defining the weights in su(4 ) multiplets .
composite fermions acquire additional labels ( quantum numbers ) @xmath44 , which come from a basis chosen in the fundamental representation of su(4 ) , and can be chosen conveniently to be @xmath45@xmath45 , @xmath45@xmath46 , @xmath46@xmath45 and @xmath46@xmath46 where the first arrow refers to valley and the second to spin .
let @xmath47 denote the incompressible ground state in which @xmath48 @xmath15 levels with label @xmath44 are fully occupied .
which of these states is the ground state depends @xcite on the values of @xmath39 and @xmath40 .
a coupling between the spin and valley degrees of freedom and ll mixing are neglected in what follows .
we neglect the anisotropy of the interaction due to underlying sublattice structure of the basis states , because such effects are small , being proportional to the ratio of the lattice constant @xmath49 to the magnetic length @xmath50 .
the study of the residual lattice effects is beyond the scope of this paper .
another important assumption of our model is the neglect of disorder .
this is a standard assumption in most quantitative studies of fqhe , because it is not known how to include the effect of disorder in a reliable manner .
it is possible that disorder might be affecting different fqhe states differently , thus affecting their relative stability . in the absence of a reliable quantitative treatment of disorder , while comparing our results with experiment we will make the simplest assumption that while the disorder causes a significant reduction of the excitation gap , it does not change the order of stability of various fqhe states .
strictly speaking , our results below ought to be considered as a prediction for future experiments in the limit of low disorder .
the kinetic term still determines the coulomb matrix elements via the single - particle orbitals when the landau level is fixed . for the @xmath2
ll the interelectron interaction is same as in gaas , because in this ll the single - particle states are identical for carriers with linear and quadratic dispersions . for @xmath51 , however , the two - component character of the single - particle states of massless dirac fermions produces different interaction pseudopotentials than in gaas . following standard practice ,
@xcite we use a lowest ll basis with an effective interaction given in @xcite , which reproduces the @xmath4 graphene pseudopotentials .
we employ the standard spherical geometry for our calculations @xcite , in which electrons move on the surface of a sphere and a radial magnetic field is produced by a magnetic monopole of strength @xmath52 ( integer or half - integer ) at the center , producing a magnetic flux @xmath53 through the surface . the general form of the cf wave functions in the spherical geometry is given by @xmath54 here , @xmath55 and @xmath56 .
@xmath57 is the wave function for interacting electrons at @xmath58 and @xmath59 is the wave function for noninteracting electrons at filling factor @xmath17 , with @xmath16 . in the spherical geometry
, @xmath60 is constructed at monopole strength @xmath61 , producing @xmath62 at monopole strength @xmath63 .
@xmath64 projects the state into the lowest ll @xcite .
the orbital part of the wave function has the same form as above even for the su(2 ) and su(4 ) generalizations , which we discuss below .
the electron - electron interaction is conveniently described in terms of pseudopotentials @xcite @xmath65 , which give the energy of two electrons in relative angular momentum @xmath17 .
as the orbitals in the lowest ll in graphene are identical to those in the conventional two - dimensional electron gas , the pseudopotentials are identical too .
the problem of interacting electrons in the @xmath31 ll of graphene can be mapped into a problem of interacting electrons in the lowest ll with the effective pseudopotentials @xcite @xmath66 where @xmath67 are laguerre polynomials .
the monte carlo evaluation of the energy of variational wave functions , on the other hand , requires a real - space interaction . to simplify calculation , we construct the wave functions in the lowest ll , and use an effective real - space potential corresponding to the effective pseudopotentials @xmath68 in the lowest ll basis .
following a well - tested procedure @xcite , we use the form @xmath69 and fit the coefficients @xmath70 to reproduce the first @xmath71 pseudopotentials @xmath68 of equation ( [ effective ] ) .
the coefficients @xmath70 obtained for @xmath72 are given in @xcite .
to weigh the relative importance of the relevant parameters , it is appropriate to consider the ratio of the zeeman splitting to some relevant energy scale in the problem . we consider @xmath73}$ ] , where the energy in the denominator is the fermi energy of the single component cf fermi sea and @xmath74 is the land factor for the background semiconductor material @xcite .
( the quantity @xmath75 is the magnetic length .
we could alternatively have chosen the gap of a prominent fqhe state . )
the fqhe states in gaas are typically fully spin polarized for @xmath76 t , i.e. , for ratio @xmath770.5 @xcite .
for graphene , this ratio is 1.8 ( 0.6 ) at @xmath78 t ( 4 t ) , suggesting that the spin degree of freedom may be frozen in the experiments of @xcite , which are performed at rather large @xmath79 .
the valley symmetry breaking in graphene is on the order of @xmath80 , where @xmath49 is the lattice spacing @xcite . at @xmath81 and 36 t ,
this ratio is @xmath82 0.02 and 0.06 , respectively .
while this issue requires further investigation , it can not be ruled out that the spin is fully polarized but the valley symmetry may not be broken @xcite .
we therefore first consider the situation where the spin degree of freedom is frozen but valley symmetry is intact .
( the analysis of the su(4 ) limit is given in the next section . ) in the su(2 ) limit , the orbital part of the wave functions at the maximal weight states @xcite of an su(2 ) multiplet at monopole strength @xmath52 are given by equation ( [ cfwf0 ] ) , with @xmath83 with @xmath84 and the superscript denotes the electron species . in other words ,
the state consists of @xmath85 and @xmath86 landau bands of up and down pseudospins , respectively occupied .
( here , pseudospin refers to the valley su(2 ) symmetry . ) one can show that this wave function satisfies fock s cyclic condition @xcite , i.e. , it is annihilated by any attempt at further antisymmetrization ; this ensures that it is an eigenstate of the total pseudospin operator .
the energies of variously valley - polarized states can be evaluated using standard methods , from which the phase diagram as a function of the valley zeeman energy can be constructed @xcite .
the valley degree of freedom behaves very much like the spin degree of freedom .
in fact , the predictions of the cf theory for two component systems have been verified semiquantitatively @xcite for another system with two valleys , namely alas quantum wells , where the valley splitting ( analogous to the zeeman energy ) can be controlled by application of in - plane symmetry breaking strain .
( a ) the extrapolation of the energy of the fully polarized and the su(2 ) singlet state at @xmath87 in the @xmath4 landau level of graphene to the thermodynamic limit .
@xmath88 is the number of particles .
only the leftmost four points , corresponding to @xmath89 , are included in the fit because smaller systems show finite size effects .
( b ) the polarization energy at @xmath87 .
the scaling is linear for @xmath90 . ] with spin frozen , 1/3 is valley - polarized , whereas 2/3 and 2/5 are valley singlets .
( these states are related by particle - hole symmetry to 5/3 , 8/5 , and 4/3 . )
that @xmath87 and @xmath33 are singlets in the @xmath2 landau level follows from the analogous result established in conventional two - dimensional electron systems such as gaas @xcite . for the @xmath87 in the @xmath4 ll of graphene
the energies of the 2/5 fqhe states in the thermodynamic limits can be obtained using the cf wave functions , and are given by @xmath91 for the singlet and @xmath92 for the polarized ground state energies , as shown in fig .
[ diff](a ) . the polarization energy , i.e.
, the energy difference between the polarized and the su(2 ) singlet states , has been estimated to be @xmath93 in the thermodynamic limit as shown in fig .
[ diff](b ) .
this confirms that , to the extent the wave functions are accurate , the energy ordering of states predicted by the cf theory remains same in the @xmath4 ll as in the @xmath2 ll .
we note that shibata and nomura @xcite conclude , based on a density matrix renormalization group calculation , that the ground state here is _ polarized _ even in the absence of a symmetry breaking field ; this conclusion , different from ours , is based on a study of smaller systems and without extrapolation to the thermodynamic limit .
we assume in what follows that the valley splitting is small but large enough to suppress the cf valley - skyrmions ( cf skyrmions are rapidly suppressed by the zeeman / valley splitting @xcite ) , and the relevant excitation is a particle - hole pair of composite fermions .
if composite fermions were completely noninteracting , the 1/3 gap would be the smallest , as it does not involve cf cyclotron energy .
the composite fermions , however , do feel a residual interaction , which necessitates a quantitative evaluation of the gaps , which can be accomplished using the standard methods of the cf theory @xcite .
the energy gaps of the lowest charged excitations shown in table [ gaps1 ] have several unexpected features @xcite : the 2/3 and 4/3 states in the @xmath31 ll are the strongest , followed by 2/3 and 4/3 in the @xmath2 ll and 1/3 and 5/3 in the @xmath31 ll .
next most stable are 1/3 and 5/3 in the @xmath2 ll .
these calculations are consistent with the observations that 2/3 and 4/3 are the strongest in the @xmath2 ll ( producing fqhe at @xmath25 and @xmath26 ) , both 1/3 and 2/3 are strong in the @xmath31 ll ( resulting in fqhe at @xmath28 , @xmath29 , @xmath94 , @xmath95 ) , and 2/5 and 8/5 are not yet observed .
.[gaps1 ] the smallest gaps for the creation of a far separated particle - hole pair of composite fermions .
the 1/3 ground state is fully spin and valley - polarized , whereas the 2/3 and 2/5 states are spin polarized but valley singlets .
neglecting ll mixing , the gaps at @xmath58 and @xmath96 are equal .
all energies are quoted in units of @xmath97 .
the 2/3 gaps come from a combination of exact diagonalization and cf theory ; the rest comes from cf theory only . [ cols="^,^,^",options="header " , ] while the observed order of stability is consistent with theory , the actual values of gaps are not .
the largest observed gap @xcite of @xmath98 k at @xmath99 and @xmath100 t is much smaller than the theoretical estimate @xmath101 k ( for @xmath102 ) .
the significant discrepancy is no doubt in large part due to disorder , and suggests that improvements in sample quality should reveal much further fqhe structure .
we next turn to the su(4 ) limit , achieved when the valley and the spin states are all approximately degenerate and may be applicable to experiments at very low @xmath79 . in this limit ,
if composite fermions were completely non - interacting , only fqhe states of the form @xmath22 with @xmath103 would be observable , but as seen above , the residual interaction between composite fermions results in a spontaneous symmetry breaking , producing fqhe at all of these fractions @xcite .
we now show that new goldstone modes become available , which are an su(4 ) generalization of the spin waves of composite fermions considered previously in the context of gaas fqhe @xcite .
( similar mechanism has been considered for qhe at @xmath104 integer fillings in graphene @xcite and in bilayer systems @xcite . ) as a result of the presence of these modes , fractions such as 2/5 and 3/7 are expected to be much weaker than their su(2 ) counterparts . in the su(4 ) generalization @xcite of cf theory @xcite , the orbital part of the ground state wave function at @xmath16 for correlated electrons in the maximal weight states @xcite of an su(4 ) multiplet at monopole strength @xmath52 is still given by equation [ cfwf0 ] , but with @xmath105 where @xmath106 is the slater determinant wave function of @xmath107 electrons of the @xmath108 species at monopole strength @xmath109 , with @xmath110 and @xmath111 .
with @xmath112 , it is easy to see that @xmath62 satisfies fock s cyclic condition @xcite , i.e. , it is annihilated by any attempt to antisymmetrize an electron @xmath113 of type @xmath114 ( @xmath115 ) with respect to the electrons of type @xmath116 , @xmath117 where @xmath118 permutes indices @xmath119 and @xmath113 . for the cf - particle hole excitation , one must in general choose @xmath120 to be an appropriate linear combination of slater determinant states to ensure the fock condition . in each su(4 ) multiplet , represented by a young tableau ( yt ) @xmath121 $ ] , where @xmath122 is the length of the @xmath123-th row ( @xmath124 and the last @xmath122 s are not shown if zero ) , the maximal weight states can be constructed with the help of elementary group theory , and the eigenstates of the total orbital angular momentum @xmath125 are obtained by standard clebsch - gordan expansion .
we show the appropriate combinations pictorially for all single exciton excitations at 2/5 , 3/7 and 4/9 , along with their yt representations , in the subsequent figures .
the interaction energies of the explicit wave functions are evaluated by monte carlo integration . for cf particle - hole excitation
we identify their orbital angular momentum @xmath126 with their wave vector @xmath119 according to @xmath127 , which yields the dispersion for the excitations , with the @xmath128 limit giving the energy of a far separated cf particle - hole pair .
all states in a given multiplet are degenerate for @xmath129 ; for nonzero @xmath39 and @xmath40 the energy splittings are straightforwardly determined .
the cf ground state at @xmath130 has a single fully occupied @xmath2 @xmath15l .
as excitation to an unoccupied @xmath2 @xmath15l is already possible in su(2 ) systems , we expect no new physics at this fraction .
the @xmath131 state has two @xmath2 @xmath15ls occupied in the ground state ; assuming for definiteness that @xmath132 , the occupied @xmath15ls are @xmath45@xmath45 and @xmath45@xmath46 .
it has two kinds of excitations depicted in fig .
[ twofifth ] : ( i ) the highly degenerate su(4 ) spin wave , in which a cf is excited sideways " to an unoccupied @xmath2 @xmath15l , as shown in fig .
[ twofifth](a ) .
this mode , represented by the yt @xmath133 $ ] , is available only because of the su(4 ) symmetry .
( ii ) excitations within the su(2 ) subspace of the ground state , in which a cf is raised up " to the second @xmath15l either flipping or preserving its valley index , represented , respectively , by yts @xmath134 $ ] and @xmath135 $ ] .
the four cases give rise to a valley triplet ( fig .
[ twofifth](b ) ) and a valley singlet ( fig .
[ twofifth](c ) ) , which are both gapped at all wavelengths .
the triplet is split by the @xmath40 and destabilizes the fqhe state when it reaches the `` valley - roton '' gap @xcite of @xmath136 .
the energy of the `` singlet - roton '' gap in mode ( c ) is unaffected by valley or zeeman splittings .
transitions that raise a cf to the first excited level of an otherwise empty band are ignored , as these involve a loss of exchange energy and a kinetic energy cost at the same time and are likely to represent higher excitations .
[ twofifth ] shows that the smallest gap to creating a far - separated pair of charged excitations , which is the one relevant for transport experiments , corresponds to a cf exciton in mode ( a ) .
because this energy is approximately half the gap in gaas for either fully spin polarized or spin singlet 2/5 state @xcite , the new goldstone mode thus results in a substantial weakening of the 2/5 state in the su(4 ) limit . from the su(4 ) spin wave dispersion at small @xmath119 we estimate the energy of the su(4 ) skyrmion , using the methods of @xcite , to be 0.030(6 ) and 0.037(9 ) in the @xmath2 and @xmath4 landau levels ; this has a higher energy than the other excitations , so will not be relevant to either transport or spin polarization . at @xmath137 there exist
sideways " excitations ( fig . [ threeseventh](a ) ) as well as those that stay within the su(3 ) sector of the ground state ( fig . [ threeseventh ] ( b ) and ( c ) ) .
the nine possible excitons within the su(3 ) sector can be reduced to a @xmath138 $ ] octet of su(3 ) ( see illustration above fig . [ threeseventh](b ) ) and an su(3 ) singlet .
the latter su(3 ) multiplets are embedded in @xmath139 $ ] and @xmath140 $ ] su(4 ) multiplets , respectively , and are all gapped .
the su(4 ) spin wave mode of fig .
[ threeseventh](a ) is not gapped in the @xmath141 limit and has the smallest energy at all wave vectors . the rather small gap to creating a far separated cf particle - hole pair in mode ( a ) makes the 3/7 state very delicate in the su(4 ) limit .
the stiffness of the su(4 ) spin wave in ( a ) could not be determined with sufficient accuracy , precluding an estimate of the su(4 ) skyrmion energy at 3/7 .
weak symmetry breaking fields @xmath39 and @xmath40 split the multiplet and select a state of reduced symmetry out of the su(2)@xmath142su(2 ) decomposition @xcite of the su(4 ) multiplet for both the ground state and the exciton . to illustrate we consider @xmath87 with @xmath132 .
the spin and pseudospin ( valley ) quantum numbers are @xmath143 for the ground state and @xmath144 , @xmath145 and @xmath146 , respectively , for modes ( a ) to ( c ) .
thus the fourfold degenerate mode ( a ) is raised by @xmath39 and split by @xmath147 ; mode ( b ) is split similarly but it is not shifted ; mode ( c ) is neither split nor shifted . for each mode as well as the ground state
there are other copies displaced to higher energies in multiples of @xmath39 , which in turn must merge into the continuum of other excitations ; the structure of these higher copies follows from the su(4 ) to su(2)@xmath142su(2 ) decomposition @xcite .
excitations of the su(4 ) singlet @xmath148 fqhe in the @xmath149 landau levels .
the weight structure of the 15-fold degenerate mode ( a ) is shown .
notice the roton minimum in mode ( a ) , which is destabilizing if @xmath150 .
3@xmath142 denotes weight multiplicity .
roton and transport gaps in mode ( a ) are indicated .
, title="fig : " ] excitations of the su(4 ) singlet @xmath148 fqhe in the @xmath149 landau levels .
the weight structure of the 15-fold degenerate mode ( a ) is shown .
notice the roton minimum in mode ( a ) , which is destabilizing if @xmath150 .
3@xmath142 denotes weight multiplicity .
roton and transport gaps in mode ( a ) are indicated .
, title="fig : " ] we have also studied @xmath151 , where there is no spin wave but the sixteen possible excitations into the second @xmath15l decouple into an su(4 ) singlet ( b ) and the fifteen - fold degenerate mode ( a ) labeled by the yt @xmath152 $ ] , see fig . [ fourninth ] .
the large @xmath119 limits of lowest energy are 0.021(1 ) and 0.026(2 ) for the @xmath2 and @xmath31 lls . as a function of @xmath39 or @xmath40
there are several possible transitions into cf states with other spin and valley quantum numbers @xcite .
the roton and transport gaps in mode ( b ) could not be obtained with sufficient accuracy in the theormodynamic limit ; but in any finite systems they are not smaller than in mode ( a ) .
it is uncertain if mode ( b ) truly develops a roton gap .
rotons in mode ( a ) destabilize the state at sufficiently high zeeman energy .
to conclude , we have argued that the spin degree of freedom may be frozen at high fields in graphene , and that many features of the fqhe in graphene at high magnetic field are consistent with su(2 ) cf theory . in particular ,
quantitative calculations are consistent with the observed order of stability of states in @xmath2 and @xmath4 lls , although the observed gaps are much smaller than the theoretical values .
we have further shown that in the su(4 ) limit , the fqhe states at @xmath131 and @xmath153 have a greatly reduced charge gap due to new su(4 ) spin wave mode , which implies more stringent conditions for their observation .
we are thankful to eva andrei , philip kim , cory dean and andrea young for stimulating discussions .
j.k.j . was supported in part by the nsf under grant no .
dmr-1005536 . c. t. was supported by science ,
innovative research teams , srop-4.2.2/08/1/2008 - 0011 and the bolyai fellowship of the hungarian academy sciences .
the authors acknowledge research computing and cyberinfrastructure , a unit of information technology services at the pennsylvania state university , and the national information infrastructure development program ( hungary ) for providing high - performance computing resources and services .
we thank lszl udvardi for assistance with the high - performance cluster at the budapest university of technology and economics . c. r. dean , a. f. young , p. cadden - zimansky , l. wang , h. ren , k. watanabe , t. taniguchi , p. kim , j. hone , and k. l. shepard , nat .
* 7 * , 693 ( 2011 ) ; f. ghahari , y. zhao , p. cadden - zimansky , k. bolotin , and p. kim , phys . rev . lett . * 106 * , 046801 ( 2011 ) r. r. du _
et al _ , phys .
lett . * 75 * , 3926 ( 1995 ) ; i. v. kukushkin , k. v. klitzing and k. eberl , _ ibid .
_ * 82 * , 3665 ( 1999 ) ; s. melinte _
et al _ , _ ibid . _ * 84 * , 354 ( 2000 ) ; n. freytag _ et al _ , _ ibid . _
* 89 * , 246804 ( 2002 ) .
we note that this issue deserves further investigation .
recent experiments suggest that the valley symmetry breaking is stronger than the spin splitting for the integral quantum hall plateaus .
see , for example , y. j. song , a. f. otte , y. kuk , y. hu , d. b. torrance , p. n. first , w. a. de heer , hongki min , s. adam , m. d. stiles , a. h. macdonald and j. a. stroscio , nature * 467 * , 185 ( 2010 ) ; r.k .
kamilla , x.g .
wu , and j.k .
jain , solid state commun .
* 99 * , 289 ( 1996 ) ; a. wjs and j. j. quinn , phys .
b * 66 * , 045323 ( 2002 ) ; d. r. leadley _ et al _ , phys . rev . lett .
* 79 * , 4246 ( 1997 ) ; a. f. dethlefsen , r. j. haug , k. vborn , o. ertk , and a. wjs , phys .
b * 74 * , 195324 ( 2006 ) . | because of the spin and dirac - valley degrees of freedom , graphene allows the observation of one- , two- or four - component fractional quantum hall effect in different parameter regions .
we address the stability of various states in the su(2 ) and su(4 ) limits . in the su(4 ) limit , we predict that new low - energy goldstone modes determine the stability of the fractional quantum hall states at 2/5 , 3/7 etc . ;
su(4 ) skyrmions are not found to be relevant for the low - energy physics .
these results are discussed in light of experiments . |
the search for quantum spin liquid states in frustrated magnets can be traced back at least as far as the early suggestion of a resonating valence bond state in the triangular lattice heisenberg model@xcite .
almost 15 years later , the suggestion of such a state in the undoped high-@xmath5 cuprates@xcite set off an explosion of interest in two - dimensional spin liquids ( i.e. mott insulators at half - filling with _ no _ broken symmetries ) .
frustrated heisenberg models on the square , triangular and kagom lattices have all received significant attention as candidate systems for quantum disordered ground states . while there has been comparatively little theoretical work on quantum spin liquids in three - dimensional frustrated magnets , materials with magnetic ions on the pyrochlore lattice ( fig .
[ fig : pyro ] ) may be good candidates for spin liquids and other exotic states .
to give one example , recent neutron scattering experiments on @xmath6 , a @xmath7 pyrochlore heisenberg antiferromagnet , suggest a nontrivial disordered state@xcite above a transition to nel order accompanied by a lattice distortion@xcite at 12.5 k. meanwhile , much work has been devoted to understanding the properties of possible spin liquid states , independent of their existence in particular microscopic models .
most of the proposed spin liquid states support deconfined @xmath8 spinons ; such states are fractionalized in that some of the elementary excitations carry quantum numbers that are fractions of those allowed in a finite - size system .
fractionalized states can be precisely characterized by their _
_ topological order__@xcite , which in the simplest scenario is associated with the topological sectors of an emergent , deconfining @xmath9 gauge field@xcite .
the `` vison , '' a gapped vortex - like excitation that carries the @xmath9 flux@xcite , must also be present .
while topological order does not require a liquid ground state and can coexist with conventional long - range order , we believe it is probably common in spin liquids and hence in some of the nearby ordered states . very recently , many of these ideas have been put on firmer ground by the emergence of several microscopic models supporting stable fractionalized phases in two and three dimensions @xcite .
despite these recent theoretical successes , an unambiguous experimental realization of these ideas is still lacking . indeed , spin liquid states seem rather rare ; is topological order rare as well ? for @xmath9-fractionalized states
this question is difficult to answer , because the gapped visons have no effect on easily measurable low - energy properties .
clever proposals have been made@xcite and carried out@xcite to directly detect topological order in the cuprates ( with negative results thus far ) , but these experiments are difficult and rely on properties of the phases proximate to a topologically ordered state . @xmath9
topological order is difficult enough to observe that it is impossible to say at present how rare or common it is .
fortunately it may be possible to shed new light on the experimental situation . in this paper , we present two models of _ three - dimensional _ @xmath8 frustrated magnets , one on the pyrochlore lattice , the other on a related network of corner - sharing octahedra ( the links of the cubic lattice , as shown in fig .
[ fig : cubic - octahedra ] ) . both of these models exhibit a novel fractionalized phase , the @xmath1 spin liquid .
this state has an emergent @xmath1 gauge structure that gives rise to several remarkable properties : there is a _
`` artificial photon '' excitation , a gapped spinon carrying `` electric '' gauge charge , a gapped `` magnetic '' monopole , an emergent @xmath10 `` coulomb '' potential between pairs of spinons and monopoles , and novel @xmath1 topological order .
if this phase exists in a real material , the gapless photon should have important implications for low - energy thermodynamics , transport and spectroscopy ; therefore @xmath1 fractionalization may be easier to find in experiments .
such states , thus far realized in large-@xmath11 spin models@xcite and bosonic hubbard - type models@xcite , arise as the deconfined or coulomb phase of compact @xmath1 lattice gauge theory .
while most work on spin liquids has focused on @xmath12 , motivated by the cuprates and the conventional wisdom that quantum fluctuations are more effective at destroying long range order in low dimensions , the @xmath1 spin liquid _ only _ occurs in @xmath13 ; for @xmath12 the coulomb phase of compact @xmath1 gauge theory ( with gapped matter ) is always unstable due to instanton effects@xcite .
both models are of intrinsic interest as examples of tractable but nontrivial frustrated magnets .
the pyrochlore model is particularly appealing due to its simplicity : its derivation begins with the nearest - neighbor @xmath8 heisenberg antiferromagnet .
taking the limit of large easy - axis exchange anisotropy @xmath14 simplifies the problem by breaking the spectrum into extensively degenerate manifolds with large separations of @xmath15 .
it is then possible to write an effective hamiltonian describing the splitting of the low - energy manifold , using standard techniques of degenerate perturbation theory in @xmath16 .
this effective hamiltonian _ has _ a @xmath1 gauge structure , which forms the foundation for our subsequent analysis@xcite .
another point of view , equivalent at the level of perturbation theory but perhaps with broader implications in more general scenarios , is that the low - energy sector of the model is _ unitarily equivalent _ to a @xmath1 gauge theory .
it is not obvious how to treat the resulting model analytically , but upon addition of an extra six - site interaction term it can be tuned to a soluble point where it is possible to write an exact ground state wavefunction with no further approximations .
the models can be reinterpreted as quantum dimer models ( qdms ) , and the extra term as the analog of the rokhsar - kivelson ( rk ) potential in the square lattice qdm@xcite . as will be explained in detail below ,
the properties of the soluble point allow us to locate the @xmath1 spin liquid adjacent to it .
since this state is stable to _ all _ zero - temperature perturbations , it persists over a finite extent of the phase diagram ( fig .
[ fig : phasediag ] ) .
furthermore , stability to large but finite @xmath17 implies that the @xmath1 gauge structure persists in the absence of microscopic local symmetries and is truly emergent . on the purely theoretical side , we believe these models give the first examples of @xmath1 gauge theories that have a deconfining phase even in the limit of infinitely strong bare coupling . the first such @xmath9 gauge theory was discovered only recently by moessner and sondhi@xcite .
is the relative strength of the rokhsar - kivelson potential and the xy ring exchange that obtains in the easy - axis limit of the heisenberg model .
the soluble point is located at @xmath18 , which is a special deconfined point of the adjacent @xmath1 spin liquid . just to the right of the soluble point the models go into an ising ordered state . sufficiently far to the left we expect ising order , while at intermediate values of @xmath19 states with broken translation symmetry but no magnetic order are also possible . immediately to the left of the soluble point , the @xmath1 spin
liquid exists over a finite ( but unknown ) extent of the phase diagram.,width=316 ] the effective theory of the @xmath1 spin liquid and the soluble rk point is simply gaussian quantum electrodynamics ( qed ) . at the rk point , which is itself a special deconfined limit of the generic phase , the `` electric stiffness , '' or coefficient of @xmath20 in the hamiltonian , vanishes .
this is a higher - dimensional generalization of the effective picture of the square lattice qdm in terms of a coarse - grained height field @xcite .
the @xmath1 spin liquid has power - law correlations with nontrivial angular dependence , novel @xmath1 topological order , and supports gapped @xmath3 spinons , a gapped topological point defect ( the `` magnetic '' monopole ) and a gapless @xmath4 collective mode corresponding to the photon of the gauge theory .
the latter excitation makes an additive @xmath21 contribution to the low - temperature specific heat , and should affect various other low - energy properties of @xmath1-fractionalized phases ( either the @xmath1 spin liquid , or phases with coexisting conventional and topological order ) .
if such a phase exists in a real material , we speculate that it may be possible to probe `` photons '' with photons via raman scattering . we begin sec .
[ sec : models ] with a derivation of the pyrochlore model starting from the heisenberg antiferromagnet . in sec .
[ sec : cubic - model ] the cubic ( or corner - sharing octahedra ) model is discussed . the remainder of sec .
[ sec : models ] is concerned with demonstrating the equivalence of the spin models to frustrated compact @xmath1 gauge theories , and developing a useful lattice version of electric - magnetic duality .
beginning from the dual description , sec .
[ sec : effective - theory ] develops the effective description of the @xmath1 spin liquid and the soluble point in terms of gaussian quantum electrodynamics .
corrections to effective action and to the scaling equalities between microscopic and effective degrees of freedom are discussed in sec .
[ sec : corrections ] .
[ sec : properties ] contains a discussion of the universal properties of the @xmath1 spin liquid , including its novel @xmath1 topological order . in sec .
[ sec : rkpoint ] we present our analysis of the soluble point ground state wavefunction , which gives strong support for the validity of our effective picture .
we conclude in sec . [
sec : discussion ] with a discussion of open issues , focusing on the challenging problems of understanding this physics in a broader range of models and looking for @xmath1-fractionalized phases in real materials .
, the fcc bravais lattice vector @xmath22 points in the direction given by looking from site @xmath23 to site @xmath24.,width=316 ] we begin with the nearest - neighbor @xmath8 heisenberg antiferromagnet on the pyrochlore lattice .
this structure is a three - dimensional network of corner - sharing tetrahedra ( fig .
[ fig : pyro ] ) .
it can be obtained by translating one `` up - pointing '' tetrahedron ( shown on the right of fig .
[ fig : pyro ] ) through the fcc bravais lattice vectors @xmath25 .
we choose @xmath26 , @xmath27 , and @xmath28 .
basis vectors for the reciprocal lattice are defined by @xmath29 , so that @xmath30 .
the four sites in each unit cell are distinguished by an index @xmath31 , as indicated in fig .
[ fig : pyro ] .
lattice sites are denoted either by single italic letters like @xmath24 , or by pairs @xmath32 when we wish to specify the position of a site within the unit cell . up to a constant
the hamiltonian can be written as a sum over tetrahedra : @xmath33 where @xmath34 is the total spin on the tetrahedron @xmath35 . following the analysis of a generalized kagom heisenberg antiferromagnet in ref .
[ ] , we introduce easy - axis exchange anisotropy : @xmath36 where @xmath14 .
this reduces the global @xmath37 invariance to @xmath38 .
we first consider the point @xmath39 , where @xmath40 reduces to a classical ising model , with ground states specified by @xmath41 on all tetrahedra .
it was argued by anderson@xcite that , almost identically to pauling s model for water ice@xcite , this ising model has an extensive ground state degeneracy ( i.e. finite @xmath42 entropy per site ) .
a small @xmath43 introduces quantum fluctuations and lifts the extensive degeneracy ; this splitting is encapsulated in an effective hamiltonian using standard techniques of perturbation theory .
the first - order contribution is easily seen to vanish .
we will need to go to third order , where we have the general expression : @xmath44}(1 - { \cal p } ) \text{.}\ ] ] here @xmath45 projects onto the orthogonal complement of the ground state manifold . to describe the processes contributing in eq .
( [ eq : generalpt ] ) , it is useful to work in the standard hardcore boson language for the spins , where @xmath46 corresponds to the presence / absence of a boson .
each term in @xmath47 hops bosons along nearest - neighbor bonds ; acting on a state in the low - energy manifold , each hop creates two tetrahedra with @xmath48 . at second order in @xmath47
, bosons can hop and then return along the same bond ( fig .
[ fig : pyro - pt]a ) .
this can always occur on 4 bonds in every tetrahedron , thus giving only a constant contribution to the energy . at third order
another constant contribution arises from single bosons ( or holes ) hopping around triangular faces ( fig .
[ fig : pyro - pt]b ) .
there is also a nontrivial _ ring exchange _ process acting on the hexagonal plaquettes ( see fig .
[ fig : pyro ] ) , where hexagons containing three evenly spaced bosons can be rotated as shown in figure [ fig : pyro - pt]c .
the resulting effective hamiltonian is @xmath49 where @xmath50 is the total number of tetrahedra , @xmath51 , and the sum is over hexagonal plaquettes .
the labelling of the spin operators inside the sum is given by moving around each hexagon in an arbitrary direction .
note that @xmath52 = 0 $ ] , as must be true for _ any _ effective hamiltonian acting in the low - energy manifold , whatever the form of @xmath47 .
we focus on the extreme easy axis limit described by @xmath53 , but note in passing that a finite but large @xmath17 would introduce small fluctuations out of the ground state manifold .
while these will not affect universal properties , they can matter for the short - distance correlation functions of some microscopic operators .
this can be understood formally by a more sophisticated execution of the perturbation theory in @xmath54 that accounts for splitting of the low - energy manifold and mixing of higher states on an equal footing@xcite .
the main result is that the problem at finite @xmath17 can be mapped , by a unitary transformation , order - by - order in @xmath54 onto a transformed hamiltonian acting only within the low - energy manifold where @xmath41 .
this mapping accounts for finite @xmath17 by generating nontrivial perturbative relations between physical and transformed operators ; however , we ignore these corrections for simplicity and use only the results of the standard degenerate perturbation theory described above .
it is possible , and will be convenient , to change the sign of the ring term by a similarity transformation . on any given site we can make the transformation @xmath55 and @xmath56 by making a @xmath57-rotation about the @xmath2-axis in spin space .
one transformation with the desired effect , consisting of @xmath57-rotations on a pattern of sites , is : @xmath58 where @xmath59 and @xmath60 .
after this transformation the hamiltonian takes the form @xmath61 where the constant terms have been dropped .
models similar to this one on the kagom@xcite , square@xcite , triangular@xcite other lattices@xcite , where xy ring exchange of spins or bosons is a dominant term , have recently been shown to exhibit a variety of unusual phases and critical behavior .
the physics of the pyrochlore ring exchange model should be accessible to quantum monte carlo studies ; while the original hamiltonian in eq .
( [ eq : pyro - easy - axis ] ) has a sign problem , @xmath62 does not .
@xmath63 can be reinterpreted as a quantum dimer model on the diamond lattice ( fig .
[ fig : diamond ] ) , with two dimers touching every site . to see this ,
observe that the centers of the pyrochlore tetrahedra form a diamond lattice .
each nearest - neighbor diamond link passes through exactly one pyrochlore site , so we can reinterpret the pyrochlore spins as diamond link variables .
the smallest closed loops in this lattice contain six links and correspond to the pyrochlore hexagons .
we say a dimer is present on a given bond if @xmath64 , or absent if @xmath65 .
@xmath41 becomes the constraint that every diamond _ site _ touches two dimers , and the ring exchange move is the most local dynamics preserving this constraint .
each term in @xmath63 acts on a `` flippable '' hexagon , one containing alternating full and empty bonds as in fig .
[ fig : diamond ] , by rotating the dimers around it .
non - flippable hexagons are annihilated .
as first realized by rokhsar and kivelson , dimer models generically have a point in their parameter space where it is possible to write down the exact ground state wavefunction@xcite . to reach this point in our model , we add the term @xmath66 , where @xmath67 is the number of flippable hexagons .
the rk point obtains for @xmath68 ( i.e. @xmath69 ) , and the ground state is an equal - weight superposition of all possible dimer coverings of the lattice that satisfy the constraint of two dimers touching every site . in the spin language , this wavefunction can be written as the projection of a transverse ferromagnet : @xmath70 ) , @xmath71 projects onto the @xmath41 manifold . for completeness
, we also express @xmath67 in terms of spin operators : @xmath72 , where @xmath73 gives unity acting on a flippable hexagon and zero otherwise .
one has : @xmath74 we will be interested in the properties of the generalized ring model @xmath75 in the vicinity of the soluble point . largely to simplify the geometry of the presentation , we introduce an alternate model that we find has many of the same properties as its pyrochlore analog .
the model is the qdm on the cubic lattice , with _ three _ dimers touching every site .
we consider only the most local dynamics , which rotates the configuration on square plaquettes with two dimers on opposite sides , and the corresponding rokhsar - kivelson potential that counts flippable squares .
reversing the mapping above , we can also think of this as a spin model with @xmath8 on the links of a cubic lattice , or , equivalently , on the sites of a lattice of corner - sharing octahedra with their centers at the cubic sites ( fig .
[ fig : cubic - octahedra ] ) .
the octahedra play the role of the pyrochlore tetrahedra , with the total spin on each @xmath76 .
we denote cubic sites by boldface letters like @xmath77 and identify the links by specifying either pairs of adjacent sites , or one site and the direction of the link .
for example , the link connecting a site @xmath77 with its nearest neighbor in the positive @xmath78-direction ( @xmath79 ) is denoted by @xmath80 or @xmath81 .
using this notation , we express the dimer kinetic term as a 4-site xy ring exchange for the spins : @xmath82 where the numbering in the first line runs around the perimeter of each square plaquette , and in the second line only one orientation of plaquette is shown explicitly .
as before , we will be interested in the vicinity of the soluble point of @xmath83 .
it is also interesting to note that , as for @xmath63 , the pure ring exchange model @xmath84 can be derived as the easy - axis limit of a heisenberg antiferromagnet . in this case one begins with spins on the network of corner - sharing octahedra with nearest - neighbor exchange , and an additional exchange of the same sign and magnitude between spins at opposite points of the same octahedron .
the hamiltonian can be written as a sum over octahedra and the analysis proceeds exactly as before .
both the cubic and pyrochlore models have an exact @xmath1 gauge invariance , as is generally the case in dimer models . in the pyrochlore model , this arises because of the local integer conserved quantity @xmath85 .
the local symmetry is generated by rotations about the @xmath2-axis in spin space on all the sites in a given tetrahedron : @xmath86 .
these generators commute with one another and the hamiltonian .
it is important to emphasize that this gauge symmetry is generated by _
physical _ transformations and has nothing to do with any redundancy in our description .
these statements also hold for the cubic model , with octahedra substituted for tetrahedra .
for now , we shall focus on the cubic model for ease of presentation .
all of the statements in this and the following section can be generalized simply to the pyrochlore case ; this is summarized in section [ sec : pyro - mappings ] below .
the gauge structure suggests we may gain insight by thinking about the model as a @xmath1 lattice gauge theory , an approach that proved helpful in understanding the square lattice qdm @xcite .
in fact , it is possible to formally rewrite @xmath84 as a pure gauge theory on the cubic lattice . to see this
, we find it convenient to soften the hardcore constraint on the bosons by explicitly introducing an on - site repulsion term in the hamiltonian .
we also go to quantum rotor variables @xmath87 and @xmath88 , with commutation relations @xmath89 = i
\delta_{ij } \delta_{{\bf r}{\bf r}'}$ ] .
( here and elsewhere we commit a standard abuse of notation and formally denote a constraint on the eigenvalues of an operator as a constraint on the operator itself . ) using the relations @xmath90 and @xmath91 , we have @xmath92 where the numbering inside the cosine proceeds around the perimeter of the given square .
this is a faithful representation of the spin model eq .
( [ eq : cubic - ring - model ] ) in the limit @xmath93 , which just imposes the hardcore constraint @xmath94 .
we now define an orientation on the cubic links , which we take to point out of the a sublattice and into the b sublattice .
this allows us to define oriented link variables by : @xmath95 here we take the plus / minus sign when @xmath77 lies in the a / b sublattice . since @xmath96 , these variables can be thought of as components of vector fields taken along the links of the lattice .
putting these definitions into the hamiltonian , we have @xmath97 the sum inside the cosine is taken in an oriented fashion around the links of the given square plaquette , and is thus a discrete line integral of @xmath98 along a smallest possible closed path .
this is a lattice version of the curl , so we define : @xmath99 eq . ( [ eq : cubic - gt ] ) is invariant under gauge transformations written in the usual form @xmath100
. therefore @xmath98 behaves like a vector potential , and since @xmath101 = i$ ] , @xmath102 plays the role of an electric field . since the vector potential is a @xmath103-periodic phase this is evidently a _ compact _ gauge theory .
the `` electric '' charge is given by the lattice divergence of @xmath102 : @xmath104 where the sum is over the sites adjacent to @xmath77 .
in the ground state there is no gauge charge since @xmath76 , and single gauge charges have a large gap of order @xmath17 .
suppose @xmath77 is a site in the a sublattice .
acting with @xmath105 creates two octahedra with @xmath106 , at @xmath77 and @xmath107 . in the gauge theory language ,
there is a positive electric charge at @xmath77 and a negative one at @xmath108 , where the sign of the charge comes from the orientation convention eq .
( [ eq : vector - vars - e ] ) .
we can now act with the infinite string operator : @xmath109,y } s^-_{[{\bf r } + ( n+1){\bf x } + n{\bf y}],x } \text{.}\ ] ] this hops the gauge charge originally at @xmath79 off to infinity , leaving an isolated @xmath110 octahedron at @xmath77 .
the two @xmath110 octahedra together carry @xmath111 , so the single remaining octahedron is evidently a @xmath3 _ spinon _ @xcite .
the spinons are single electric gauge charges and can propagate freely in a deconfining phase of the gauge theory ; such a state is therefore fractionalized .
we note that , because spinons can not hop from one sublattice to the other , there are in fact two spinon flavors . because the theory is compact , magnetic charge is also allowed .
define dual lattice sites by serif characters @xmath112 ; these are located at the centers of the cubic `` boxes '' of the direct lattice .
the links of the dual lattice are naturally associated with the square plaquettes of the direct lattice . using this correspondence
we define a magnetic field on the dual links by @xmath113 .
the sense of the lattice curl is taken counterclockwise looking from the point @xmath114 to @xmath115 .
the lattice divergence @xmath116 gives the magnetic charge inside the box at @xmath115 .
naively this divergence vanishes since each term @xmath98 occurs twice , with opposite signs , but this is not the case because @xmath117 is a periodic variable invariant under @xmath118 .
in fact we have @xmath119 for integer @xmath120 ; the magnetic charge is automatically quantized .
it is convenient to take @xmath121 , so that @xmath122 on each box .
it is also possible to have @xmath123 in our convention , but this measure zero point in the configuration space of magnetic fields should be ignored . while electric charges are locally conserved in the low - energy manifold , there is no such conservation law for magnetic charge and we expect @xmath124 even in the ground state .
while this means the ground state always has some local fluctuations of magnetic charge , it does _ not _ necessarily contain monopoles , which are smoothly - varying defect configurations unaffected by a small amount of coarse - graining . the model eq .
( [ eq : cubic - gt ] ) looks identical to the standard hamiltonian formulation of compact @xmath1 lattice gauge theory , but there is one difference of critical importance in the limit of interest @xmath125 .
here , the electric field takes on half - integer rather than integer values . in the case of integer electric fields ,
the vacuum in the large-@xmath126 limit is trivial : @xmath127 everywhere with small fluctuations . in our model
this limit enforces the nontrivial constraint @xmath128 .
this is another expression of the inherent frustration , so we refer to the model as a `` frustrated gauge theory . ''
this situation is essentially the same as that arising in the large-@xmath11 limit of bipartite @xmath129 antiferromagnets@xcite , and in the gauge theory description of the square lattice qdm@xcite .
there , one considers an integer - valued electric field in a background of static charge ; of course , our model can also be viewed this way by making a shift of the electric field . to gain insight into the possible phases of @xmath84 , we briefly review the properties of the standard unfrustrated gauge theory , which has the same hamiltonian as our model but an integer electric field@xcite .
there are two phases , separated by a transition at @xmath130 . for @xmath131 , the `` strong coupling '' side of the transition
, the model enters a confining phase smoothly connected to the trivial @xmath132 vacuum @xmath133 . in this phase
all excitations are gapped , and static sources of the electric field are confined by a linear potential .
for @xmath134 one enters the deconfining coulomb phase , so named because static gauge charges interact via a @xmath10 coulomb potential and are thus free to propagate . at low energies ,
the effective description of the coulomb phase is simply gaussian qed , so there is a gapless , linearly dispersing photon with two transverse polarizations . in the coulomb phase ,
magnetic monopoles are gapped and interact via a @xmath10 magnetic coulomb potential .
the monopoles incorporate the compact nature of the magnetic field , which is not important at low energies in the deconfined phase . in the confined phase
, however , the magnetic field fluctuates wildly , its periodicity is important , and the monopoles have proliferated and condensed .
this distinguishes the two phases : in the coulomb phase the monopole propagator decays exponentially , while in the confined phase it goes to a constant .
this distinction is more robust than the wilson loop , which fails to differentiate between the phases in the presence of matter fields .
returning to the frustrated gauge theory , we can infer that it also has a coulomb phase at small @xmath135 .
this should be so because in the deconfined phase the discrete nature of the electric field is unimportant , so the half - integer nature of @xmath102 will not play a role
. we will be interested in whether the coulomb phase survives in the opposite limit of strong coupling , perhaps stabilized by additional terms in the hamiltonian . in recent work on other models with xy ring exchange
, it has been useful to make a novel `` plaquette duality '' transformation@xcite . in the case of the frustrated gauge theory ,
plaquette duality is in fact identical to the more familiar electric - magnetic duality for @xmath1 gauge theories .
we shall make use of the dual cubic lattice defined in the preceeding section .
we define oriented variables on the dual links @xmath136 and @xmath137 , which we take to be canonically conjugate : @xmath138 = i$ ] on the same dual link , zero otherwise . as discussed above , @xmath139 is the magnetic field , defined by : @xmath140 where the sense of the curl is again taken counterclockwise looking from @xmath114 to @xmath115
. the conjugate variable will play the role of a `` dual vector potential , '' and is related to the electric field by : @xmath141 where the curl is taken around the _ dual _ square plaquette encircling the direct link joining @xmath77 to @xmath142 , and again the orientation is given by the right hand rule . here
@xmath143 is a static , divergenceless background field ; this is necessary for consistency with @xmath144 . with these definitions ,
the dual commutation relations are consistent with the original ones . in dual variables
the hamiltonian is @xmath145 where the first sum is over dual plaquettes .
the constraint of zero electric charge is now automatically satisfied , since @xmath146 .
however , magnetic charge can take on continuous values in the dual variables , and one should impose the constraint @xmath147 .
this constraint commutes with @xmath148 since it is invariant under the dual gauge transformations @xmath149 , where @xmath150 .
note that because magnetic charge is not locally conserved , we are not allowed to demand @xmath151 .
it is useful , and enlightening , to write down the euclidean action obtained by a trotter expansion in eigenstates of the dual vector potential . as usual , one begins with the partition function @xmath152 here @xmath153 is a projection operator imposing the quantization of magnetic charge : @xmath154 breaking the exponential in eq .
( [ eq : partition - fn ] ) into @xmath155 timeslices , and inserting @xmath45 once in every timeslice , one has : @xmath156 where the index @xmath157 .
the imaginary time component of @xmath158 comes from the poisson - resummed form of the projector eq .
( [ eq : magnetic - quantization - projector ] ) .
the spatial components enter as the eigenvalues of the states @xmath159 used to form the resolution of the identity at each timeslice .
following very similar manipulations to those in appendix a of ref .
( ) , one obtains the dual action : @xmath160 where @xmath161 .
the action eq .
( [ eq : cubic - dual - action ] ) is essentially a higher - dimensional generalization of the height model partition function arrived at by similar manipulations in the context of the square lattice qdm@xcite .
significantly , it differs in having a local rather than a global invariance , under spacetime - dependent dual gauge transformations of the form @xmath162 .
in fact , @xmath163 has the same structure as non - compact lattice qed , except for the discrete nature of the fields .
this encodes the important physics of the magnetic monopoles . in sec .
[ sec : effective - theory ] below , it will be useful to imagine softening the constraint of discreteness on the fields to interpolate between the dual partition function and an effective description of the coulomb phase . with the soft constraint ( implemented by the `` corrections '' in eq .
( [ eq : non - ginvt - corrections ] ) ) , the theory is identical to a more familiar dual representation of @xmath1 gauge theory consisting of a non - compact gauge field minimally coupled to scalar monopoles .
we now return to the pyrochlore ring model @xmath63 . in this case
the diamond lattice with sites at the centers of the tetrahedra ( discussed in sec .
[ sec : pyro - model ] ) plays the role the cubic lattice did for the cubic model . denoting _ diamond _ sites by boldface characters ,
we soften the hardcore constraint on the bosons and go to quantum rotor variables living on the diamond links : @xmath164 here the second sum is over the hexagonal loops of the diamond lattice ( fig .
[ fig : diamond ] ) , and the numbering inside the cosine proceeds around the perimeter of the given hexagon .
the diamond lattice is bipartite , so we define an orientation by declaring that links naturally point out of the `` up - pointing '' sites and into the `` down - pointing '' ones ( corresponding to up- and down - pointing tetrahedra , respectively ) .
we define an oriented electric field and vector potential exactly as in eqs .
( [ eq : vector - vars - e],[eq : vector - vars - a ] ) .
the hamiltonian then takes the form of the diamond lattice frustrated gauge theory : @xmath165 it is evident that the lattice curl now naturally lives on the hexagons of the diamond lattice .
again the electric charge has a simple interpretation in the spin language : @xmath166 tetrahedra with @xmath167 are now the @xmath3 spinons carrying unit gauge charge .
single spinons can be created by a string operator similar to eq .
( [ eq : string ] ) . for
small @xmath135 the model should again enter a deconfining phase where the spinons are free to propagate .
we define a dual lattice of plaquette variables by putting a site at the center of every pyrochlore hexagon .
this is also a pyrochlore lattice , and it will be useful to think of its sites as the links of a dual diamond lattice with sites labelled by serif characters @xmath115 .
each hexagon of the dual lattice encircles a link of the direct lattice , and vice versa . as before ,
magnetic charge lives on the dual lattice sites .
the dual variables , again with the commutator @xmath138 = i$ ] on the same link , are defined on the dual links by : @xmath168 with the sense of the lattice curls determined as in the cubic case .
here @xmath169 , with @xmath170 .
the hamiltonian takes the form @xmath171 one can derive an action in eigenstates of the dual vector potential as in the previous section , with the result : @xmath172 and @xmath173 the constraint of magnetic charge quantization @xmath174 enters as before , giving rise to the temporal dual vector potential fields in @xmath175 .
it is well - known that the low - energy description of the coulomb phase of compact @xmath1 gauge theory is non - compact qed with no matter fields . in our case it will be convenient to work in dual variables to formulate the effective theory , which more naturally allows the inclusion of magnetic charge fluctuations . depending on the purpose at hand , different formulations of the effective action will be useful . to fix notation ,
we show them all here . in each case
the partition function is of the form : @xmath176 \operatorname{exp}(-{\cal s}^0)\text{,}\ ] ] where @xmath177 is a real - valued field
. it will often be convenient to keep the full spatial lattice structure , in which case we can define @xmath177 in terms of the microscopic variables by the _
temporal _ coarse - graining : @xmath178_f - \frac{\alpha^0_{{\sf r}\mu}}{2}\text{.}\ ] ] the brackets @xmath179_f$ ] denote an average over high - frequency modes in imaginary time , and we have subtracted the time - independent , non - fluctuating background @xmath180 defined by : @xmath181 with @xmath182 . if ( as we always do in practice ) we restrict attention to spatially periodic @xmath183 with zero average electric flux in every direction , @xmath184 can also be taken periodic .
this subtraction simplifies the relation between the electric field and dual vector potential , since @xmath185 on a spacetime lattice , the action looks almost identical to @xmath186 in eq .
( [ eq : cubic - dual - action ] ) : @xmath187 the only difference from the microscopic model is that the fields are now continuous .
we will also have occasion to retain the spatial lattice structure but take the time - continuum limit , in which case we write @xmath188 the parameters here are related to those in eq .
( [ eq : discrete - cubic - effective - theory ] ) by @xmath189 and @xmath190 , and @xmath191 now has units of inverse time .
this action corresponds to the effective dual hamiltonian ( in the sector with no magnetic charge ) : @xmath192 which can be obtained simply by expanding the cosine in eq .
( [ eq : cubic - dual - hamiltonian ] ) .
the lattice effective actions for the pyrochlore model look identical to eqs .
( [ eq : discrete - cubic - effective - theory ] , [ eq : time - cont - cubic - effective - theory ] ) , where the only change necessary is the replacement of the sums over dual square plaquettes with sums over dual hexagons .
finally , in order to take the spatial continuum limit , we introduce a continuum 4-vector field @xmath193 , formally defined by the replacements @xmath194 here @xmath195 are the vectors connecting the site @xmath115 to its nearest neighbors , in either the cubic or diamond lattice , and @xmath196 is a microscopic length on the order of the lattice spacing .
navely it seems we have thrown away too much information in the pyrochlore case , since there are four sites per unit cell , but only three spatial components of the continuum vector field . however , there are also _ two _ gauge degrees of freedom per unit cell of the diamond lattice , corresponding to changing @xmath197 on the two different sublattices .
this leaves us with two transverse degrees of freedom , the same number as in the cubic model and the continuum theory .
the full spacetime continuum theory is @xmath198 where @xmath199 and @xmath200 ( @xmath196 is the microscopic length in eq .
[ eq : biga - defn ] ) .
the spatial / temporal components of @xmath201 have units of inverse length / time .
the form is identical to the familiar ( dual ) maxwell action for electromagnetism with photon velocity @xmath202 .
this action can be used to obtain long - wavelength properties of the coulomb phase of either microscopic model , but at the end of any calculation the allowed spatial components of all vector fields are determined by the lattice structure . to calculate using any of these effective actions , it is convenient to implement a gauge - fixing procedure .
the well - known manipulations of faddeev and popov@xcite tell us that we may add any function of the vector potential 4-divergence to the action .
the standard choice , in our nonstandard notation , is : @xmath203 if we choose @xmath204 , as usual the off - diagonal terms in the action are cancelled and we obtain the simple photon propagator : @xmath205 where the euclidean metric is defined by @xmath206 , @xmath207 and @xmath208 . while these expressions hold in the continuum , it is amusing to note that the obvious lattice regularization of eq .
( [ eq : fp - action ] ) also cancels all off - diagonal terms on the cubic lattice . working in the time - continuum action , we find the lattice propagator : @xmath209 where @xmath210 . on the diamond lattice ,
even more complicated faddeev - popov terms seem only to cancel some of the off - diagonal terms . finally , it is important to note that this procedure is only legitimate for calculating expectation values of operators invariant under _ continuous _ dual gauge transformations .
terms with only discrete dual gauge invariance ( see sec . [
sec : corrections ] ) must be handled more carefully ; this issue arises in calculating the monopole propagator in section [ sec : monopole - prop ] .
given the microscopic gauge structure of our models , it is reasonable to conjecture that the low - energy effective degrees of freedom are simply noncompact @xmath1 gauge fields .
the effective action should take the form of an expansion in the lowest - order terms involving the dual vector potential consistent with the symmetries .
generically this would take the form of eq .
( [ eq : dual - continuum - action ] ) ; however , as our analysis of the microscopic models takes advantage of the special properties of the rokhsar - kivelson point to conclude that the `` stiffness '' or @xmath211 term for the electric field vanishes there , we must include a higher term .
this leads us to propose that the rk point of both models is described by the effective action @xmath212 here @xmath213 is the coefficient of the new term , which ( when @xmath214 ) is marginal in the renormalization group sense and must be included .
while the theory remains gauge invariant , the photon dispersion now vanishes _ quadratically _ : @xmath215 .
the @xmath211 term is relevant , as can be seen by the usual power - counting procedure , so this action can not describe a stable phase .
a small positive @xmath211 will drive the system into the coulomb phase , while a small negative @xmath211 will result in an electric field `` crystal ; '' in the dimer language this is a state with long - range dimer order and no flippable plaquettes .
as we argue in section [ sec : rkpoint ] , precisely this picture describes the physics of the microscopic models near the rk point , with the relation @xmath216 .
a very similar story is known to apply to the @xmath217 square lattice qdm @xcite . in that case
, electric - magnetic duality leads to a height model partition function , which describes the fluctuations of a discrete field living on the square lattice plaquettes in 2 + 1 dimensions . just as above , softening
the constraint of discreteness naturally leads to the gaussian effective action : @xmath218 when @xmath219 we have a description of the rk point , which has power - law dimer - dimer correlations . for small negative @xmath220
the system goes into the staggered valence bond crystal state .
these have natural analogs in the three - dimensional case .
however , a small positive @xmath220 leads to a confining state with broken translation symmetry@xcite via an instability of the gaussian theory .
this occurs because eq .
( [ eq : height - effective ] ) with @xmath221 would describe the unstable coulomb phase of pure 2 + 1-dimensional @xmath1 gauge theory . in three dimensions , however , the analogous phase is stable and should exist adjacent to the rk point . to calculate with eq .
( [ eq : rk - effective - action ] ) we again carry out a gauge - fixing procedure .
we add the faddeev - popov term : @xmath222 although this does not cancel all off - diagonal terms , it leads to the relatively simple photon propagator : @xmath223 where @xmath224 \nonumber\\ { \cal m}_{i j } & = & \frac{1}{f(\omega_n,{\bf k } ) } \big [ k_i k_j ( k^2 - { \cal k}_c^2 m^2 ) \big ] \quad \text{(}i \neq j\text{)}\nonumber\text{,}\end{aligned}\ ] ] and @xmath225 with @xmath226 . while all gauge - invariant corrections to @xmath227 are irrelevant in the renormalization group sense , in order to understand the detailed realization of the @xmath1 spin liquid ( and the nearby rk point ) in any microscopic model it will generally be important to consider corrections to the effective action .
furthermore , we need to specify relations between the microscopic and effective degrees of freedom .
navely we could write @xmath228 , but any corrections allowed by symmetry will generically be present . for ease of presentation , we focus on the cubic model ; the lattice structure is unimportant for these results .
the symmetries of the microscopic model are listed in detail in appendix [ app : symmetries ] .
it is important to note that , because the background @xmath184 is not invariant under all lattice symmetries , the dual vector potential transforms with additional shifts under these operations .
furthermore , the dual effective degrees of freedom @xmath177 and @xmath229 transform exactly as their microscopic counterparts . rather than attempt a painstaking enumeration of all allowed corrections to @xmath230
, we discuss some representative examples .
the corrections naturally fall into two classes ; those invariant under _ continuous _ dual gauge transformations , and those not .
in hamiltonian language on the lattice , some typical terms in the first class are : @xmath231 terms involving discrete line integrals of both vector potentials around closed loops larger than single plaquettes are also allowed , as are terms containing the divergence of the fields . while these terms are irrelevant , they can have quantitative effects , presumably accessible in perturbation theory .
all such corrections are also irrelevant at the rk point , except the @xmath232 term , which contributes to the @xmath213 term in eq .
( [ eq : rk - effective - action ] ) .
more interesting are those terms lacking continuous dual gauge invariance .
here we only consider the single site terms ; in the spacetime lattice action these take the form : @xmath233 in the first term the @xmath184-dependence is necessary to compensate the shifts in @xmath177 under lattice symmetries . with appropriate choices of coefficients ,
when these corrections become large they have the effect of pinning the dual vector potential to take on discrete values .
this allows us to interpolate explicitly between the effective theory @xmath234 and the microscopic partiton function .
physically , the spatial @xmath235 part of @xmath236 is a _ magnetic charge hopping _ term .
its temporal companion is related to the discreteness of magnetic charge .
therefore , these corrections introduce magnetic charge fluctuations into the effective theory .
as should be expected when magnetic monopoles are gapped , when these fluctuations are small there is no associated instability .
formally , this should be the case because the correlation functions of @xmath237 are local in space and time .
for example , @xmath238 because continuous dual gauge invariance in @xmath230 makes each cosine into an independent random variable .
finally , we consider corrections to the scaling equalities between microscopic and effective degrees of freedom .
some representative contributions in the case of the dual vector potential are , in operator language : @xmath239 this should really be interpreted as an instruction for what effective theory operators to use in the calculation of gauge invariant expectation values .
these corrections are not important , and we will neglect them : the second term clearly only leads to local corrections , while the third term and others like it lead only to subdominant power laws , which will always be present on the lattice .
the conclusion is the same for corrections to the magnetic field , but for completeness we show some of the representative terms : @xmath240
using the effective theory developed above , we now treat some of the more striking properties of the @xmath1 spin liquid . since the effective theories for both the generic coulomb phase and the adjacent rk point are quadratic in the fields , all of the calculations can be done by simple gaussian integrals .
as already mentioned , the @xmath1 spin liquid supports a gapped @xmath3 spinon carrying electric gauge charge , a gapped topological point defect that plays the role of a magnetic monopole , and a gapless , linearly dispersing `` photon '' with velocity @xmath241 .
the spinon was discussed in section [ sec : frust - gt ] , and in the microscopic models takes the form of a tetrahedron ( or octahedron ) carrying nonzero @xmath242 .
the gap to the spinons is very large , of order @xmath17 .
the monopole is a classical configuration of the magnetic field emanating from a point with nonzero magnetic charge at @xmath243 . for a compact gauge field on , say , the cubic lattice , the classical configuration @xmath244 of the vector potential is given by minimizing @xmath245 = -\sum_{\langle{\sf r}{\sf r}'\rangle } \operatorname{cos}(\pi{\cal b}_{{\sf r}{\sf r}'})\ ] ] in the presence of a specified distribution of magnetic charge , which enters via the constraint @xmath246 . in these expressions @xmath247 .
far from the center of the monopole the magnetic field will be small , the energy is approximately a sum over @xmath248 , and the minimum is obtained by solving the ( discrete ) laplace equation of classical electrostatics .
the compact nature of the theory only modifies the field near the core . in the effective theory
, we can create a monopole with the operator @xmath249 the exact choice of @xmath250 in this operator is somewhat arbitrary ; the requirement for monopole creation is that the surface integral of the magnetic field at large distances indicates the presence of one magnetic charge .
this arbitrariness does not matter when calculating long - time monopole correlators , which will be dominated by the contribution from the lowest - energy monopole eigenstate ( it is not hard to see that , in the effective theory , our operator has very good overlap with this state ) .
as long as the magnetic field is chosen to spread the flux uniformly from the monopole center , long - distance correlations should also be unaffected . because magnetic charge is created at @xmath243 , this operator is _ not _ invariant under continuous dual gauge transformations .
the monopole gap is of order @xmath251 , and in the microscopic models is likely to be of order @xmath252 as long as @xmath251 does not renormalize too much from its bare value .
therefore the monopoles have a much smaller gap than the spinons . using the correspondence between microscopic and effective variables
, we can write down a creation operator that should have at least some overlap with the true microscopic monopole eigenstate : @xmath253 this is written in terms of the direct variables to connect with the spin language , where the monopoles are defect configurations in the xy component of the spin . because of the half - integer electric field , the sign of @xmath254 is changed by a @xmath103-shift of any one of the @xmath255 . however , this overall sign does not fluctuate since the electric fields change only by integer steps .
both the monopoles and spinons feel an emergent @xmath10 interaction , even though the microscopic hamiltonians contain only local operators .
consider the field due to a configuration of a few static electric gauge charges ( the same discussion could be repeated for magnetic charge ) .
this is given by solving poisson s equation @xmath256 to obtain coulomb s law . by gauge invariance
the longitudinal part of @xmath257 does not fluctuate , so the field is simply given by its classical value plus _ transverse _ fluctuations .
this zero point energy does not contribute to the energy difference between the charged state and the charge - free ground state , and one recovers the familiar @xmath10 coulomb potential . in the rk point
effective theory , this interaction is absent for the spinons , but still present for monopoles , because there is no energy cost for longitudinal electric fields .
a word about the statistics of the charged excitations is in order .
certainly , both monopoles and electric charges as discussed above are bosonic .
however , a simple argument shows that a _ dyon _ , a bound state of electric and magnetic charge , has fermi statistics@xcite .
if such bound states exist and have lower energy than pure electric charges , then one could say that there are fermionic spinons .
while this is a nonuniversal question with no bearing on the low - energy physics , it can be relevant in understanding the possible transitions out of the @xmath1 spin liquid .
the emergence of fermi statistics in local bosonic models is also of great conceptual interest@xcite , although in the present case it is not clear what , if any , local microscopic interaction will bind electric and magnetic charge .
finally we turn to the photon . in terms of the spins , it is a linearly dispersing @xmath4 collective mode oscillating between the ising and xy parts of the spin vector . as for _ phonons
_ , these gapless excitations make the contribution to the low - temperature specific heat @xmath258 .
since in real materials it should be possible to quantitatively understand the phonon contribution to @xmath259 by measurement of elastic moduli@xcite , this signature of the @xmath1 spin liquid should be easily accessible to experiments .
other potential , though likely more delicate , probes of the photon are low - temperature thermal conductivity and raman scattering . finally , the photon manifests itself in the power - law correlations discussed in section [ sec : power - laws ] . in the theory of @xmath9-fractionalized phases , the notion of topological ground state degeneracy@xcite
has been very useful both as a conceptual tool , and as a property that can be directly probed by experiments@xcite .
the degeneracy is associated with the topological sectors of a deconfining @xmath9 gauge field , so it is natural to ask about the generalization of these ideas to @xmath1-fractionalized states .
we will work with the microscopic hamiltonian of cubic frustrated gauge theory ( eq . [ eq : cubic - gt ] ) on a 3-torus , a cube with @xmath260 lattice sites on a side and periodic boundary conditions : @xmath261,j } & = & e_{{\bf r}j } \\
\operatorname{exp}(i a_{[{\bf r}+l{\bf e}_i],j } ) & = & \operatorname{exp}(i a_{{\bf r}j } ) \nonumber \text{.}\end{aligned}\ ] ] we define an operator to measure the electric flux through a plane perpendicular to each of the three independent cycles of the torus .
for example , @xmath262,x } \text{,}\ ] ] with similar definitions for @xmath263 .
these fluxes are constants of the motion , and are in fact conserved by _ any _ local , gauge invariant dynamics .
furthermore , because there is no electric charge in the ground state , by gauss law we can make arbitrary incremental deformations of the specified plane without changing the flux .
these fluxes define the electric topological sectors of the gauge theory@xcite .
now imagine the model is in the coulomb phase , where in the ground state @xmath264 .
we imagine threading one quantum @xmath265 of electric flux through the system in , say , the @xmath78-direction . because the energy is proportional to @xmath266 ,
the most favorable situation is for the flux to spread itself uniformly through the system , so that @xmath267 .
this electric field is purely longitudinal and does not fluctuate , so there is a total energy cost @xmath268 . therefore , in the thermodynamic limit , all states with a finite number of electric flux quanta threading each direction of the 3-torus become degenerate with the ground state . just as in the case of @xmath9 topological order , these states are locally identical to the ground state ( local correlation functions will be unaffected ) , but globally distinct .
what about _ magnetic _ topological sectors of the gauge field ? because @xmath269 = 0 $ ] , these can clearly be specified simultaneously with the electric sectors .
we define magnetic fluxes @xmath270 just as above : @xmath271,x } \text{,}\ ] ] and similarly for @xmath272 .
this expression is a ( lattice ) surface integral over a plane bisecting the system , which has the topology of a 2-torus due to the periodic boundary conditions .
because this is a closed surface and @xmath117 is the curl of a compact vector potential , the magnetic flux is quantized : @xmath273 .
however , there is an important difference from the case of electric flux : because magnetic charge fluctuates in the ground state , @xmath270 is _ not _ a constant of the motion in the microscopic theory . within the coulomb phase , though , the magnetic flux defined in the effective theory does commute with @xmath274 , the effective hamiltonian written in dual variables .
magnetic charge fluctuations can mix states with different values of the flux by creating a monopole - antimonopole pair and separating the particles along one lattice direction , only to have them annihilate when they complete their periodic journey around the system ( fig .
[ fig : flux - thread ] ) .
because the monopoles are gapped , the rate for this process is suppressed exponentially in the system size , and it does not occur for a large enough sample . as with the electric sectors ,
the magnetic flux sectors become degenerate ground states as @xmath275 with energies scaling to zero as @xmath276 . in the microscopic theory
, we expect that @xmath277 will label the topological sectors within the coulomb phase ; in confining phases magnetic charge fluctuates wildly and @xmath270 is no longer a good quantum number , even approximately . to summarize , the @xmath1 spin liquid on a 3-torus has a topological degeneracy characterized by six _ integers _ , with energies that vanish as @xmath278\\ & + & { \cal k}_c \frac{(\phi^b_0)^2}{l } \big[(n^b_x)^2 + ( n^b_y)^2 + ( n^b_z)^2 \big]\nonumber \text{.}\end{aligned}\ ] ] in the rk point effective action @xmath211 vanishes , so the electric topological sectors have zero energy , with only corrections from irrelevant terms possible .
we first consider the two - point @xmath242 correlation function both in the coulomb phase and at the rk point ; the transverse part of the spin - spin correlator is not gauge - invariant and vanishes .
the mappings of section [ sec : frust - gt ] tell us this is given by the electric field correlator . in the cubic case
we use the three - site unit cell containing @xmath279 . because of the orientation convention for the electric field , we have @xmath280 , where @xmath281 and we take @xmath282 in the @xmath283 sublattice . therefore the spin correlation function will be shifted from the electric field one by @xmath284 in the brillouin zone .
for the pyrochlore we use the unit cell of section [ sec : pyro - model ] containing the four sites @xmath32 , where @xmath285 labels the centers of up - pointing tetrahedra ( up - pointing diamond sites ) . in this case @xmath286 .
we calculate the equal - time correlator @xmath287 in the gauge - fixed continuum theory for the coulomb phase , where @xmath288 .
we illustrate the calculation with a single component : @xmath289 where we have imposed the hard momentum cutoff @xmath290 , @xmath291 is the angle between @xmath292 and @xmath285 , @xmath293 is the angle between @xmath292 and the @xmath2-axis , and the @xmath294 integral is over the angular direction of @xmath292 .
expanding the exponential in legendre polynomials of @xmath295 , and using the addition theorem to rewrite these in terms of spherical harmonics , the integral can be done to find : @xmath296 here @xmath297 is the angle between @xmath285 and the @xmath2-axis .
we have dropped terms oscillating at the cutoff wavevector , which are unphysical artifacts of the hard cutoff .
alternatively , we can use a soft cutoff by inserting @xmath298 in the integrand , integrating over all @xmath299-space , taking the limit @xmath300 and keeping only the dominant powers of @xmath301 to recover the same result . in the general case
we find : @xmath302 the striking angular dependence of this correlator is a manifestation of the inherent vectorial structure of the @xmath1 spin liquid , which comes in turn from the vector fields of @xmath1 gauge theory .
for the pyrochlore model in the @xmath1 spin liquid phase we have : @xmath303 where @xmath195 are unit vectors connecting @xmath285 with its nearest - neighbor _ diamond _ sites . in the cubic case : @xmath304 in the rk point effective theory , a similar calculation shows that both the radial and angular dependence of the correlator changes .
it takes the familiar `` dipole ''
form : @xmath305 falling off as @xmath306 .
it is also interesting to calculate the correlations of the kinetic energy density . in our ring exchange models
this naturally lives on the plaquettes and in the cubic case has the form @xmath307 in the pyrochlore model eq .
( [ eq : ke - density ] ) looks the same , but with ring exchange on the hexagonal plaquettes as in eq .
( [ eq : pyro - heff ] ) . long range order at @xmath308 in @xmath309 would indicate a plaquette density wave state with broken translation symmetry . to simplify notation we work out the cubic case , where @xmath310 the prefactor involving @xmath311 is a nonuniversal constant . at large separation
, the second factor can be evaluated in the continuum theory , giving : @xmath312\text{.}\ ] ] we can immediately write down the magnetic field correlator by duality , which simply interchanges @xmath211 and @xmath313 within the coulomb phase . therefore @xmath314 and the kinetic energy density exhibits power - law @xmath315 correlations and a nontrivial angular dependence .
the rk point theory does not have this self - duality property , so we need to evaluate another integral to find the magnetic field correlator . using a hard cutoff ,
one finds _ only _ unphysical oscillatory terms . using the @xmath298 soft cutoff , the result is @xmath316 .
this cutoff - dependence indicates that the angular dependence is probably nonuniversal , since there should be other @xmath317 contributions with nontrivial angular factors that have been left out of the continuum theory . in this section
we calculate the monopole propagator and verify that it falls off exponentially in space and imaginary time , both in the coulomb phase and at the rk point .
this is one of the key properties of the @xmath1 spin liquid , since at the transition to any nearby confining phase the monopoles will proliferate and condense .
it is most convenient focus on the equal - time monopole propagator @xmath318 , working on the spacetime lattice . for @xmath319 , this expectation value _
vanishes _ taken with respect to @xmath234 because it creates magnetic charge at two points . to understand the true behavior we need to include magnetic charge fluctuations and add the corrections eq .
( [ eq : non - ginvt - corrections ] ) to the gaussian action . for simplicity
, we consider only the restricted set of correction terms : @xmath320 we want to calculate the propagator in perturbation theory in @xmath321 : @xmath322 \bigg[\\ & & \operatorname{exp}(i \sum_{\langle{\sf r}{\sf r}'\rangle } { \cal b}_{{\sf r}{\sf r } ' } \tilde{\alpha}_{{\sf r}{\sf r } ' } ) \operatorname{exp}(- { \cal s}^0_{lat } - \delta{\cal s})\bigg ] \text{,}\nonumber\end{aligned}\ ] ] where @xmath250 is the classical magnetic field due to a monopole - antimonopole pair at @xmath323 and @xmath324 , respectively . to obtain a nonvanishing contribution
, we need to bring down correction terms until the complex exponential has been modified to create no magnetic charges .
to lowest order we obtain : @xmath325 where the product of cosines is taken over the shortest path connecting @xmath323 and @xmath324 ; for simplicity we restrict our attention to geometries where this is unique , although this is inessential .
we have also dropped the background field @xmath184 , which gives only an overall sign .
the gauge - invariant part of the expectation value comes from the term in the product of cosines that threads one magnetic flux quantum from @xmath323 to @xmath324 , giving a new magnetic field @xmath326 with zero divergence , and allowing us to express the result as the expectation value of a gauge - invariant operator : @xmath327 what the perturbation theory has done is exactly to connect the monopole and antimonopole with a dirac string .
the exponential decay is clear from the prefactor in eq .
( [ eq : mono - prop - calc-3 ] ) . to evaluate the corrections to this
, we apply the faddeev - popov procedure to the functional integral for the gauge - invariant expectation value , and calculate using the photon propagator .
we could _ not _ have done this at the outset since the original operator was not gauge - invariant .
it is most convenient to integrate by parts inside the exponential to obtain a result in terms of the classical vector potential and the electric field : @xmath328 the necessary integral is quite difficult to evaluate , so we resort to power - counting to determine the largest possible contribution . consider a spatial separation of @xmath301 between monopole and antimonopole .
one contribution will come from the region near the pair , giving a factor of @xmath329 from the integration , @xmath330 from the two vector potential factors , and @xmath331 from the electric field correlator in the coulomb phase .
these factors multiply to give a constant , so we expect that the largest possible contribution to the integral is logarithmic in @xmath301 , which contributes only a power - law prefactor to the propagator .
other contributions involving regions far away from the pair make subdominant contributions . at the rk point
there is an extra power of @xmath301 from the electric field correlator , and the dominant possible contribution is linear in @xmath301 and gives a correction to the correlation length .
finally it is clear from these considerations that the unequal - time propagator is not substantially different , and decays exponentially in space and time .
in this section we return to the microscopic spin models , and use the exact ground state at the rokhsar - kivelson point to extract information about the physics nearby .
we begin with a discussion of the structure of the wavefunction and some simple properties that can be seen analytically , then proceed to a discussion of the numerical evaluation of several quantities .
in addition to various equal time correlation functions , we make use of a remarkable property of rk - type points discovered by henley@xcite to approximately evaluate the imaginary - time monopole propagator . we will consider finite - size cubic and pyrochlore lattices in this section . to fix notation , for the cubic lattice
@xmath260 denotes the number of sites on a side of a periodic system . in the pyrochlore case
we take periodic boundary conditons so that @xmath285 and @xmath332 are identified , and we can think of @xmath260 as the number of tetrahedra along one periodic direction .
the rk wavefunction is an equal - weight superposition ( with positive coefficients ) of all possible dimer coverings consistent with the local dimer constraint . in the spin models the electric field
can only take on the values @xmath333 , and it is convenient to visualize dimer coverings as zero - divergence configurations of an electric field that can point only forward or backward on each link . within the vast superposition
there are components of every electric topological sector .
since these are not mixed by the dynamics they should be thought of as degenerate but distinct ground states .
the rk point exhibits the electric sectors of @xmath1 topological order , and as in the gaussian rk action they are exactly degenerate even in a finite - size system . to understand what sector to focus on
, it is profitable to consider first - order perturbation theory in @xmath334 away from the rk point .
the first - order shift in the ground state energy is @xmath335 , where @xmath336 is the average number of flippable plaquettes _ in a given sector_. for @xmath337 , sectors with _ no _ flippable plaquettes , and hence no dynamics , will have the lowest energy , and we reach the usual conclusion that a valence bond crystal obtains on this side of the rk point . for @xmath338
, however , the sector with the greatest average flippability wins out .
this is in some sense the most disordered sector , and should have @xmath264 ; several numerical checks support this conclusion .
we are interested in @xmath338 , so we focus on the zero - flux sector .
we also wish to consider sectors that violate the zero - divergence constraint and contain spinons . within each such sector ,
the lowest energy state is again given by an equal - weight superposition of all electric field configurations consistent with a background charge density specifying the spinon positions .
these are exact eigenstates with energy @xmath339 times the number of spinons , which in a finite - size system is always even .
consider in particular a sector with two spinons .
there is clearly zero energy cost no matter how the spinons are moved around , so the rk point is deconfined and has no coulomb potential between static electric charges .
again , this is consistent with the effective action eq .
( [ eq : rk - effective - action ] ) .
finally , we calculate the variational energy of the magnetic topological sectors . in general
, we create a classical configuration of the vector potential by acting on the rk ground state with the operator @xmath340 , just as for the special case of monopole creation in eq.([eq : micro - monopole - creation ] ) .
of course we expect only that the resulting state has some overlap with the desired eigenstate of the gauge theory . denote the rk wavefunction by @xmath341 where the sum is over all configurations of the electric field in the desired sector and @xmath342 is the number of states contributing to the sum . defining @xmath343 $ ] to be the number of flippable plaquettes in a given configuration , and
@xmath344 $ ] to be one if the specified plaquette is flippable and zero otherwise , the variational energy of interest is given by : @xmath345 & = & \langle \psi_{rk } | { \cal o}^{\dagger } { \cal h}_{rk}\ , { \cal o } | \psi_{rk } \rangle \\ & = & \frac{j_{ring}}{{\cal n } } \sum_{\{e\ } } \big [ n_f[\{e\ } ] \nonumber \\ & & - \sum_{\square } p_f[\{e\},\square ] \operatorname{exp}(2 i \sum_{{\bf r}{\bf r } ' \in \square } { \cal a}_{{\bf r}{\bf r } ' } e_{{\bf r}{\bf r } ' } ) \big ] \nonumber\\ & = & \frac{j_{ring}}{{\cal n } } \sum_{\{e\},\square } p_f[\{e\},\square ] \big[1 - \operatorname{cos}\big((\operatorname{curl } { \cal a})_{\square } \big ) \big]\text{. } \nonumber\end{aligned}\ ] ] here we use notation appropriate to the cubic model ; the generalization to the pyrochlore is obvious . specifically , consider the case where @xmath346 threads one quantum of magnetic flux through the @xmath78-direction in the cubic model .
then @xmath347 for plaquettes in the @xmath78-direction and zero otherwise , and @xmath348 where @xmath349 is the average flippability per plaquette . for @xmath275
this energy goes to zero as @xmath276 as expected , and further gives a rough value for the `` magnetic stiffness '' at the rk point : @xmath350 . using the numerical methods discussed below , for the cubic model in the zero - flux sector we find @xmath351 . for the pyrochlore a similar calculation shows @xmath352 ; the factor of two arises because the flux passes through two kinds of hexagons . in this case
we find numerically @xmath353 .
because the rk wavefunction has positive and equal weights , equal - time properties can be evaluated by infinite temperature monte carlo for the associated _ classical _ dimer model .
the simplest possibile monte carlo step is to : 1 ) randomly choose a plaquette .
2 ) if the plaquette is flippable , flip it , otherwise do nothing .
as desired , this algorithm preserves the electric flux @xmath354 . for the measurement of the quantities discussed below
, the algorithm was run for as many as @xmath355 monte carlo steps at a given system size . in some cases
these long runs were necessary to achieve good accuracy because the desired quantity was very small .
it is not clear whether the single ring move algorithm is ergodic within each electric field sector . for @xmath356
cubic and pyrochlore systems we have performed an exact enumeration of all allowed configurations . in the cubic case the zero - flux sector contains 880 states ; 864 of these have flippable plaquettes ( with @xmath357 on average ) , and are connected under single ring moves .
the other 16 states have no flippable plaquettes . for the pyrochlore model
there are 384 zero - flux states connected under single ring moves ( @xmath349 = 1/4 ) , and 12 zero - flux states with no flippable hexagons .
evidently the single ring move algorithm is not ergodic within the zero - flux sector , but it does not matter if we fail to access states with no flippable plaquettes . in our simulations
we generate initial configurations for larger lattices by periodically repeating a state from the flippable part of the @xmath356 zero - flux sector , so more insidious problems could occur if this sector breaks up into multiple flippable subsectors in larger lattices , each closed under single ring moves . for an analogous two - dimensional model ,
it is possible to prove that single ring moves are ergodic in each electric flux sector ( see appendix [ app:2d - ergodic ] ) . in the absence of a similar result in three dimensions
, we have checked some of our results for the cubic model with an algorithm due to barkema and newman that we believe is probably ergodic@xcite .
this algorithm is illustrated in figure [ fig : ice - algorithm ] ; in the present case the basic idea is to flip loops of arbitrary length ( including single rings ) , keeping only non - contractible loops that do not change @xmath354 .
all properties measured using both algorithms gave the same results ; we present two examples in the following section . while we have not implemented an algorithm with `` loop moves '' for the pyrochlore model , due to the great similarity of all measured properties to their cubic analogs it seems unlikely that ergodicity is an issue .
we measured the electric field - electric field correlator in both the cubic and pyrochlore models . to extract the asymptotic dependence most simply we focused on the correlators measured at a separation of half the system size .
it should be noted that one can not trivially extract the angular dependence discussed in section [ sec : power - laws ] by this method , since the two electric field vectors in the correlator will be connected by many paths with length scaling as @xmath260 but with different angles . for the cubic model
we show our results for @xmath358 in figure [ fig : cubic - ee ] .
data from both algorithms are shown and the agreement with the @xmath306 decay predicted by the effective theory is very good .
the same conclusion obtains for the pyrochlore model ( but with only the single ring move algorithm ) . in that case
the correlator measured was @xmath359 ; the data are shown in figure [ fig : pyro - ee ] .
measurements of other orientations of the electric field correlator in both models all showed the same @xmath306 decay .
+ 6 mm to show the very good agreement with the @xmath306 decay expected from the effective action.,width=316 ] -4 mm [ fig : cubic - ee ] to measure the equal time monopole propagator it is necessary to address some subtleties that arise when putting objects with gauge charge in a finite - size system .
first , it is clear that because the system is a compact manifold with no boundary , it must have zero total magnetic charge .
since the operator @xmath360 creates a monopole - antimonopole pair , this is not a problem for the propagator .
however , consider the cubic lattice and let @xmath361 . for any value of @xmath362
there will be planes where @xmath363 , which is not allowed for a compact vector potential as discussed in section [ sec : toporder ] .
we can ameliorate this problem by creating a _
double strength _
monopole - antimonopole pair , but even in this case the only allowed separation is @xmath364 . because of these complications , in the cubic case we only measure the double - strength propagator at a separation of half the system size : @xmath365 by time - reversal or ising symmetry , @xmath366 is real .
we obtained the vector potential appropriate for eq .
( [ eq : finite - size - cubic - mono - prop ] ) by first solving a discrete poisson s equation by numerical matrix inversion , then feeding the solution into eq .
( [ eq : monopole - classical - energy ] ) .
this was minimized by a combination of simulated annealing and direct minimization ( i.e. zero - temperature simulated annealing ) .
because of the very rapid decay , it was only practical to measure @xmath366 for @xmath367 ; results are shown in figure [ fig : cubic - mono ] and are consistent with exponential decay with power - law corrections . + 6 mm ,
the line is a guide to the eye with slope @xmath368.,width=316 ] -4 mm [ fig : pyro - ee ] on the pyrochlore lattice it is not possible to have double - strength magnetic charges in the microscopic model ( see sec . [
sec : pyro - mappings ] ) , so for consistency with flux quantization we must consider a more complicated geometry .
we look at the following propagator , which creates two monopoles on adjacent up- and down - pointing ( dual ) diamond sites , and similarly two antimonopoles separated by a distance @xmath369 : @xmath370 where @xmath323 is an up - pointing diamond site and @xmath324 is the down - pointing site directly above .
the results are shown in figure [ fig : pyro - mono ] , and are consistent with exponential decay ; we do not believe the apparent lack of substantial power - law corrections is significant , since the coefficient of these corrections is presumably nonuniversal .
+ 6 mm -4 mm [ fig : cubic - mono ] in the cubic model , we also measured the potential between a pair of static spinons in first - order perturbation theory away from the rk point . in the notation of sec .
[ sec : rk - general - props ] , we consider @xmath338 . the coefficient of @xmath371 in the spinon - spinon potential is given by : @xmath372 to measure this quantity , we generated sectors with one @xmath3 spinon at @xmath282 , and another at @xmath373 for @xmath374 .
the results are shown in figure [ fig : cubic - spinon - potential ] , and are consistent with a @xmath10 coulomb potential . while the perturbation theory does not necessarily `` know '' whether a deconfining phase with a coulomb potential can be stable
, it does indicate @xmath375 for small @xmath338 .
since the coulomb phase _ is _ stable , we conclude it indeed exists adjacent to the rk point over a finite range @xmath376 .
+ 4 mm ) .
the results are consistent with exponential decay , with no indication of power - law corrections.,width=316 ] ) for several system sizes of the cubic model .
the horizontal axis is the distance between the spinons , @xmath377 .
the curve is a fit of the data to the functional form @xmath378.,width=316 ] -4 mm in our models it is possible to approximately measure quantum imaginary time correlation functions using only the classical monte carlo dynamics .
the key observation , due to henley@xcite , is that at the rk point the master equation describing the monte carlo dynamics of the single ring move algorithm is _ identical _ to the imaginary - time schrdinger equation
. we will be interested in monopole correlation functions built from the equal - time correlators constructed above . in the cubic case ,
for example , we measure : @xmath379 where @xmath380 .
the analogous definition of @xmath381 for the pyrochlore model is constructed from the equal - time propagator eq .
( [ eq : finite - size - pyro - mono - prop ] ) . to extract information about the spectrum from the simulation , it is useful to relate the units of classical monte carlo time to quantum imaginary time .
let @xmath382 be the total number of plaquettes in the system .
if we begin with some electric field configuration @xmath383 and execute one monte carlo step , the probability of remaining in the same state is @xmath384/n_p)$ ] , while that of making a transition to each of the states accessible by flipping one plaquette is @xmath385 .
we now consider the time - discretized imaginary - time schrdinger equation and fix the timestep @xmath386 to recover the same values , which now enter as probabillity _
we consider : @xmath387 where the final sum is over those electric field configurations connected to @xmath383 by a single ring move .
it is only valid to neglect the higher - order terms in @xmath388 when @xmath389 ; in fact , the correspondence between the _
discrete_-time quantum dynamics and the monte carlo dynamics ( which are necessarily discrete in our simulation ) , is only strictly valid in this limit . in our models @xmath390 is of order @xmath391 , so we expect to at best make quantitative errors of about @xmath392 , and at worst to get the wrong answer
. this problem can not be alleviated by the means at hand because the discrete classical algorithm _
fixes _ the time - step for the quantum dynamics .
a more careful treatment would require a direct simulation of the master equation with control over @xmath386 , or quantum monte carlo . if we nevertheless expand the exponential , setting @xmath393 correctly matches the classical probabilities and quantum amplitudes .
using this relation , we write down the appropriate quantity to simulate to measure the imaginary time correlator : @xmath394 \nonumber\text{.}\end{aligned}\ ] ] here @xmath35 is an integer number of steps in monte carlo time , and @xmath395 labels all possible monte carlo time evolutions of time @xmath35 starting from the configuration @xmath383 .
the number of such evolutions is @xmath396 , and @xmath342 is the total number of electric field configurations in the zero - flux sector .
finally , we have defined @xmath397 .
+ 6 mm for the cubic model .
the horizontal axis is imaginary time in units of @xmath398 .
data is shown for @xmath399 ; as discussed in the text , as @xmath260 increases the exponential decay in imaginary time becomes cleaner , suggesting the monopole is indeed gapped.,width=316 ] , the approximate monopole - antimonopole imaginary time correlator in the pyrochlore model .
the horizontal axis is imaginary time in units of @xmath398.,width=316 ] -4 mm we have measured the monopole imaginary time correlators for both the cubic and pyrochlore models ; results are shown in figures [ fig : cubic - mgap ] and [ fig : pyro - mgap ] .
note that the extreme long time behavior is not related to the gap ; instead , @xmath400 will have some overlap with the ground state , causing @xmath401 to approach a constant as @xmath402 . as we go to larger system sizes and the _ spatial _ separation of the monopole - antimonopole pair in the correlator increases
, there is a clean exponential decay persisting for longer times .
this is illustrated for several system sizes of the cubic model in fig .
[ fig : cubic - mgap ] . from the numerical data
one can extract a rough value for the energy of the monopole - antimonopole configuration put in at @xmath403 ; in both cases this is about @xmath404 .
while these results are approximate , it is unlikely that the monopoles are somehow gapless given the circumstantial evidence for a gap , particularly the very rapid exponential decay of the equal - time propagator .
in this paper we have argued for the existence of a novel fractionalized quantum disordered state , the @xmath1 spin liquid , in spin models with a global @xmath1 symmetry . while we found it necessary to add an extra interaction to the easy - axis limit of the pyrochlore heisenberg model to proceed analytically
, we speculate that this may be only a crutch and that the pure easy - axis model is in the @xmath1 spin liquid phase . both the pyrochlore and cubic models have no sign problem , so this issue could be resolved by quantum monte carlo simulations@xcite .
while it would be a remarkable result if the easy - axis pyrochlore heisenberg antiferromagnet were a @xmath1 spin liquid , we believe the results in this paper are enough of an `` existence proof '' that the phase is likely to be present in less tractable but more realistic microscopic models .
the converse problem , of determining the general requirements for a spin model to exhibit an emergent @xmath1 gauge structure , is of course more challenging .
it would appear that , since monopole excitations are required to enable a transition to a neighboring confined phase , any such gauge theory must be compact , and hence appear _ on the lattice_. thus , unlike emergent global symmetries , which can appear at low energies and long wavelengths , an emergent @xmath1 gauge structure would appear to admit only a limited degree of spatial coarse - graining . given the additional need for spin - carrying degrees of freedom , the unit cell of an emergent gauge theory in a pure spin model would appear to necessarily contain more than a single @xmath0 spin . based on the present examples , which take advantage of the bipartite structure of the cubic and diamond lattices , it is tempting to speculate that the models should contain a natural bipartite sublattice .
it is easy to see , however , that the emergent gauge structure in these models is stable to adding arbitrary weak additional multi - spin interactions , including those that break any bipartite symmetries .
further understanding , including the important issue of whether the @xmath1 spin liquid is possible in a magnet with global @xmath37 symmetry , and if yes , of what its properties might be , must await further study .
certainly , the most interesting issue is whether the @xmath1 spin liquid can be found in a real material .
the simplest signature is the additive @xmath21 contribution to the specific heat from the photon .
because it is likely to be possible to quantitatively understand the _ phonon _ specific heat via independent measurements@xcite , this could provide a relatively clear and simple test for emergent photons .
further theoretical work may be necessary to understand more delicate probes of the photon , like heat current and ( possibly ) raman scattering ; phonon-``photon '' interactions and disorder may play a nontrivial role . while disorder may be important for some experimental properties
, we remark that simple arguments demonstrate that disorder does not destabilize the @xmath1 gauge structure or the gapless photon . since topological order can coexist with conventional broken symmetry , spin liquids are not the only good places to look for emergent photons .
one can imagine , for example , condensing spin - carrying but gauge - neutral excitations ; in our models this would lead to an ordered magnetic state with gapless spin waves and a @xmath1 gauge structure .
understanding the possibilities for and properties of phases near the @xmath1 spin liquid is another problem worthy of consideration if contact is to be made with experiment .
very recently , senthil , vojta and sachdev have made the interesting suggestion that @xmath1-fractionalization may provide the answer to some of the puzzles of heavy fermion materials@xcite .
we leave these intriguing issues aside to remark that @xmath1-fractionalized states are a remarkable and novel possibility for the physics of strongly correlated electrons in _ three _ dimensions , heretofore a relatively unexplored area .
finally , we note that huse , krauth , moessner and sondhi are considering a model closely related to ours@xcite .
we thank c. broholm , c. l. henley , r. g. melko , r. moessner , o. i. motrunich , r. d. sedgewick , t. senthil and a. vishwanath for numerous stimulating and helpful discussions .
m.h . is grateful for support from the department of defense through the ndseg program , and from uc santa barbara .
was generously supported by the nsf under grants dmr-0210790 and phy-9907949 .
l.b . was supported by the nsf through grant dmr-9985255 , and by the sloan and packard foundations .
we here enumerate the action of the various symmetries of the cubic model on all the microscopic operators , both in dual and original variables . *
discrete translations : @xmath405 . the electric field transforms as @xmath406 , with similar expressions for @xmath117 and @xmath407 .
because of the non - translation - invariant background @xmath184 , the dual vector potential obeys the more complicated transformation law : @xmath408 where the shift @xmath409 is defined to satisfy @xmath410 * lattice rotations : let @xmath411 rotate the lattice into itself about some fixed origin .
for example , take the origin to be @xmath282 and make a @xmath412-rotation about the @xmath2-axis
. then @xmath413 = [ -n_y { \bf x } + n_x { \bf y } + n_z { \bf z},y]$ ] , and so on .
we have @xmath414}$ ] , and similarly for @xmath117 and @xmath407 . again
@xmath415 } + \delta\alpha_{{\cal r}[{\sf r}i]}$ ] , with @xmath416 } = \pi ( e^0_{{\cal r}[{\bf r}i ] } - e^0_{{\bf r}i})$ ] . *
reflections : let @xmath417 denote reflection about a plane with normal vector @xmath418 , @xmath419 or @xmath420 . for definiteness , choose the plane with normal @xmath418 at @xmath421 , then : @xmath422 for @xmath423 .
the vector potential obeys a similar transformation law .
the magnetic field and dual vector potential are _ pseudovectors _ and transform under reflections with an additional minus sign .
again , @xmath424 transforms with an appropriate shift @xmath409 to compensate for changes in the background under reflections . *
gauge and dual gauge invariance : under gauge transformations , @xmath425 , where @xmath426 is a phase defined on the cubic sites .
similarly we have the dual gauge transformations @xmath427 , with @xmath150 . in the action
the dual gauge transformations can be spacetime - dependent , with @xmath428 * ising ( or particle - hole ) symmetry : @xmath429 , @xmath430 and @xmath431 . also @xmath432 .
* time reversal : since spin operators obey @xmath433 under time - reversal , we have : @xmath429 , @xmath434 and @xmath435 . for consistency with the electric field transformation law , @xmath436 . in the action
the latter relation continues to hold , as long as we also send @xmath437 .
furthermore , the temporal component of the dual vector potential transforms as @xmath438 . note that the situation here is the reverse of that in real electromagnetism , where @xmath439 changes sign under time reversal and @xmath257 is invariant . these symmetries are all present in analogous forms in the diamond lattice gauge theory .
there is also an additional global symmetry , which exchanges the two sublattices of up- and down - pointing sites .
we consider the classical dimer model on the square lattice at infinite temperature , with two dimers touching every site .
this is the @xmath217 analog of the rk point of our cubic model .
we work in the electric field language , where @xmath440 with @xmath441 , and @xmath442 .
consider an @xmath443 system ( @xmath260 even ) with periodic boundary conditions : @xmath444,j } = e_{{\bf r}j}$ ] , where @xmath445 .
the zero - flux sector is specified by the condition : @xmath446 we will show that any two states in this sector are connected by a sequence of ring exchange moves on single square plaquettes . the key step is to go to a height representation on the dual lattice with sites at the plaquette centers @xmath447 .
we define : @xmath448,y } \\
h({\sf r } + { \bf y } ) - h({\sf r } ) & = & 2
e_{[{\sf r } + ( { \bf y } - { \bf x})/2],x } \nonumber\end{aligned}\ ] ] the content of this definition is that the height increases / decreases by one if we cross a link of the direct lattice with electric field pointing to the right / left .
if we fix the value of the height on one site of the dual lattice these definitions determine it uniquely everywhere .
the height is well - defined because the electric field has zero divergence , and can consistently be taken to have periodic boundary conditions because we are in the zero - flux sector .
a flippable plaquette has a height either above or below all of its four neighbors , depending on its orientation . therefore , in a given configuration , the plaquettes with minimum and maximum height will always be flippable , and every configuration has at least two flippable plaquettes .
there are also two states with every plaquette flippable ; up to overall shifts of the height these have @xmath449 these two configurations are connected by single ring moves , since we can flip all the maximum height plaquettes in one to go to the other . to complete the proof , we label configurations by @xmath450 .
this clearly takes on the minimum possible value of unity in only the two maximally flippable states .
suppose we are in some other state with @xmath451 .
then we can flip plaquettes of maximum height until @xmath452 is reduced by 1 .
this procedure can be repeated until @xmath453 and we reach one of the maximally flippable states .
we have thus shown that any two states are connected by a sequence of single ring moves , because the maximally flippable states are connected to all states and to each other .
the presence of such a gauge structure and the possibility of a coulomb phase has been realized independently by moessner , for a pyrochlore ising antiferromagnet in a weak transverse field .
( r. moessner , private communication . ) another point of potential interest in this regard is that the pyrochlore ising antiferromagnet with a small _ ferromagnetic _ transverse exchange or small transverse field ( see ref . ) is likely to have a @xmath1 spin liquid phase ; the lowest - order splitting of the low - energy manifold is identical to that in our model .
these models have no sign problem and should also be accessible to numerical work . | we study the @xmath0 heisenberg antiferromagnet on the pyrochlore lattice in the limit of strong easy - axis exchange anisotropy .
we find , using only standard techniques of degenerate perturbation theory , that the model has a @xmath1 gauge symmetry generated by certain local rotations about the @xmath2-axis in spin space .
upon addition of an extra local interaction in this and a related model with spins on a three - dimensional network of corner - sharing octahedra , we can write down the exact ground state wavefunction with no further approximations . using the properties of the soluble point we show that these models enter the @xmath1 spin liquid phase , a novel fractionalized spin liquid with an emergent @xmath1 gauge structure .
this phase supports gapped @xmath3 spinons carrying the @xmath1 `` electric '' gauge charge , a gapped topological point defect or `` magnetic '' monopole , and a gapless `` photon , '' which in spin language is a gapless , linearly dispersing @xmath4 collective mode .
there are power - law spin correlations with a nontrivial angular dependence , as well as novel @xmath1 topological order .
this state is stable to _ all _ zero - temperature perturbations and exists over a finite extent of the phase diagram .
using a convenient lattice version of electric - magnetic duality , we develop the effective description of the @xmath1 spin liquid and the adjacent soluble point in terms of gaussian quantum electrodynamics and calculate a few of the universal properties .
the resulting picture is confirmed by our numerical analysis of the soluble point wavefunction .
finally , we briefly discuss the prospects for understanding this physics in a wider range of models and for making contact with experiments .
= 10000 |
emergence of new active regions ( ars ) is believed to be manifestation of the solar dynamo action deep in the convective zone @xcite .
@xcite suggested that a magnetic @xmath7-shaped flux tube emerges due to magnetic buoyancy under the resistance of aerodynamic drag .
the process is not available for direct observations , so methods of numerical simulations are still the main source of information on the details of the emergence process .
significant progress was made since the first two - dimensional magnetohydrodynamic ( mhd ) simulations of flux emergence performed by @xcite .
thus , @xcite suggested a thin - flux - tube approximation , and @xcite and @xcite performed compressible mhd simulations for a twisted flux tube .
flux emergence in the turbulent convective zone ( cz ) was also simulated in the framework of the anelastic approximation ( _ e.g. _ , @xcite and references therein ) .
however , according to @xcite , the approximation does not work for the upper cz ( above 0.95 solar radius ) due to large gradients of physical parameters .
the upper 20 mm of the cz thus became the focus of intense research efforts .
thus , @xcite proposed a `` two - step - emergence '' model where a twisted flux rope , driven by buoyancy emerges from the depth of -20 mm and slows down as it reaches the photosphere because the unmagnetized plasma becomes trapped between the rising flux and the photosphere hindering further emergence . during the intermediate phase , slow rise of the magnetized plasma
is accompanied by enhanced expansion in the transverse direction .
the second step of emergence is a runway expansion into the corona due to buoyant instability ( see figures 5 and 7 in @xcite ) .
the model was later elaborated in the framework of 2d and 3d compressible mhd simulations by @xcite .
@xcite and @xcite explored the flux emergence process through the top layers of the cz and the photosphere using 3d radiative mhd simulations and taking into account the effect of compressibility , energy exchange via radiative transfer , and partial ionization . only ars with the total magnetic flux less than 10 ^ 20^ mx ( a typical flux of an ephemeral region @xcite ) were considered .
the comparison of the simulation results with _ hinode _
observations @xcite demonstrated the success of the simulations .
counterparts of observed high - speed down - flows , bright points , anomalous darkenings , transient kilogauss horizontal fields were reproduced by the model and explained .
moreover , @xcite have shown that the rate of emergence ( `` the rate of transport of the longitudinal flux '' ) is higher for the more twisted flux tubes .
the undulating behavior of the field lines connecting the poles of a magnetic dipole ( serpentine field lines ) was demonstrated in the simulations as well .
authors argue that convective dynamics of photospheric granules is responsible for this effect that is ubiquitous in observations of flux emergence . in @xcite the serpentine pattern as well as the lateral expansion of magnetic flux
is demonstrated in case of emergence of a semi - torus .
the process of flux emergence , sunspots formation , and subsequent decay was recently modeled by @xcite starting from the rise of a semi - torus shaped magnetic flux of 1.7@xmath810 ^ 22^ mx from the depth of 15.5 mm .
authors have shown that large - scale subsurface flows can affect the resulting emerged magnetic structure .
the above brief discussion of the most relevant theoretical works is by no means a comprehensive review of numerical simulation efforts aimed to model the flux emergence on the sun . for an extended review ,
we recommend the _ living review in solar physics _ by @xcite .
impressive achievements of numerical simulations should be compared with observations and explain observed features , as they do frequently . however , for the further progress , efforts in opposite direction are desirable , _
i.e. _ , characteristics of flux emergence that follow from observations , should complement the modeling progress . when analyzing flux emergence using observations , one of the essential parameters to study is the total unsigned flux and its time variations .
frequently the time derivative of the total unsigned flux , the flux emergence rate or flux growth rate , is explored as well .
@xcite studied 101 flux emergence events ranging from small ephemeral regions to large ars by using _
hinode_/sot filtergrams and magnetograms .
they calculated the flux growth rate as the ratio of the saturated unsigned flux to the total emergence time and considered this value as the average flux growth rate , @xmath9 . a relationship between @xmath10 and @xmath5
was found : @xmath11 . under a set of simplifications
it was shown that the index should be 0.5 , which is close to 0.57 .
authors consider this inference as a partial confirmation of an elaborated physical view of the solar flux emergence .
@xcite studied the early emergence of two moderate - size ars using sdo / hmi vector magnetic field data and doppler velocities focusing on moving dipolar features observed between the leading and following parts of an ar .
their observed characteristics can be well explained in the framework of @xcite modeling that predicts unavoidable appearance of serpentine - like field lines responsible for near - surface reconnection and the discharge of mass from rising magnetic flux tubes .
@xcite studied emergence during the first @xmath12 hours since the first appearance of a new flux using soho / mdi magnetograms and dopplergrams of 224 emerging ars .
the flux emergence rate , @xmath13 , was derived as the ratio of the maximum ( during @xmath14 ) unsigned magnetic flux , @xmath5 , and the interval length , @xmath14 . their scatter plot
( figure 8 in @xcite ) allows to infer a power law relationship between the @xmath5 and @xmath13 with an index of about 0.4 .
author explained the weak dependence between the parameters suggesting that ars emerge slowly and gradually fragment after fragment during a rather long time period ( frequently up to several days ) , and the flux growth rate derived from the first half - day of emergence is not a comprehensive characteristic of an ar as a whole .
@xcite utilized sdo / hmi data for 23 emerging ars to study horizontal divergent flows that are expected to accompany the emergence process according to the `` two - step '' emergence scenario @xcite . in 13 cases ,
these flows were detected 30 - 100 minutes prior to the emergence onset .
authors suggested that the observed consecutive peaks are likely to indicate bifurcation of the emerging magnetic flux , which may consist of separate flux bundles @xcite .
however , these authors did not remove the well known artificial 24-hour oscillations in the magnetic flux ( see , _ e.g. _ , @xcite ) so that the resulting time profiles of the flux growth rate are contaminated by artificial strong peaks separated by 24-hour intervals ( see figures 1 and 7 in @xcite ) making the growth rate measurements uncertain .
nevertheless , the @xcite suggestion is promising , and it should be investigated using improved and more accurate detection routines . the magnetic flux emergence process is usually accompanied by interaction between emerging and pre - existing fluxes .
this topic was scrutinized by @xcite who explored the problem in both theoretical and observational way . as a part of their investigation
, they determined the emergence time for 116 flux emergence cases .
authors considered only the total unsigned flux time variations and did not calculate the flux growth rate .
a two - step emergence behavior , when a gradual emergence stage is followed by rapid emergence , was detected in 54 cases , whereas 46 cases were consistent with a single - phase ( rapid ) emergence process . in the present study ,
we focus on the analysis of the time variations of the flux growth rate .
using sdo / hmi line - of - sight magnetograms , we derive this parameter as a time derivative of the total unsigned flux , which was preliminary filtered to remove the artificial 24-hour oscillations @xcite .
our aim is to reveal how the flux growth rate varies during the rising phase in various active regions and to make an attempt to infer some information about the emerging structure basing on the flux growth rate profile .
for this study , we used line - of - sight magnetic field data taken by the _ helioseismic and magnetic imager _ ( hmi , @xcite ) on board the _ solar dynamics observatory _
( sdo , @xcite ) .
continuous 5 - 7 day long data series for 42 emerging ars were selected for the study .
we chose only isolated ars , emerging amidst the quiet sun .
the list of the ars , the analyzed time interval , the longitude of emergence , and calculated emergence parameters are listed in table [ table1 ] .
we selected only those ars that emerged at heliographic longitudes less than 60 degree ( except a one event ) to avoid errors that can be introduced by the foreshortening effect .
all but one ar ( noaa ar 11678 ) first appeared in the eastern hemisphere which allowed us to track the emergence process during an extended period of time .
the 720-second magnetograms were carefully aligned using the idl fast fourier transform algorithm to remove the solar rotation effect .
the total unsigned flux was calculated from each magnetogram as a sum of absolute flux values measured at each pixel .
only those pixels with the modulus of the flux density twice exceeding the noise level were taken into account .
the standard deviation of noise for hmi magnetograms is about 5 - 6 g @xcite .
the observed total unsigned flux was then corrected for the foreshortening effect by dividing it with the cosine of the heliographic longitude ( see , @xcite ) .
the resulting time profiles of the total unsigned flux for selected ars are shown in figures [ fig1 ] , [ fig2 ] , and [ fig3 ] with green curves .
we observed an active region until the moment when the total unsigned flux saturates , _ i.e. _ , its derivative changes the sign from positive to negative , see figures [ fig1 ] , [ fig2 ] , [ fig3 ] .
we calculated the maximal total unsigned flux , @xmath5 , at the moment when the derivative equals zero .
cccccccc ar & observation & emerg .
& @xmath5 & @xmath15 & @xmath1 & @xmath2 & @xmath4 + no .
& window & long .
& 10 ^ 22^ mx & hours & hours & 10 ^ 20^ mx hour^-1^ & hour^-1^ + 11130 & 2010.11.26 - 12.01 & e10 & 1.27 & 43.1 & 27.4 & 1.78(0.21 ) & 0.014 + 11158 & 2011.02.10 - 02.18 & e22 & 2.49 & 20.8 & 02.2 & 4.97(0.01 ) & 0.020 + 11184 & 2011.04.01 - 04.05 & e16 & 1.07 & 27.6 & 11.3 & 2.42(0.12 ) & 0.023 + 11416 & 2012.02.07 - 02.14 & e48 & 1.86 & 50.6 & 08.2 & 3.30(0.06 ) & 0.018 + 11422 & 2012.02.16 - 02.22 & e22 & 1.39 & 42.3 & 00.1 & 3.60(0.07 ) & 0.026 + 11455 & 2012.04.10 - 04.15 & e31 & 0.80 & 44.8 & 21.8 & 1.16(0.16 ) & 0.015 + 11554 & 2012.08.22 - 08.28 & e43 & 0.80 & 31.0 & 05.0 & 2.12(0.08 ) & 0.027 + 11560 & 2012.08.28 - 09.04 & e51 & 1.61 & 28.5 & 60.6 & 1.63(0.18 ) & 0.010 + 11620 & 2012.11.23 - 11.28 & e01 & 2.13 & 21.6 & 09.4 & 3.58(0.20 ) & 0.017 + 11660 & 2013.01.18 - 01.24 & e14 & 1.46 & 24.4 & 13.1 & 2.04(0.18 ) & 0.014 + 11670 & 2013.02.06 - 02.12 & e50 & 1.01 & 28.7 & 31.8 & 1.14(0.06 ) & 0.011 + 11675 & 2013.02.15 - 02.20 & e57 & 0.58 & 23.0 & 04.3 & 1.70(0.09 ) & 0.029 + 11678 & 2013.02.15 - 02.20 & w23 & 1.90 & 26.3 & 02.0 & 4.38(0.07 ) & 0.023 + 11696 & 2013.03.10 - 03.16 & e48 & 0.98 & 15.4 & 38.9 & 1.37(0.21 ) & 0.014 + 11726 & 2013.04.17 - 04.24 & e25 & 2.62 & 38.6 & 18.3 & 4.25(0.23 ) & 0.016 + 11764 & 2013.06.01 - 06.05 & e04 & 0.61 & 22.9 & 08.4 & 1.86(0.03 ) & 0.030 + 11765 & 2013.06.03 - 06.11 & e39 & 0.98 & 27.9 & 39.9 & 1.29(0.17 ) & 0.013 + 11781 & 2013.06.27 - 07.01 & e21 & 0.84 & 39.7 & 37.7 & 1.00(0.16 ) & 0.012 + 11837 & 2013.08.30 - 09.05 & e02 & 1.04 & 33.5 & 35.1 & 1.23(0.11 ) & 0.012 + 11855 & 2013.09.29 - 10.04 & e38 & 0.90 & 33.7 & 12.9 & 1.83(0.02 ) & 0.020 + 11916 & 2013.12.03 - 12.09 & e18 & 2.25 & 24.8 & 25.0 & 2.23(0.19 ) & 0.010 + 11928 & 2013.12.15 - 12.21 & e35 & 2.12 & 27.1 & 04.8 & 3.57(0.05 ) & 0.017 + 11946 & 2014.01.03 - 01.10 & e50 & 1.58 & 19.1 & 58.5 & 1.68(0.26 ) & 0.011 + 12036 & 2014.04.11 - 04.19 & e55 & 1.79 & 42.3 & 03.6 & 3.21(0.10 ) & 0.018 + 12085 & 2014.06.05 - 06.11 & e41 & 2.35 & 22.4 & 60.9 & 2.41(0.25 ) & 0.010 + 12175 & 2014.09.22 - 09.28 & e16 & 2.78 & 23.8 & 26.7 & 3.38(0.35 ) & 0.012 + 12203 & 2014.10.31 - 11.06 & e22 & 1.06 & 21.5 & 06.6 & 2.86(0.14 ) & 0.027 + 12257 & 2015.01.03 - 01.11 & e28 & 2.00 & 39.5 & 33.3 & 2.63(0.36 ) & 0.013 + 12266 & 2015.01.16 - 01.22 & e37 & 1.15 & 26.6 & 19.8 & 2.17(0.06 ) & 0.019 + 12273 & 2015.01.23 - 01.30 & e22 & 0.51 & 20.2 & 12.6 & 1.49(0.06 ) & 0.029 + 12275 & 2015.01.23 - 01.30 & e00 & 0.77 & 25.6 & 03.6 & 2.34(0.15 ) & 0.030 + 12353 & 2015.05.20 - 05.26 & e24 & 0.51 & 11.9 & 64.5 & 0.56(0.07 ) & 0.011 + 12414 & 2015.09.09 - 09.14 & e01 & 1.25 & 19.5 & 06.7 & 3.80(0.16 ) & 0.030 + 12494 & 2016.02.02 - 02.09 & e36 & 1.14 & 25.8 & 06.6 & 3.12(0.13 ) & 0.027 + 12579 & 2016.08.20 - 08.27 & e37 & 0.71 & 19.3 & 08.3 & 2.55(0.12 ) & 0.036 + 12597 & 2016.09.20 - 09.27 & e29 & 0.63 & 19.6 & 28.3 & 1.03(0.13 ) & 0.016 + + 11682 & 2013.02.24 - 03.02 & e34 & 1.24 & - & - & - & - + 11768 & 2013.06.10 - 06.15 & e14 & 1.50 & - & - & - & - + 11776 & 2013.06.17 - 06.22 & e24 & 0.95 & - & - & - & - + 12003 & 2014.03.08 - 03.13 & e18 & 1.29 & - & - & - & - + 12271 & 2015.01.23 - 01.30 & e43 & 1.13 & - & - & - & - + 12390 & 2015.07.23 - 07.30 & e68 & 1.03 & - & - & - & - + sdo / hmi magnetic data exhibit artificial oscillations with period of 24 hours caused by instrumental effects @xcite .
following the routine suggested by @xcite , we performed low - frequency filtering to remove these artificial oscillations and the resulting profiles , @xmath16 , are shown in figures [ fig1 ] , [ fig2 ] , and [ fig3 ] with red curves .
one can see that artificial flux oscillations are not negligible . these projection - corrected and filtered time profiles of the total unsigned flux , @xmath16 , will be used in our study to calculate the flux growth rate .
for sake of simplicity , further in the text we will refer to it as the total flux , @xmath16 .
the flux growth rate was calculated as a time derivative , @xmath0 , of the total flux , @xmath16 : @xmath17 .
since the time derivative does not depend on any constant component of the flux itself , we ignored the contribution from the quiet sun areas within a magnetogram , contrary to a routine suggested in @xcite .
time variations of the flux growth rate are presented in figures [ fig1 ] , [ fig2 ] , and [ fig3 ] with turquoise curves . having the time profiles of the @xmath0 function during the emergence , for each ar we performed a piecewise continuous linear fitting under the following requirements : the fit should consist of four linear pieces
; the first and third segments are horizontal , whereas the second and the fourth segments may have an arbitrary slope .
the parameters of the linear pieces are calculated under the requirement of the minimum mean square deviations from the observed @xmath0 curve .
the start time was set to the first magnetogram acquisition time , and the end time of the fitting interval in the most of the cases was chosen as the moment where @xmath18 . in cases of complex @xmath0 profiles , the end time was modified in order to preserve the quality of the fit .
the fitting procedure gave us two characteristic time intervals : @xmath15 - the interval of continuously increasing flux increment , and @xmath1 - the interval of quasi - constant flux increment ( see table [ table1 ] , 5^th^ and 6^th^ columns , respectively ) .
the latter is followed by an interval of decreasing flux increment .
we were successful in fitting 36 out of 42 ars , while the remaining 6 ars display a complex @xmath0 profile ( figure [ fig3 ] ) , so that the fitting algorithm failed .
we include these ars in table [ table1 ] , however , no fitting parameters are provided in these cases .
as it follows from figures [ fig1 ] and [ fig2 ] , during the @xmath1-interval , the flux growth rate , @xmath0 , undulates around the flat fitline reaching its maximum at some moment .
the undulations can be rather low ( see , _
e.g. _ , figure [ fig2 ] , ar 12266 ) and high ( _ e.g. _ , ar 11946 in figure [ fig2 ] ) , and the absolute maximum of @xmath0 does not seem to adequately describe the ar emergence process .
instead , we accept the averaged over the @xmath1 value of @xmath0 as a characteristic of the maximum flux growth rate of an ar .
their magnitudes , calculated as @xmath19 ~@xmath1~@xmath20 are listed in the 7^th^ column of table [ table1 ] . as can be seen from table [ table1 ] ,
the @xmath2 varies within a broad range of values and , presumably , depends on the saturated ( maximal ) total unsigned flux , @xmath5 ( 4^th^ column in table [ table1 ] ) , determined at the time when @xmath18 . to compare the flux growth rate in strong and weak active regions
, we calculated a flux growth rate of an ar normalized by the maximal total flux of that ar : @xmath21 .
this parameter listed in the 8^th^ column of table [ table1 ] indicates the fraction of the maximal total flux added to the ar per hour during the @xmath1-stage . * * * * * * * * * * * * * * * *
the @xmath1 and @xmath4 histograms using data from 36 ars are shown in figures [ fig4]a and [ fig4]b , respectively . figure [ fig4]a shows a gap at approximately 13 hours , while in figure [ fig4]b we see a similar separation of the data points into two subsets with @xmath4<0.024 hour^-1^ and @xmath4>0.024 hour^-1^. are these @xmath1 and @xmath4 subsets independent ? to answer the question , we plot @xmath1 versus @xmath4 in two ways . in figure [
fig5]a open circles represent cases with @xmath1<13 hour , whereas in figure [ fig5]b open circles represent cases with @xmath4>0.024 hour^-1^. we see that only for 8 cases the symbol coding has changed to the opposite , while the symbol coding for the rest 28 cases remains unchanged .
this means that for majority of ars the @xmath1 interval and the normalized flux emergence rate , @xmath4 are inversely proportional . for sake of simplicity
further in the text the ars with short @xmath1 and high @xmath4 will be referred to as `` rapid '' emergence events , and , correspondingly , the ars with long @xmath1 and low @xmath4 will be called `` gradual '' emergence events .
there exists a well pronounced statistical dependence between the @xmath5 and the flux growth rate , @xmath2 ( figure [ fig6 ] ) .
when all 36 cases are considered , a linear regression in the double - logarithmic plot with the slope [email protected] is well defined with pearson correlation coefficient of 0.75 . when the `` rapid '' and `` gradual '' emergence cases are considered separately , again
, the two subsets do not overlap for both methods of segregation , see panels a and b in figure [ fig6 ] .
moreover , the linear regressions show different slopes .
thus , the `` gradually '' emerging ars ( filled circles ) display a steep slope @[email protected] for all @xmath1>13 hours events and [email protected] for all the @xmath4<0.024 hour^-1^ events , and both differ significantly from the @xmath22=0.69 slope derived for the entire ensemble . the `` rapidly '' emerging active regions ( open circles ) in both groups
are tightly distributed around the linear ( dashed line ) fit ( see figure [ fig6 ] ) , although the slope varies : @[email protected] for the @xmath1<13 hours events and @[email protected] for the @xmath4>0.024 hour^-1^ events .
further studies with larger statistics are definitely needed to clarify the value of the slope .
an analysis of time series of magnetograms shows that for the `` rapid '' emergence ars coherent appearance of a bundle of elongated mixed - polarity magnetic threads over a fast - expanding area is a characteristic feature ( see figure [ fig7 ] and movies in the online version ) . at the very
beginning of emergence , it is impossible to identify where the leading and following parts will be formed .
an oval - like structure forms by the moment of maximal flux growth rate , @xmath23 ( see bottom panel in figure [ fig7 ] ) . after the peak time
, @xmath23 , the structure continues to emerge , however , in a different way : the leading and following parts are now well formed and separated from each other , and the mixed - polarity filaments / threads ( presumably , `` u '' shaped parts of serpentine field lines ) in the middle of the oval area become less ordered and either disappear at the location , or drift to the leading or following magnetic concentrations .
the entire picture resembles the passage of a coherent flux tube through the photosphere in accordance with the `` two - step '' emergence model ( see conclusions and discussion section ) .
finally , in figure [ fig8 ] we show emergence of an ar where the piecewise fitting failed because of two well pronounced peaks in the @xmath0-profile .
the pairs of numbers ( 1 and 1 ) , ( 2 and 2 ) , _ etc . _ denote five consecutively emerging magnetic dipoles which emerged one after another .
these data suggest that in case of `` gradual '' emergence , we may deal with a series of flux tubes , emerging along the same channel in the cz , and appearing near the same place in the photosphere .
each one produces an @xmath0-profile similar to that in figure [ fig7 ] ( a single - peak profile ) , and their superposition in time results in the variety of @xmath0-profiles shown in figures [ fig2 ] and [ fig3 ] .
in this paper , we analyse the flux growth rate for 42 emerging ars .
the flux growth rate function , @xmath0 , was derived from the total unsigned flux , which was filtered to remove the artificial 24-hour oscillations in hmi data .
this allowed us to obtain a more confident estimate of the @xmath0 than those reported in previous studies ( _ e.g. _ , @xcite ) .
a continuous piecewise linear fit to the @xmath0-profiles allowed us to determine an interval @xmath15 of increasing @xmath0 and an interval @xmath1 of nearly - constant @xmath0 followed by the decrease of @xmath0 .
this fitting routine failed to perform well for 6 ars . during the interval @xmath1 , the @xmath0 function typically exhibited one or several local maxima .
this allowed us to consider the averaged over the @xmath1 growth rate as an estimate of the maximal value of the flux growth rate , @xmath2 .
the 36 successfully fitted events form two subsets ( with an 8 event overlap ) : i ) ars with short ( < 13 hours ) @xmath1 interval and a high ( > 0.024 hour^-1^ ) normalized flux emergence rate , @xmath4 which we call `` rapid''-emergence events and ii ) `` gradual''-emergence events characterized by long ( > 13 hours ) @xmath1 interval and a low @xmath4 ( < 0.024 hour^-1^ ) .
one more evidence for the separation is offered in the diagrams @xmath2 _ versus _
@xmath5 , where the events from different subsets are not overlapped and each subset displays an individual power law .
the power law index for the `` rapid''-emergence events varies in a range of 0.62-0.97 , and in the range of 0.87-0.89 for the `` gradual''-emergence events .
the power law index as derived for the entire ensemble of 36 events equals [email protected] .
we found that the maximal flux growth rate @xmath2 varies in a range of ( 0.5 - 5)@xmath310 ^ 20^ mx hour^-1^ for active regions with the maximal total unsigned flux of ( 0.5 - 3)@xmath310 ^ 22^ mx .
our data points ( see figure [ fig6 ] ) can be considered as an extension ( toward larger fluxes and higher rates ) of the diagram in figure 5a in @xcite , where the flux growth rate varies in the range of ( 10 ^ 18 ^ -10 ^ 20^ ) mx hour^-1^ for regions with the total flux of 10 ^ 18 ^ -0.7@xmath310 ^ 22^ mx
. however , the power index @xmath24 in the relation @xmath25^@xmath24^ is higher ( 0.69 ) in our study as compared to that ( 0.57 ) reported in @xcite
. the reason might be either a difference in the calculation of the flux growth rate , or a real change of the index with the total flux .
for six large active regions studied by @xcite , the flux growth rate chaotically varies in a range of ( 0.4 -1.0)@xmath310 ^ 20^ mx hour^-1^ when the total flux varies in a range of ( 0.8 - 2)@xmath310 ^ 22^ mx .
the reported flux growth rate is lower than our estimates for the same values of the flux , see figure [ fig6 ] .
the reason might be again the calculation method : in @xcite , as well as in @xcite , the flux growth rate was derived by dividing the saturated flux over the entire emergence time , whereas in our study , the time derivative @xmath0 of the total flux was used to calculate the maximal flux growth rate . scrutinized inspection of movies of emergence
allows us to detect a typical difference between the two emergence types .
the `` rapid''-emergence scenario is consistent with a `` two - step '' emergence of a single twisted flux tube @xcite and with magnetic flux `` shredding '' described in @xcite .
frequently , for these active regions , a rapid expansion of an oval - shaped pattern of mixed polarity magnetic threads is observed .
the leading and following sunspots are formed later , approximately near the peak of the flux growth rate , t^*^ time in figure [ fig7 ] . before the t^*^-time ,
the flattened and horizontally expanded ( during the first step of sub - photospheric rising ) flux tube blows up into the photosphere ; by the t^*^-time , the cross - section of the tube already rose above the photosphere , and later the leading and following footpoints of the tube become more vertical ; the flux growth rate diminishes .
it is possible to suggest , at least , two reasons for the decrease of @xmath0 at this stage of emergence .
first , as the apex of the tube rises , the contribution to the total flux from the serpentine field lines diminishes due to cancellation , magnetic diffusion between opposite polarities , retraction of u - loops , _ etc . _ @xcite .
second , the azimuth component @xmath26^@xmath27^ of the magnetic field of the tube predominantly contributes to the line - of - sight flux at the very early stage of emergence , when the tube is nearly horizontal .
as the apex rises and the footpoints become more vertical , @xmath26^@xmath27^ contributes to the transverse flux @xcite .
the events of the `` gradual '' emergence can then be explained as a consecutive rising of several flux tubes at the same location , in a full agreement with the suggestions by @xcite and @xcite .
each tube is then experiencing the `` two - step '' emergence process : indeed , bubbles of mixed polarity serpentine - line features always accompany the consecutive emergence of dipoles of any size .
a question why the `` rapid''-emergence active regions display the higher normalized flux emergence rate ( in other words , they are able to experience the more intensive emergence ) than the `` gradually '' emerging active regions , is still open .
a possible explanation can be offered that the sub - photospheric twisting motions of the roots of emerging flux tube intensify during the emergence and thus generate additional azimuthal component of the magnetic field .
the effect can be more pronounced in the emergence of a single strong flux tube than in an ensemble of rather weak flux tubes . in the more general way , the sub - photospheric turbulent dynamo action ( see , _
e.g. _ , @xcite ) can manifest itself in this way , at the emergence stage of magnetic structures .
another explanation can be suggested as well .
we see that `` rapid''-emergence ars usually exhibit emergence of a flattened single bundle , while `` gradual''-emergence ars are formed by gradual , extended in time appearance of several dipoles .
this observation can be explained by different spatial orientation of magnetic flux bundle before the start of emergence .
a horizontally - oriented flattened subsurface bundle would produce `` rapid '' emergence .
correspondingly , vertically - oriented nearly - flat bundle would result in consecutive appearance of the dipoles , one - by - one , as they rise through the photosphere .
if this explanation is correct it raises another question why different ars are oriented in different way in the upper layer of the photosphere during their pre - emergence stage .
sdo is a mission for nasa s living with a star ( lws ) program .
the sdo / hmi data were provided by the joint science operation center ( jsoc ) .
the reported study was supported by the rfbr research projects 16 - 02 - 00221 a and 16 - 42 - 910493 and the presidium of the russian academy of science program 7 .
vyu acknowledges support from afosr fa9550 - 15 - 1 - 0322 and nsf ags-1250818 grants and korea astronomy and space science institute .
we thank international space science institute for enabling interesting discussions .
we are thankful to anonymous referee for critical comments and suggestions that helped very much to improve the paper .
archontis , v. , moreno - insertis , f. , galsgaard , k. , hood , a. , oshea , e. : 2004 , _ astron .
_ * 426 * , 1047 .
http://dx.doi.org/10.1051/0004-6361:20035934[doi ] http://adsabs.harvard.edu/abs/2004a%26a...426.1047a[ads ] liu , y. , hoeksema , j.t . ,
scherrer , p.h .
, schou , j. , couvidat , s. , bush , r.i . ,
duvall , t.l . ,
hayashi , k. , sun , x. , zhao , x. : 2012 , _ solar phys .
_ * 279 * , 295 .
http://dx.doi.org/10.1007/s11207-012-9976-x[doi ] http://adsabs.harvard.edu/abs/2012soph..279..295l[ads ] scherrer , p.h . ,
schou , j. , bush , r.i . ,
kosovichev , a.g . ,
bogart , r.s .
, hoeksema , j.t . ,
liu , y. , duvall , t.l . ,
zhao , j. , title , a.m. , schrijver , c.j . , tarbell , t.d .
, tomczyk , s. : 2012 , _ solar phys . _ * 275 * , 207 .
http://dx.doi.org/10.1007/s11207-011-9834-2[doi ] http://adsabs.harvard.edu/abs/2012soph..275..207s[ads ] schou , j. , scherrer , p.h . ,
bush , r.i . ,
wachter , r. , couvidat , s. , rabello - soares , m.c . ,
bogart , r.s .
, hoeksema , j.t . , liu , y. , duvall , t.l . , akin , d.j . ,
allard , b.a .
, miles , j.w . , rairden , r. , shine , r.a .
, tarbell , t.d . ,
title , a.m. , wolfson , c.j . , elmore , d.f . , norton , a.a . , tomczyk , s. : 2012 , _ solar phys .
_ * 275 * , 229 .
http://dx.doi.org/10.1007/s11207-011-9842-2[doi ] http://adsabs.harvard.edu/abs/2012soph..275..229s[ads ] smirnova , v. , efremov , v.i . ,
parfinenko , l.d .
, riehokainen , a. , solovev , a.a . : 2013 , _ astron .
astroph . _
* 554 * , a121 .
http://dx.doi.org/10.1051/0004-6361/201220825[doi ] http://adsabs.harvard.edu/abs/2013a%26a...554a.121s[ads ]
van driel - gesztelyi , l. , dmoulin , p. , mandrini , c.h .
, harra , l. , klimchuk , j.a . : 2003 , _ astrophys .
j. _ * 586 * , 579 .
http://dx.doi.org/10.1086/367633[doi ] http://adsabs.harvard.edu/abs/2003apj...586..579v[ads ] | we studied the emergence process of 42 active region ( ars ) by analyzing the time derivative , @xmath0 , of the total unsigned flux .
line - of - sight magnetograms acquired by the _ helioseismic and magnetic imager _ ( hmi ) onboard the _ solar dynamics observatory _ ( sdo ) were used .
a continuous piecewise linear fitting to the @xmath0-profile was applied to detect an interval , @xmath1 , of nearly - constant @xmath0 covering one or several local maxima .
the averaged over @xmath1 magnitude of @xmath0 was accepted as an estimate of the maximal value of the flux growth rate , @xmath2 , which varies in a range of ( 0.5 - 5)@xmath310 ^ 20^ mx hour^-1^ for active regions with the maximal total unsigned flux of ( 0.5 - 3)@xmath310 ^ 22^ mx .
the normalized flux growth rate , @xmath4 , was defined under an assumption that the saturated total unsigned flux , @xmath5 , equals unity . out of 42 ars in our initial list ,
36 event were successfully fitted and they form two subsets ( with a small overlap of 8 events ) : the ars with a short ( < 13 hours ) interval @xmath1 and a high ( > 0.024 hour^-1^ ) normalized flux emergence rate , @xmath4 , form the `` rapid '' emergence event subset .
the second subset consists of `` gradual '' emergence events and it is characterized by a long ( > 13 hours ) interval @xmath1 and a low @xmath4 ( < 0.024 hour^-1^ ) . in diagrams of @xmath2 plotted versus @xmath5 , the events from different subsets are not overlapped and each subset displays an individual power law . the power law index derived from the entire ensemble of 36 events is [email protected] .
the `` rapid '' emergence is consistent with a `` two - step '' emergence process of a single twisted flux tube .
the `` gradual '' emergence is possibly related to a consecutive rising of several flux tubes emerging at nearly the same location in the photosphere . |
the analysis of real lisa data will necessarily involve a detailed modeling of instrument response and noise ; for the purposes of the mldcs , it is desirable to decouple this aspect from the inherent complexity of gw analysis , by distilling the lisa measurements into a standard idealized model . thus , the mldc task force has developed a set of _ pseudo - lisa _ assumptions and conventions , which we lay out in this section ( see also ref .
@xcite ) . both of the lisa response simulators used in challenge 1 ( the lisa simulator @xcite and synthetic lisa @xcite ) comply with these assumptions , and adhere to these conventions . in later challenges , as the craft of lisa data - analysis matures , so will the pseudo - lisa model , becoming increasingly realistic .
the pseudo - lisa orbits are obtained by truncating exact keplerian orbits for a small mass orbiting the sun to first order in the eccentricity ( see the appendix of ref .
@xcite ) . in solar - system barycentric ( ssb ) coordinates ( with the @xmath0 axis aligned with the vernal point ) , we set @xmath1 where @xmath2 ( @xmath3 ) is the relative orbital phase of each spacecraft , @xmath4 au is the semi - major axis of the guiding center , and @xmath5 is its orbital phase . in this approximation , the spacecraft form a rigid equilateral triangle with side length @xmath6 km for @xmath7 .
( in fact , the lisa simulator and synthetic lisa implement @xmath8-accurate orbits , but the additional terms make very little difference to the instrument response . ) the parameters @xmath9 and @xmath10 ( ` initialposition ` and ` initialrotation ` in lisaxml , see sec . [ sec : lisaxml ] ) set the initial location and orientation of the lisa constellation ; in challenge 1 , @xmath11
. this choice places lisa at the vernal point , with spacecraft 1 directly below the guiding center in the southern ecliptic hemisphere .
see ref .
@xcite for expressions to convert to other lisa orbit specifications .
all times are measured by an ideal clock at the ssb .
the basic ( individual - link ) lisa response to gws is taken to be the _ phase response _
@xmath12 used in the lisa simulator and discussed in sec .
ii of ref .
@xcite , or the _ fractional frequency response _ @xmath13 used in synthetic lisa and discussed in sec .
ii b of ref .
( see the tdi rosetta stone @xcite for translations between index notations . ) the phase and fractional - frequency formalisms are equivalent , and are related by a simple time integration .
the former has the advantage of representing more closely the actual output of the lisa phasemeters ; the latter of being directly proportional to ( differences of ) the gravitational strains at the spacecraft .
( in fact the _ lisa simulator
_ produces _ equivalent - strain _ data , with a nominal length of @xmath14 m. to convert equivalent strain to fractional frequency , differentiate and multiply by @xmath15 . )
lisa will employ time - delay interferometry ( tdi ; see refs .
@xcite ) to cancel the otherwise overwhelming laser phase noise .
in essence , tdi observables are constructed from time - delayed linear combinations of individual - link measurements , and they represent _ synthesized _ interferometers where laser phase fluctuations move in closed paths across the lisa arms .
more complicated paths are required to deal with the variations of the armlengths due to the finer details of the lisa orbits , giving rise to the three tdi `` generations . ''
it is expected that high - level lisa data - analysis tasks ( such as those targeted in the challenges ) will be performed directly on tdi observables , and not on the underlying phase measurements .
thus , for the initial challenges we elect to represent the lisa output as _ tdi 1.5 _ observables @xcite , and in particular as the _ unequal - arm michelson _ observables @xmath16 , @xmath17 , and @xmath18 defined in refs .
@xcite . strictly speaking
, tdi 2.0 would be required to cancel laser noise completely in a rotating and flexing lisa array such as our pseudo - lisa ; however , the upgrade from tdi 1.5 to 2.0 changes little in the response to gw signals , but it requires more careful numerical treatments and adds to the complexity of analysis codes .
thus , the initial challenge data sets contain tdi 1.5 observables _ without _ laser noise .
the model of lisa instrument noise adopted in challenge 1 includes only contributions from optical noise ( assumed white in phase , with one - sided spectral density @xmath19 ) , and from acceleration noise ( assumed white in acceleration , but increasing as @xmath20 below @xmath21 hz , with one - sided spectral density @xmath22^{1/2}\ , \mathrm{m}\ , \mathrm{s}^{-2}\ , \mathrm{hz}^{-1/2}$ ] ) , but not from laser phase noise , as discussed above .
the six optical noises and six acceleration noises ( for the two optical benches on each spacecraft ) are modeled as independent gaussian random processes , and are realized in practice with sequences of pseudo - random numbers .
specifically , synthetic lisa generates independent gaussian deviates ( i.e. , white noise ) in the time domain , and then filters them digitally to obtain the desired spectral shape ; the lisa simulator generates independent gaussian deviates in the frequency domain , multiplies them by @xmath23 , and ffts to the time domain .
another source of complexity that we wish to exclude from the initial challenges is the uncertainty about the true shape of the gravitational waveforms that nature will provide to lisa .
however , we can already begin prepare for their detection and analysis , while waiting for theory to provide more and more accurate models , by working with fully known waveforms of comparable structure and increasing complexity .
this section describes the standard simplified waveforms used in challenge 1 to embody the signals emitted by the three kinds of gw sources under consideration : galactic binaries , massive black - hole binaries , and extreme mass - ratio inspirals .
a special care was devoted to choosing standard source parametrizations that could be used by mldc participants to report their analysis results and compare them easily .
.common source parameters .
note that in the initial challenges we do not deal explicitly with the redshifting of sources at cosmological distances ; thus , @xmath24 is a _ luminosity _ distance , and the masses and frequencies of tables [ tab : bbh ] are those measured at the ssb , which are red / blue - shifted by factors @xmath25 w.r.t . to those measured locally near the sources.[tab : common ] [ cols= " < ,
< , < , < " , ]
all mldc training and challenge data sets are distributed from the mldc website @xcite in a standard file format , developed by the mldc task force with the goal of facilitating the use of the data sets on different computing platforms , and of enabling their identification , perusal , tracking , and archival .
the mldc file format is also used internally by the mldc task force within the workflow that leads from the choice of random source parameters to the generation of tdi data sets . * a _ prolog _ including file metadata such as author , generation date ,
the name and version of the computer code used to create the data , and any other relevant comments . *
a _ lisa data _ section describing the model of the lisa orbits used in the simulations ; for the initial challenges , this amounts to the ` initialposition ` and ` initialrotation ` needed to fully specify the pseudo - lisa orbits of sec . [
sec : orbits ] . *
a _ noise data _ section describing the models of the lisa noises used in the simulations ; for the initial challenges , this amounts to the power spectral densities and generation timesteps of the pseudo - random sequences described in sec .
[ sec : noises ] . *
a _ source data _ section describing the gravitational waveform(s ) included in the simulation .
the sources may be specified in terms of the standard source parameters defined in sec .
[ sec : sources ] ; otherwise , they may be represented as explicit time series of the @xmath26 and @xmath27 gravitational strains at the ssb , plus a minimal set of parameters including the source s sky position and the time offset , cadence , and length of the time series . *
a _ tdi data _ section containing one or more time series of tdi observables , assembled from lisa s response to noises and/or sources , as described in sec .
[ sec : tdi ] ; for the initial challenges , the observables of choice are @xmath16 , @xmath17 , and @xmath18 of tdi 1.5 , plus ( trivially ) the ssb time .
the standard names of these observables are ` xp ` , ` yp ` , and ` zp ` for the equivalent - strain version of the data sets ( generated with the lisa simulator ) , and ` xf ` , ` yf ` , and ` zf ` for the fractional - frequency fluctuation version ( generated with synthetic lisa ) . thus , training data sets will be represented by mldc files including all types of sections ; on the other hand , challenge data sets ( which must be `` blind '' ) will omit the source - data section .
different combinations of the sections appear in the intermediate files used in the mldc workflow , shown in fig .
[ fig : workflow ] .
the mldc file format is implemented using xml ( the extensible markup language ) , a simple , flexible text format related to html , and widely used in the exchange , manipulation , and storage of many kinds of data , especially on the world - wide web @xcite .
software libraries to handle xml are readily available for most computer languages .
xml documents consist of a nested hierarchy of _ elements _ , enclosed by opening and closing _ tags _ ( ` < tagname > ` and ` <
/tagname > ` ) , and containing textual data ; elements may also have textual _ attributes _ ( as in ` < tagname attrname = attrvalue > ` ) .
white space and newlines have no meaning within xml ( any sequence of such characters is in fact equivalent to a single white space ) , but they are usually added liberally to help human parsing of the files .
the xml implementation of the mldc file format ( known as lisaxml ) is based on xsil ( the extensible scientific interchange language ) , an xml dialect developed at caltech to represent scientific data in multiple applications @xcite .
xsil is very terse ; its data structures consist of few simple building elements : * ` < xsil name= ... type= ... > ` acts as a hierarchical container .
for instance , the lisa - data section of an mldc file is represented by an ` < xsil type = lisadata > ` element , and the source - data section is represented by a ` < xsil type = sourcedata > ` element containing one or more ` < xsil name= ... type = planewave > ` elements . * ` < param name= ... unit= ... > ` is used to describe parameter values and their units . for gw sources , the ` name ` and ` unit ` attributes are those found in tables [ tab : common][tab : emri ] .
for instance , ` galacticbinary ` may have ` < param name = frequency unit = hertz>1.0e-3</param > ` . * ` < array name= ... type= ... > ` is used to specify arrays of homogeneous data , such as the time series of tdi observables . in lisaxml ,
the actual array elements are stored in separate binary files , but the ` array ` element contains information about the location , organization , and encoding of the binary files .
the advantage of this hybrid xml / binary arrangement is that bulk data are represented in binary format , which can be saved , stored , and loaded very efficiently , while file metadata and the lisa , noise , and source parameters are contained in textual xml files that are easily parseable and editable by humans ( or by powerful xml libraries ) .
in fact , lisaxml files can be viewed in standard - compliant web browsers such as firefox and safari , which will use a set of special lisaxml _ stylesheet transformations _ @xcite to render the metadata and nested parameters structures ( but not the bulk data ! ) as pleasantly formatted tables .
* _ c / c++ . _
the lisa tools subversion archive @xcite includes a directory ` lisaxml / io - c ` , which contains lisaxml input
output routines in c ; tdi and strain time series are returned as simple c arrays , while source and time series parameters are stored in c structures . * _ python .
_ routines to input and output lisaxml are built into the python - steered synthetic lisa @xcite , which can read ( write ) strain and tdi time series into ( from ) standard numpy arrays @xcite .
in addition , the synthetic lisa python / c++ objects that describe the lisa geometry , the lisa noises , and gw sources can also be translated to and from their lisaxml representation . *
_ matlab . _
the program ` xml2matlab.c ` , found in directory ` lisaxml / c - exam``ples ` in the lisa tools subversion archive @xcite , can be compiled into a matlab mex function that will read lisaxml tdi time series into a matlab array . * _ ascii . _
the program ` xml2ascii.c ` , in the same location , can be compiled into a command - line utility that will dump lisaxml tdi time series into a tab - separated ascii file , or to standard output .
all these tools are still in development at the time of writing , but all are already quite functional in reading mldc data sets into analysis applications . for more detailed instructions on reading lisaxml files , see the mldc omnibus documents @xcite and the lisa tools sourceforge website @xcite ; these include also the instructions and scripts necessary to reproduce the mldc workflow and create additional training sets .
m.v.s work was supported by the lisa mission science office and by the human resources development fund at the jet propulsion laboratory , california institute of technology , where it was performed under contract with the national aeronautics and space administration .
99 mock lisa data challenge homepage , astrogravs.nasa.gov/docs/mldc . mock lisa data challenge task force , `` document for challenge 1 , '' svn.sourceforge.net/viewvc/lisatools/docs/challenge1.pdf .
mock lisa data challenge task force wiki , www.tapir.caltech.edu/dokuwiki/listwg1b:home .
n. j. cornish and l. j. rubbo , _ phys .
d _ * 67 * , 022001 ( 2003 ) ; erratum , _ ibid .
_ , 029905 ( 2003 ) ; lisa simulator website , www.physics.montana.edu/lisa ; the canonical mldc releases of synthetic lisa are also available from the sourceforge mldc site @xcite .
m. vallisneri , _ phys .
d _ * 71 * , 022001 ( 2005 ) ; synthetic lisa website , www.vallis.org/syntheticlisa ; the canonical mldc releases of synthetic lisa are also available from the sourceforge mldc site @xcite .
m. vallisneri , the _ tdi rosetta stone _ v. 1/5/2005 , www.vallis.org/tdi .
( tdi 1.0 ) j. w. armstrong , f. b. estabrook , and m. tinto , _ astrophys . j. _ * 527 * , 814 ( 1999 ) ; f. b. estabrook , m. tinto , and j. w. armstrong , _ phys . rev .
d _ * 62 * , 042002 ( 2000 ) .
( tdi 1.5 ) n. j. cornish & r. w. hellings , _ clas .
* 20 * , 4851 ( 2003 ) .
( tdi 2.0 ) d. a. shaddock , m. tinto , f. b. estabrook , and j. w. armstrong , _ phys .
d _ * 68 * , 061303(r ) ( 2003 ) ; m. tinto , f. b. estabrook , and j. w. armstrong , _ phys .
d _ * 69 * , 082001 ( 2004 ) .
l. blanchet , g. faye , b. r. iyer , and b. joguet , _ phys .
* 65 * , 061501 ( 2002 ) ; erratum , _ phys .
* 71 * , 129902(e ) ( 2005 ) .
t. damour , b. r. iyer , and b. s. sathyaprakash , _ phys .
d _ * 63 * , 044023 ( 2001 ) .
a. buonanno , y. chen , and m. vallisneri , _ phys .
d _ * 67 * , 104025 ( 2003 ) ; erratum , _ phys .
d _ * 74 * , 029904(e ) ( 2006 ) .
l. barack and c. cutler , _ phys .
d _ * 69 * , 082005 ( 2004 ) .
p. c. peters and j. mathews , _ phys .
_ * 131 * , 435 ( 1963 ) ; p. c. peters , _ ibid . _ * 136 * , b1224 ( 1964 ) .
world wide web consortium ( w3c ) , `` extensible markup language ( xml ) activity and information , '' www.w3.org/xml .
caltech center for advanced computing research , `` xsil : extensible scientific interchange language , '' www.cacr.caltech.edu/sda/xsil .
world wide web consortium ( w3c ) , `` the extensible stylesheet language family ( xsl ) , '' www.w3.org/style/xsl .
lisa tools sourceforge project , sourceforge.net/projects/lisatools ; the lisa tools subversion code archive is accessible from sourceforge.net/svn/?group_id=161442 ; the subversion version control system is distributed and documented at subversion.tigris.org .
numpy website , numpy.scipy.org . | the lisa international science team working group on data analysis ( list- wg1b ) is sponsoring several rounds of mock data challenges , with the purpose of fostering development of lisa data - analysis capabilities , and of demonstrating technical readiness for the maximum science exploitation of the lisa data .
the first round of challenge data sets were released at this symposium .
we describe the models and conventions ( for lisa and for gravitational - wave sources ) used to prepare the data sets , the file format used to encode them , and the tools and resources available to support challenge participants .
address = gravitational astrophysics laboratory , nasa goddard space flight center , + 8800 greenbelt rd . , greenbelt , md 20771 , usa address = max - planck - institut fr gravitationsphysik ( albert - einstein - institut ) , + am mhlenberg 1 , d-14476 golm bei potsdam , germany address = gravitational astrophysics laboratory , nasa goddard space flight center , + 8800 greenbelt rd . , greenbelt , md 20771 , usa address = center for gravitational wave astronomy , university of texas at brownsville , + brownsville , tx 78520 , usa address = department of physics , montana state university , bozeman , mt 59717 , usa address = jet propulsion laboratory , california institute of technology , pasadena , ca 91109 , usa address = center for gravitational wave physics , 104 davey laboratory , university park , pa 16802 , usa , altaddress = department of physics , weber state university , 2508 university circle ,
ogden , ut 84408 , usa address = school of physics and astronomy , cardiff university , cardiff , cf243yb , uk address = jet propulsion laboratory , california institute of technology , pasadena , ca 91109 , usa address = school of physics and astronomy , university of birmingham , + edgbaston , birmingham b152tt , uk address = department artemis , observatoire de la cte dazur , bp 429 , 06304 nice , france the objectives , structure , and timeline of the mock lisa data challenges ( mldcs ) are discussed in the other contribution in this volume by the mldc task force .
here we concentrate on the technical side of the challenges , and in particular on the theoretical models used to embody lisa and gravitational - wave ( gw ) sources in the first challenge , and on the file format used to distribute the challenge datasets .
more details can be found on the official mldc website @xcite , in the living _ omnibus _ document for challenge 1 @xcite , and on the mldc task force wiki @xcite . |
the seyfert 1 galaxy ngc 4151 is well known to have an extended narrow line region ( enlr ) in the optical ( e.g. , perez et al .
1989 ; robinson et al . 1994 ; crenshaw et al .
2000 ) , indicating interaction between the active galactic nucleus ( agn ) and the host galaxy ( unger et al .
thanks to its proximity ( @xmath9 mpc , @xmath10 pc ; mundell et al .
1999 ) and extensive coverage at all wavelengths ( ulrich 2000 and references therein ) , ngc 4151 thus represents an excellent laboratory to study observationally nuclear feedback .
presence of extended emission in the soft x - rays is also well established ( elvis et al .
1983 ; morse et al . 1995 ;
ogle et al .
2000 ; yang et al . 2001
; wang et al .
2010a,2011a ) .
the nature of the kpc - scale extended x - ray emission remains uncertain ; suggestions include collisionally ionized plasma , photoionized gas , a hybrid of both , or electron - scattered nuclear emission ( elvis et al .
1983 , 1990 ; heckman & balick 1983 ; morse et al .
1995 ; schulz & komossa 1993 ; ogle et al .
2000 ; komossa 2001 ; yang et al . 2001 ) , although electron - scattering is less favored , because the required scattering column and spectral variability do not match the observations ( weaver et al . 1994
; yang et al . 2001 ) .
chandra _ imaging studies of ngc 4151 ( ogle et al .
2000 ; yang et al . 2001 ; gonzalez martin 2008 ; wang et al .
2009 , 2010a ) found good morphological agreements between the extended x - ray emission and the optical forbidden line emission ( in particular [ oiii ] @xmath35007 ) .
such a close spatial correspondence between the soft x - rays and the [ oiii ] emission appears to be common in nearby seyfert 2 galaxies , leading bianchi et al .
( 2006 ) to propose that optical and x - ray features arise from a single photoionized medium .
high spectral resolution grating observations of ngc 4151 with _
chandra _ and _ xmm_-newton further show that the soft x - ray emission is due to both blended emission lines and radiative recombination continua ( rrc ) from he - like and h - like transitions of carbon , oxygen , neon , and nitrogen ( ogle et al .
2000 ; schurch et al .
2004 ; armentrout et al .
the x - ray line ratio diagnostics and electron temperatures of the rrc features support the conclusion that most of the gas is photoionized ( @xmath11 k ) by the agn ( schurch et al .
2004 ; armentrout et al . 2007 ) , although photoexcitation by the agn continuum may also play an important role ( kinkhabwalla et al .
2003 , ogle et al .
2003 ) .
however , these grating studies of ngc 4151 can not provide the spatial distribution of each spectral feature .
the extracted spectra consist of both nuclear emission and extended emission from various physical scales .
therefore , based on the high resolution spectra alone , one can not exclude that the photoionized emission may be associated with the bright narrow - line region ( nlr ) gas clouds close to the nucleus ( @xmath12 pc ) , while a significant contribution from collisionally ionized gas may be present at larger radii ( e.g. , ngc 1365 , wang et al .
2009 , guainazzi et al .
2009 ) . although yang et al .
( 2001 ) presented acis imaging data , the limited signal to noise ratio ( @xmath13 ) of these data prevented detailed high spatial resolution comparisons of optical [ oiii ] and x - ray morphology , or the spectral analysis of x - ray emission from the spatially resolved features . the x - ray emission mechanisms in this prototypical seyfert 1 galaxy deserve further investigation with high quality x - ray data that allow spatially resolved spectral analysis of the circum - nuclear region .
this is the aim of the study presented in this paper , the sixth and last in our series of studies on the nuclear and circum - nuclear region of ngc 4151 using deep _ chandra _
acis and hrc observations ( pi : fabbiano ) . using the hrc data and image deconvolution techniques , wang et al .
( 2009 ) resolved the nuclear emission on spatial scale of @xmath530 pc and made comparisons with _
hst _ observations .
wang et al . (
2010a ) presented the x - ray spectral analysis of the ngc 4151 nuclear emission .
wang et al .
( 2010b ) reported the discovery of faint soft x - ray emission extending 2 kpc out from the active nucleus and filling in the cavity of the h i material .
in paper i of the present series ( wang et al .
2011a ) , based on in - depth analysis of the acis data set , we have examined the kpc scale x - ray morphology and the relations between x - ray absorption features and the cold ism in the host galaxy . in paper ii ( wang et al .
2011b ) we have reported on strong evidence of jet cloud interaction in the nuclear region ( @xmath14 pc ) and compared these findings to our previous hrc study . here
we focus on the comparison between the x - ray line emission of the ionization cone and the ionized gas traced by the optical line emission , and we report on the spectral analysis of spatially resolved x - ray features .
the paper is organized as follows . in 2 , we first briefly describe the data reduction , present the x - ray emission line morphologies ( 2.1 ) and the spectral modeling of all extended emission within a @xmath15-radius ( @xmath16 kpc ) and the spatially resolved emission ( 2.2 ) .
the results are discussed in 3 and summarized in 4 .
ngc 4151 was observed by _
chandra _ for a total of 180 ks ( after screening for high background intervals ) with the spectroscopic array of the advanced ccd imaging spectrometer ( acis - s ; garmire et al .
2003 ) in 1/8 sub - array mode during march 27 - 29 , 2008 .
we have presented the details of our acis observations of ngc 4151 and the data reduction in paper i. the data were analyzed following the standard procedures using ciao ( version 4.2 ) with the caldb 4.2.1 provided by the _
chandra _ x - ray center ( cxc ) .
subpixel event repositioning and subpixel binning techniques ( paper i and references therein ) have been applied to the acis images to improve the spatial resolution .
the complexity in the data analysis caused by the bright nuclear emission was described in papers i and ii . we established that for the @xmath17 x - ray emission ( @xmath18 kev ) , photon pile - up is most severe in the inner @xmath19 , mild at @xmath20 , and not an issue at @xmath21 .
_ chandra _ psf simulations were performed to provide an estimate of the expected contamination from the nuclear emission in an extended feature .
we have taken into account this information in the following analysis .
figure [ define ] ( top ) shows the 0.32 kev acis image ( @xmath50.5 per pixel ) .
bright structured soft x - ray emission along the northeast ( ne ) southwest ( sw ) direction is apparent , together with fainter , less extended emission along the northwest ( nw ) southeast ( se ) direction .
the inner @xmath22 nuclear region , where we identified interactions between the radio jet and the optical clouds , has been discussed in paper ii . here
we focus on the kpc - scale extended emission .
we first extracted counts from every @xmath23 sectors to create azimuthal surface brightness profiles for the extended x - ray emission ( @xmath24 ) , which provides a clear view of its angular distribution , and to identify sectors that contain the bright outflows ( figure [ define ] , bottom ) .
this plot was used to guide further spectral extraction ( see [ spec_analysis ] ) . the extended emission along the nw
se direction is interesting as this is the direction of the putative obscuring torus .
figure [ contour ] demonstrates the clear elongation perpendicular to the bicone ( in particular the nw sector ) , using the 0.32 kev acis image smoothed with the ciao tool @xmath25 .
we extracted the 0.32 kev radial profiles of the nw sector , the bright sw cone , and the faint emission in between ( the `` control region '' ) , shown in figure [ 3profile ] .
the nw emission is on average 5 times fainter than the bright sw cone , but brighter than the control region at @xmath26 ( @xmath27 significance ) .
beyond @xmath28 , it becomes indistinguishable with the faint emission .
the implication on the nuclear obscuring structure will be discussed later in [ evidence ] .
a quick examination of the spectrum of the extended emission extracted from a circular region between a radius of @xmath29 and @xmath15 shows the clear presence of strong blended emission lines ( figure [ conti ] ) , corresponding to the emission lines ( @xmath302 kev ) seen in hetg observations ( ogle et al .
these lines can not be uniquely resolved with the spectral resolution of the acis imaging observation .
most notably the blended lines appear as three strong emission features approximately centered at 0.57 kev ( ovii ) , 0.68 kev ( oviii ) , 0.91 kev ( neix ) and a few weak lines between 12 kev ( mg xi , si k@xmath31 , sixiii ; figure [ conti ] ) .
the blended emission features seen in the acis spectrum and the corresponding hetg lines are summarized in table 1 ( see 2.2.1 ) . following paper ii , we created adaptively smoothed images of the ngc 4151 circum - nuclear region using the subpixel resolution data .
we extracted images from the merged data in three spectral bands dominated by emission lines below 2 kev : 0.30.7 kev ( `` soft band '' containing the ovii and oviii emission ) , 0.71.0 kev ( `` medium band '' containing the neix emission ) , and 12 kev ( `` hard band '' ) . figure [ 3color]a presents a false color composite image of the central @xmath32 ( @xmath53 kpc on a side ) , circum - nuclear region of ngc 4151 , where the soft , medium , and hard band smoothed images are shown in red , green , and blue , respectively .
figure [ 3color]b zooms in to the central 1 kpc - radius region , emphasizing the bicone emission . for visualization of the faint features , the nuclear region ( see paper ii ) is saturated and excluded from this figure . the x - ray biconical morphology has been described in paper i , and here we focus on the spectral differences .
the extended bicone emission ( @xmath33 ) appears rich in 0.31 kev emission ( red and green ) .
similar color emission ( mixed red / green ) is also seen perpendicular to the bicone direction with a hint of harder ( blue ) emission .
clumps of 0.30.7 kev ( red ; ovii and oviii ) emission are also seen , some particularly prominent features are seen in the sw cone @xmath34 and @xmath35 from the nucleus . to further highlight the regions where the strong emission lines arise
, we also extracted the x - ray emission in narrow energy intervals ( 0.530.63 kev ; 0.630.73 kev ; 0.850.95 kev ) to create line strength images .
this is a reasonable approach since the line emission dominates over the weak underlying continuum ( schurch et al .
2004 ) in these narrow bands .
the resulting images are shown in figure [ lines ] .
the position of the nucleus is indicated with a black cross .
ovii and oviii line emission is prevalent in the extended regions .
neix emission , while extended along the general p.a . of the large scale extended emission , is bright in the central @xmath36 but becomes fainter quickly at @xmath37 compared to the ovii emission .
we examined similar emission line maps for the weak lines in the 12 kev range ( nex , neix rrc , mgxi ) and found that they are highly concentrated within the inner @xmath36 region .
this is better demonstrated in figure [ lines]d , where the radial profiles of the line emissions are shown .
spectra and instrument responses were generated using ciao version 4.2 and analyzed with xspec version 12.6 ( arnaud 1996 ) .
background spectra were taken from a source - free region from the same chip .
spectra were grouped to have at least 20 counts per energy bin to allow @xmath38 fitting . unless otherwise noted , we restricted our modeling to photon energies above 0.3 kev , where the acis calibration is good , and below 2 kev , above which the nuclear continuum dominates @xcite . the 90% confidence interval for a single interesting parameter
is reported for all fitting results .
we first characterize the x - ray spectrum of bright extended emission extracted from the @xmath39 circular region , exploring a phenomenological spectral fitting approach that was broadly adopted in previous studies at ccd resolution ( yang et al .
2001 ; see also ogle et al .
2000 , smith & wilson 2001 ) .
the `` continuum '' in the acis spectrum ( a combination of true underlying continuum and highly blended weak emission lines ) was described by a smooth bremsstrahlung emission .
we emphasize that the bremsstrahlung emission here should not be assigned any physical origin , whereas the derived line emission should be considered as a lower limit in such an approach . repeating the technique used to fit the spectrum of the nucleus ( wang et al .
2010 ) , narrow emission lines were then added to the model .
hetg and _ xmm_-newton rgs spectra ( ogle et al . 2000 ; schurch et al .
2004 ; armentrout et al .
2007 ) were used as a guide for the identification of the soft x - ray lines .
the x - ray spectrum of the nucleus of ngc 4151 is notoriously complex and variable .
the hard x - ray continuum is dominated by a power law component suppressed by a partially covering absorber and a compton reflection component ( weaver et al .
1994 ; schurch et al .
2004 ; de rosa et al . 2007 ;
wang et al .
2010b ; lubinski et al . 2010 ) , and the absorption consists of both a mildly - ionized `` warm absorber '' and a cold absorber @xcite .
the psf scattered contribution from the nuclear emission is taken into account in the spectral fitting , using the nuclear component fixed at the best fit nuclear spectral model and normalization @xcite scaled to the expected fraction from psf simulations .
the extraction region contains 4.7% of the 0.32 kev nuclear emission due to psf scattering , corresponding to @xmath525% of the extracted counts .
this can be taken as the upper limit of contamination to the extended line emission , as the line emission dominates the nuclear spectrum .
the identified emission line blends are summarized in table [ tab : extended_flux ] and the spectrum shown in figure [ 4reg2 ] ( see also figure [ conti ] ) .
the absorption column required by the fit is consistent with the line - of - sight galactic column towards ngc 4151 ( @xmath40 @xmath41 ; murphy et al .
the x - ray luminosity of the extended emission ( between 130 pc and 2 kpc from the nucleus ) corrected for the galactic absorption is @xmath42 erg s@xmath1 , compared with @xmath43 erg s@xmath1 for the unabsorbed nuclear source ( wang et al .
2010b ) .
however , such a simplified model does not provide constraints on the emission mechanism , and more physically meaningful spectral models are needed .
fitting the data with combinations of absorbed optically - thin thermal emission with solar abundances ( @xmath44 model ; smith et al .
2001 ) was attempted but gave poor results ( reduced @xmath45 ) .
the abundances were then allowed to vary from solar abundance in the apec model using the variable abundance apec model ( vapec ) , in which @xmath46 , @xmath47 , @xmath48 , and @xmath49 were left free while other elements were fixed at solar abundance .
when @xmath50 is allowed to vary , the fit generally improves but still is far from satisfactory ( @xmath51 ) .
moreover , the required abundances become unphysically low ( @xmath52 ) as the fit attempts to reproduce a smooth continuum . to have a self - consistent model , we have made use of the cloudy photoionization modeling code ( ferland et al .
although the resolution of the acis ccd spectrum prevents us from performing line - based diagnostics , e.g. deriving line ratios of the helium - like triplets , spectral fitting with photoionization models still provides useful constraints about the photoionization status ( as in bianchi et al .
2006 , 2010 ; gonzalez - martin et al .
2010 ) . using cloudy version c08.00 , which enables a cloudy / xspec interface ( porter et al .
2006 ) , we modelled the @xmath53 kev x - ray spectrum assuming an open plane - parallel geometry ( `` slab '' ) .
the dimensionless ionization parameter ( osterbrock & ferland 2006 ) is defined as @xmath54 , where @xmath55 is the hydrogen number density , @xmath56 is the distance to the inner face of a model slab , @xmath57 is the speed of light , and @xmath58 is the emitting rate of hydrogen ionizing photons ( s@xmath1 ) by the ionizing source .
we assumed the broken power - law used by kraemer et al .
( 2005 ) for the agn continuum ( see also armentrout et al . 2007 ) .
we varied @xmath59 and @xmath60 , the column density of the slab to create spectral models covering a grid of parameters ( @xmath61 , @xmath62 @xmath41 ) , which were fed to xspec .
armentrout et al .
( 2007 ) successfully modeled the xmm - newton / rgs soft x - ray spectrum of ngc 4151 with three photoionization components : a high ionization component ( @xmath63 , @xmath64 ) , a medium ionization component ( @xmath65 , @xmath64 ) , and a low ionization component ( @xmath66 , @xmath67 ) . in our initial attempt
, we included all these components , with the ionization parameters and the column densities fixed at these values and the normalizations set free .
in addition , the photoionized emission was absorbed with a line - of - sight column density , @xmath68 .
the fit to the acis spectrum was poor ( @xmath69 ) .
this is not a total surprise , given that these best fit parameters , are optimized for the brightest photoionized material close to the nucleus ( 10@xmath701 pc ; armentrout et al . 2007 ) that dominates the soft x - ray spectrum of ngc 4151 , whereas the emission of interest here arise on the @xmath71130 pc1.5 kpc scale .
when the ( @xmath59 , @xmath60 ) parameters for the photoionized components were set free to vary , we were able to obtain a statistically satisfactory fit ( @xmath72 ) , reproducing the observed acis spectrum of the large scale extended x - ray emission well ( figure [ phot3 ] ) .
the line - of - sight column density required by the fit is @xmath73 @xmath41 , consistent with the galactic column towards ngc 4151 @xcite .
two components , a high ionization phase ( @xmath74 , @xmath75 ) and a low ionization phase ( @xmath76 , @xmath77 ) , are required to produce both the hydrogen - like and helium - like neon and oxygen line emissions .
the cloudy model predicts gas pressures of @xmath78 dynes @xmath41 and @xmath79 dynes @xmath41 for the high @xmath59 component and low @xmath59 component , respectively .
if they are collocated , additional pressure perhaps from a hot intercloud medium in a multiphase ism ( elvis et al .
1983 , ogle et al .
2000)is needed to reach pressure equilibrium .
note that the @xmath76 component closely resembles the `` low '' model component in armentrout et al .
( 2007 ) and the `` d+ea '' x - ray absorption component in kraemer et al .
( 2005 ) .
these model components still left small residuals at @xmath50.5 kev ( the nvii ly@xmath31 ) and 1.02 kev ( the nex ly@xmath31 ) . adding in a third photoionized component
can further improve the fit ( best fit parameters @xmath80 , @xmath81 ) , but the improvement appears not significant when an @xmath82-test is performed ( at @xmath83 significance level ) . thus the two - component model is used as a template for the following analysis of subregions .
given the satisfactory spectral fits with photoionized components and lack of clear signatures of collisional ionization in the enlr , how much thermal emission from hot gas may be present in the circum - nuclear region of ngc 4151 ?
we obtain a constraint on the possible presence of the hot ism by invoking a collisionally ionized component in the best fit photoionization model . after adding a thermal emission component ( @xmath44 ) to the photoionized emission
, the fit is improved with a @xmath84 kev ( at @xmath85 significant level by f - test ) and the residual at 0.91.1 kev is mostly accounted for .
this thermal component may contribute @xmath86 of the total soft x - ray emission , @xmath87 erg s@xmath1 . as a cautionary note , although @xmath44 component improves the fit statistically , it remains possible that the residuals are due to photoexcitation of fe l - shell and the enhanced resonance lines ( e.g. , ogle et al .
2003 ; kinkhabwala et al .
2003 ) that are inadequately accounted for with the cloudy photoionization model components . this will be explored with the high resolution hetg spectrum in future work .
we will discuss the possible presence of the thermal component further .
we emphasize that , because of likely complex geometrical dependencies and absorption between these components in the kpc scale emission , this modeling may still be over - simplistic to simultaneously reproduce the observed features in the broad band . to perform spectral analysis of the extended x - ray emission , the large scale region
was divided into four sectors with position angles ( p.a . ) as outlined in figure [ define ] .
extraction regions are termed the `` sw '' , `` nw '' , `` se '' and `` ne '' regions respectively ( table 2 ; see figure [ define ] ) .
the bisectors of ne and sw correspond to the approximate p.a . of the enlrs ( e.g. , robinson et al . 1994 ) .
they are further divided into subregions along the radial direction ( e.g. , sw 1 , 2 , 3 , and 4 ; ne 1 , 2 , and 3 ) , with @xmath51000 counts ( 0.32 kev ) in the fainter outer regions ( sw4 and ne3 ) .
the nw and se sectors cover regions ( @xmath88 ) perpendicular to the bicone , which is the direction of the large scale weak bar ( asif et al .
1998 ) .
corresponding to the variations in color , spectral differences are clearly visible when the acis spectra across sectors are compared ( figure [ compare4 ] ) .
the fitting results are summarized in table 2 , and figures [ sw ] , [ ne ] , and [ se - nw ] show the corresponding spectra and the fits . we find that : 1 . except for the outer - most regions ( ne3 and sw4 ) , a high ionization component is present in all regions ( @xmath891.26 ) .
lower ionization components are present along the bicone direction ( ne sw ) , with @xmath90 values ranging between -0.5 and 0 .
recall that the fit to the integrated spectrum , which includes emission from all the subregions , finds the @xmath91 and @xmath92 components .
2 . in the ne3 and sw4 regions ,
a very low ionization component ( @xmath93 ) and a high column density ( @xmath94 ) are required by the fit to improve the residuals in the mgi k@xmath31 and sii k@xmath31 .
an f - test indicates that this fluorescent component significantly improve the fit at @xmath95 confidence .
the second photoionized component in the nw se regions , perpendicular to the ionization cone , shows a lower ionization ( @xmath96 ) .
we also extracted spectra of the x - ray emission in the bicone regions that are prominent in 0.30.7 kev emission ( or `` ovii '' rich , visually red in figure [ 3color ] ) and in 0.71 kev emission ( or `` neix '' rich , visually green in figure [ 3color ] ) .
the spectral fits are shown in figure [ red - green]a , b respectively . whereas the fitted ionization parameter for the `` neix '' rich emission is close to the average value , the `` ovii '' rich emission is generally of lower ionization , likely due to higher number density .
this is consistent with the appearance of clumpy material in the optical .
kiloparsec - scale extended x - ray emission is ubiquitously observed in nearby obscured seyfert galaxies with enlrs ( bianchi et al .
2006 ; guainazzi et al . 2009 ; bianchi et al . 2010 ; gonzalez - martin et al .
2010 ) , but the ionization mechanism for this emission has been a subject of active debate . for ngc 4151 , yang et al .
( 2001 ) proposed a two phase model for the extended x - ray emission , including a photoionized warm gas and a hotter collisionally ionized gas ( see also komossa 2001 ) . the relatively low electron temperatures ( @xmath97 k ) inferred from the radiative recombination continua ( rrc ; see liedahl 1999 for review ) and the line ratios from the he - like triplets strongly support the presence of a photoionized component ( ogle et al .
2000 ; schurch et al .
2004 ; kraemer et al .
2005 ) . however , does the observed soft spectrum entirely consist of photoionized emission , or does collisional ionization become more important than the photoionization mechanism beyond the nuclear region ( armentrout et al .
2007 ) ? several of our results point to photoionization as the dominant mechanism . *
_ emission line gas morphologies _
first of all , the morphological agreement between soft x - ray and optical [ oiii ] emission is very strong . in particular
, there is a good spatial correlation between [ oiii ] line emission and the three strong x - ray emission lines ( figures [ lines]a - c ) .
there is clearly an overall morphological coincidence with the ionized gas bicone traced by [ oiii ] emission , as noted in previous x - ray studies ( e.g. , ogle et al .
2000 , yang et al .
2001 ) . a good agreement between the peaks of optical and x - ray line emissions in the inner @xmath98 is found . * _ [ oiii]-to - soft x - ray ratio of the _ hst_-resolved clouds _ bianchi et al .
( 2006 ) studied a sample of eight seyfert 2 galaxies , and suggested that the same kpc - scale gas photoionized by the agn continuum can simultaneously produce the observed x - ray and [ oiii ] emission .
our hrc and acis studies of the inner 2 ( @xmath99 pc ) emission in ngc 4151 ( wang et al . 2009 and paper ii ) have shown that , for most of the clouds close to the nucleus , the ratios of [ oiii ] to soft x - ray emission are consistent with the typical ratios in bianchi et al .
( 2006 ) , the exception being clouds collisionally ionized by the radio jet . +
our subpixel - processed acis images of ngc 4151 enable us to directly compare the extended x - ray emission and those of the resolved [ oiii ] clouds ( figure [ lines ] ) .
the main cloud features are labeled in figure [ oiii_cloud ] . using the calibrated continuum - subtracted hst / wfpc2 f502n image ( kaiser et al .
2000 ) , we measured the [ oiii ] fluxes for the clouds and listed them in table [ fluxratio ] .
acis spectra were extracted from the same regions and fitted with the continuum plus emission line model to obtain their 0.52 kev x - ray fluxes .
figure [ ratio ] shows the [ oiii ] to soft x - ray ratio for the distinct cloud features at various radii to the nucleus ( @xmath5150 pc1500 pc ) .
the ratios listed in table [ fluxratio ] are consistent with a constant [ oiii]/x - ray flux ratio of @xmath515 for the 1.5 kpc range spanned by these measurements .
the radially constant [ oiii]/x - ray ratios indicate that the relative contributions of the low- and high - ionization gas phases at different radii are similar , as pointed out by our referee . if the [ oiii ] and x - ray emission arise from a single photoionized medium , as discussed in bianchi et al .
( 2006 ) , this further implies that the ionization parameter does not vary significantly to a large radial distance and the density is approximately decreasing as @xmath100 , as expected for a freely expanding nuclear wind based on mass conservation ( kraemer et al .
2000 ; bianchi et al . 2006 ) .
* _ spectral fitting _ the spectral analysis shows that , a model with just two photoionized components of high- and low - ionization parameters can describe well the observed x - ray spectra in various regions across the circum - nuclear area , whereas in the spectral fits that invoke only collisionally ionized components strong emission line residuals persist .
this supports the dominant role of nuclear photoionization . *
_ kinematics of the ionized gas _ the full velocity field of the nlr for ngc 4151 was mapped by kaiser et al .
( 2000 ) , which supports a hollow biconical outflow model for these clouds . at distances exceeding 4 ( @xmath5260 pc ) from the nucleus , kaiser et al .
( 2000 ) find that the gas cloud kinematics are consistent with circular rotation in a gravitational field , identical to the host ism .
therefore , except for a few clouds in the nuclear region that appear disturbed by the impact of the radio jet ( * ? ? ?
* ; * ? ? ?
* see also paper ii ) , there is no evidence for strong kinematic shocks or any strong shocks in the outflow , which is consistent with the gas being photoionized by the nuclear radiation .
based on all the above considerations , we conclude that photoionization by the nucleus is most likely the dominant ionization mechanism in the enlr of ngc 4151 , confirming the suggestions of previous studies ( ogle et al .
2000 ; schurch et al .
2004 ; kraemer et al . 2005
; armentrout et al . 2007 ; wang et al .
it is reassuring to establish a role for nuclear photoionization in the extended nlrs that is consistent with the grating studies .
however , there is also evidence that points to a more complex emission mechanism than a single photoionized medium ( e.g. , bianchi et al .
we discuss these results and their implications below in sections 3.2 through 3.4 . in [ sec - large_scale ] , we estimated the extent of thermal emission from hot gas ( @xmath101 erg s@xmath1 ) that could be present in the circum - nuclear region of ngc 4151 .
chandra _ grating study of ngc 4151 by ogle et al .
( 2000 ) provided the first direct evidence of x - ray line emission from a collisionally ionized hot plasma .
armentrout et al .
( 2007 ) also noted that the poor fit to the ovii line profile could be due to a non - photoionized component .
wang et al .
( 2011b ) further show that a @xmath102 kev thermal emission component is present in the inner @xmath36 region , which likely originates from jet - cloud interaction .
although star formation activity in the circum - nuclear region is at a low level ( asif et al .
2005 ; dumas et al .
2010 ) , hot gas may originate from the ngc 4151 galactic disk , since its temperature ( @xmath103 kev ) and luminosity ( @xmath104 ) are comparable to the disk diffuse emission in some star forming galaxies , @xmath102 kev , @xmath105 erg s@xmath1 ( e.g. , ngc 253 ; strickland et al . 2002 ) .
another plausible alternative is that the hot plasma is the intercloud medium uniformly filling in the outflow ( i.e. , filling factor @xmath106 ) , providing pressure confinement for the nlr clouds ( elvis et al .
1983 ; cf .
krongold et al .
2007 for warm absorber outflows ) .
the volume emission measure for the thermal component is @xmath107 cm@xmath70 , derived from the normalization after fitting an additional thermal component ( apec ) to the photoionization models of the bright biconical emission ( 2.2.1 ) . assuming the hollow bi - conical nlr geometry ( crenshaw et al .
2000 ; das et al . 2005 ) , we obtain @xmath109 cm@xmath70 and so an average density of @xmath110 cm@xmath70 .
thus the thermal pressure of the @xmath111 k gas is @xmath112 dyne @xmath41 . comparing with the thermal pressure of @xmath11 k and @xmath113 cm@xmath70 nlr clouds ( robinson et al .
1994 ) gives @xmath114 dyne @xmath41 .
thus pressure equilibrium between the hot and cool phase is highly likely , confirming the findings on pressure confinement in ogle et al .
( 2000 ) .
figure [ lines]a clearly shows that one distinct ovii emission blob `` a '' ( at @xmath115 , @xmath5390 pc sw from the nucleus ) does not align with a bright [ oiii ] emitting cloud , designated as `` k1 '' in robinson et al .
( 1994 ) and offset by @xmath116 to the se of `` a '' ( see yang et al .
we find this is also the case for a more distant x - ray blob `` b '' ( @xmath117 sw ) where the closest [ oiii ] emission cloud is 2 to the sw .
similar displacements between the x - ray enhancement and the [ oiii ] blob are also clearly found in the oviii line emission ( figure [ lines]b ) .
note that the _ chandra _ astrometry is accurate to 0.3(paper i ) , therefore the mis - alignment between features at this spatial scale is significant and indicates that the origin of the optical / x - ray emission needs further investigation .
the radial location of x - ray blob``a '' lies at the boundary between nlr and enlr identified in the optical ( robinson et al .
1994 ) , where the density drops rapidly from @xmath118 cm@xmath70 to @xmath119 cm@xmath70 @xcite . we can estimate the ionization parameter of k1 . from the nuclear spectrum @xcite
, we obtain an ionizing luminosity @xmath120 erg s@xmath1 . assuming a number density of 200 cm@xmath70 ( penston et al .
1990 ; robinson et al .
1994 ) and a radial distance of @xmath121 cm to the nucleus , we obtain an ionization parameter @xmath122 . this is lower than , but in agreement with , the value of @xmath123 derived from the [ oiii]/[oii ] ratio in robinson et al .
( 1994 ) , supporting the photoionization assumption .
since the x - ray blob and the [ oiii ] cloud are located at the same radial distance , their spatial offset suggests that the x - ray blob is in a higher ionization phase than the [ oiii]-bright cloud .
a likely explanation is that the x - ray blob has a lower density in the clump .
if their densities are similar , the difference might arise from a higher ionizing flux received at the surface of the x - ray blob , perhaps due to less screening than in the direction towards the [ oiii ] cloud .
our spectral fitting results show that a range of ionization states are needed to obtain good fits , in agreement with the multiple ionization parameters required in the grating spectra ( e.g. , armentrout et al .
2007 ) . a high ionization ( @xmath124 ) component is found in the spectral fitting for all regions surrounding the nucleus , including the nw and se sectors perpendicular to the ionization cone ( table 2 ) .
a low ionization ( @xmath125 ) component is also present in the nw
se direction that has lower @xmath59 than found in the cone direction ( ne sw ) .
this strongly indicates that there is no continuous absorbing `` torus '' blocking all the ionizing photons .
this hypothesis is supported by the agreement between the observed h@xmath126 line flux in the `` torus '' direction and the value predicted from x - ray - irradiation by the active nucleus ( paper i ) , and by the measurement of hi absorption in the nuclear region , where high levels of clumping on the smallest scales is found ( mundell et al .
we note that a `` patchy torus '' or clumpy wind arising from the accretion disk in which the clouds are optically thick ( e.g. , * ? ? ?
* ) , may allow leakage of nuclear continuum for such a high @xmath59 component .
however , it can not reproduce the @xmath125 component in the directions where the continuum is blocked .
extended emission along the `` torus '' direction is clearly present in the large scale x - ray images shown in figures [ contour ] ( indicated by the arrows ) and [ 3color]a .
the base of this emission is at the location of the nucleus , thus may indicate ionization by the leaked photons in this direction .
moreover , we found an intriguing coincidence between radial regions of enhanced x - ray surface brightness and the p.a .
of outflowing optical clouds identified by the _
hst _ study ( das et al .
they pointed out that these faint clouds flowing close to the plane of the torus would need to be ionized by leaked nuclear radiation , which is attested by our x - ray data .
the low @xmath59 suggests a lower ionizing flux received by some clouds , likely due to nuclear continuum filtered by `` warm absorbers '' ( e.g. , * ? ? ? * ) or the cone walls ( e.g. , * ? ? ?
evaluating the difference between the ionization parameters in the direction of torus and in the bicone , we estimate that approximately 1% of the total ionizing flux is seen by the nw se clouds . previous work on the nlr kinematics of ngc 4151 have established that the bi - cone geometry in ngc 4151 results in one side of the conical outflow exiting the galactic plane at a small angle , partly intersecting the host disk ( e.g. , robinson et al .
1994 ; crenshaw et al . 2000
; das et al . 2005 ; storchi - bergmann et al . 2009 ) .
thus part of the host ism is likely illuminated by the agn radiation .
our x - ray work finds that multiple phases of photoionized gas are present in the enlr of ngc 4151 .
the fitting results for the `` ovii '' rich ( red in figure [ 3color ] ) and `` neix '' rich ( green in figure [ 3color ] ) regions indicate that there is a range of ionizations , with a lower ionization phase associated with the clumpy clouds and the higher ionization with the diffuse outflow .
there is a lowest ionization ( @xmath127 ) component at larger radii in the sw and ne sectors .
this component may come from high density material in the host disk plane , perhaps the nuclear spirals , illuminated by the nuclear radiation .
we further note that interactions between the biconical outflow and the host galaxy ism , similar to the radio jet cloud interaction in the nuclear region , may contribute to the heating of any hot gas present in the host disk .
currently there is no evidence in the gas kinematics that supports any strong interaction ( kaiser et al .
however , as we discuss below , there is some morphological evidence that , at the largest radii , the outflow runs into a clump in the inhomogeneous ism and perhaps results in local shock heating of the gas .
figure [ morph ] gives a large scale ( 1 arcmin@xmath1281 arcmin ) view of the key ism components in the circum - nuclear region of ngc 4151 , illustrating the directions of the kinematic major axis of the host galaxy ( green solid line ;
p.a.@xmath522@xmath129 , pedlar et al .
1992 ; mundell et al .
1999 ) , the large scale `` weak fat bar '' ( green dotted line ; p.a.@xmath5130@xmath129 ; mundell & shone 1999 ) , the enlr bicone ( cyan line ; p.a.@xmath565@xmath129 , evans et al .
1993 ) , and the distribution of hi ( mundell et al . 1999 ) and co ( dumas et al .
2010 ) gas , respectively .
in paper i we noted that the ne soft x - ray emission reaches as far as 12 ( 780 pc ) from the nucleus , but that the surface brightness then decreases significantly by a factor of 5 . comparing the continuum - subtracted h@xmath31 emission ( knapen et al .
2004 ) with the extended soft ( 0.31 kev ) x - ray emission ( figure [ fig8]a ) , the ionized gas traced by bright h@xmath31 emission is mainly located in the @xmath130 biconical region along the ne sw direction centered on the nucleus , which also closely follows the soft x - ray emission . an arc - like feature in the h@xmath31 image ( @xmath131 ) is clearly present at the @xmath132 location ( figure [ morph ] ) at the edge of the x - ray emission , and the northern co gas lane is also found here ( figure [ fig8]b , and dumas et al .
2010 ) .
figure [ fig8]b provides a multiwavelength , enlarged view of this region , showing the spatial relations between the soft x - ray emission , the h@xmath31 , the co gas lanes , and the hi distribution @xcite .
the termination of the ne x - ray emission appears to be closely associated with the presence of the dense cold neutral material ( traced by the co and hi emission ) in the galactic disk .
no similar feature is seen to the sw of the nucleus .
the origin of the h@xmath31 arc ( figure [ fig8]a ) at the terminus of the ne x - ray emission is intriguing .
it is spatially very close to the northern co gas lane and is one of the dust arcs identified in the optical @xmath133 color map @xcite , which are gaseous compressions driven by the bar potential @xcite . in this context
, the h@xmath31 arc may be a more clumpy continuation of the dense co arc and photoionized by the agn radiation .
the rate of ionizing photons from the agn is approximately @xmath134 s@xmath1 ( kraemer et al . 2005 ; c.f . @xmath120
erg s@xmath1 , wang et al .
2010b ) . assuming that the arc has a line - of - sight depth that is comparable to its projected length ( @xmath51 kpc ) , the arc s intercepted ionizing photon rate is @xmath135 s@xmath1 at its distance to the nucleus ( 970 pc ) , a covering factor of almost @xmath136 .
this photon budget should be treated as an upper limit as some fraction of the ionization photons will be absorbed or scattered on the way to the arc . if we assume that the gas is in photoionization equilibrium , for the case b effective recombination rate of h@xmath31 @xcite , a rate of @xmath137 s@xmath1 ionizing photons s@xmath1 is needed to produce the observed h@xmath31 luminosity of the arc ( @xmath138 erg s@xmath1 ) , which is only 1% of the maximal rate of available ionizing radiation in the direction of the arc ( for a covering factor @xmath139 ) .
the scenario where part of the co gas lane is photoionized by the agn and produces the h@xmath31 arc seems plausible with a modest covering factor of @xmath140 , implying that the arc has a filamentary structure with a thickness of a few pc . on the other hand
, we consider the possibility that the arc could be a bow shock feature from interaction between the biconical outflow and dense gas in the host disk piled in the co gas lanes by the large scale stellar bar . assuming that the x - ray emission is due to shock heating as the outflow encounters the dense co lane , and that the shock is strong and adiabatic
, we estimate the shock velocity @xmath141 , adopting a postshock temperature @xmath142 ( lehnert et al .
1999 ) , where @xmath143 is the mass per particle ( co and hi ) and @xmath144 the boltzmann constant . for the @xmath145 kev x - ray temperature measured at the arc , the required @xmath141 is @xmath5150 km s@xmath1 , which is much higher than the local sound speed @xmath146 km s@xmath1 in the @xmath147 k h@xmath31-emitting gas .
although there is no direct measurement at the exact position of the h@xmath31 arc , the measured outflow velocity in [ oiii ] at @xmath148 ne of the nucleus is fully consistent with this value of @xmath141 , e.g. , cloud#28 which has @xmath149 km s@xmath1 ( kaiser et al .
2000 ) . for two of the brighter ,
spatially resolved [ oiii]-emitting clouds seen in the @xmath150 images ( clouds # 1 and # 2 in figure [ oiii_cloud ] ; see also figure 4 in kaiser et al .
2000 ) , we were able to constrain the number density for the x - ray emitting gas with cloudy modeling of their x - ray spectra when an inner radius @xmath151 pc to the nucleus and an ionizing photon rate @xmath152 are set .
the best - fit density is @xmath153 cm@xmath70 .
we use this number density to estimate the mass loss rate and the kinematic power of the hot phase outflow , which are key quantities that gauge the impact of agn outflow on the host galaxy ism . following barbosa et al .
( 2009 ) and @xcite , we calculate the mass loss rate using @xmath154 = @xmath155 , where @xmath156 is the radial velocity ( taken as the approximate outflow velocity ) , @xmath157 the proton mass , @xmath158 the gas filling factor , and @xmath159 the area in the cross section of the outflow . for a radial distance of @xmath160
cm ( 130 pc ) and the hollow cone geometry of das et al .
( 2005 ) , we obtain @xmath161 @xmath162 . adopting @xmath163 @xcite and @xmath164 km s@xmath1 ( crenshaw et al . 2000 ; storchi - bergmann et al . 2010 ) , we obtain a mass outflow rate @xmath165 yr@xmath1 and a kinetic luminosity for the outflow @xmath166 ergs s@xmath1 , or @xmath167 of the bolometric luminosity of the agn ( @xmath168erg s@xmath1 ; kaspi et al .
2005 ) .
the @xmath154 derived from the hot phase outflow is comparable to the value derived from nir studies of ionized gas in ngc 4151 @xcite , which estimated mass outflow rate @xmath1691.2 @xmath6 yr@xmath1 along each cone .
the @xmath154 value is over 10 times higher than the outflow rate derived from @xmath170 and optical spectra ( 0.16 @xmath6 yr@xmath1 ; crenshaw & kraemer 2007 ) , but the latter is measured much closer to the nucleus ( at 0.1 pc ) .
the optical emitting gas likely has a much smaller filling factor than the x - ray emitting gas , which explains the difference in the measurements .
if our study measured an outer part of the same outflow , it could have accelerated to a higher velocity or have loaded ism between 0.1 pc and 100 pc , hence a higher @xmath154 .
how does @xmath154 compare to the accretion rate of the smbh in ngc 4151 ?
the bolometric luminosity of ngc 4151 , @xmath171 erg s@xmath1 ( kaspi et al . 2005 ) , is about 1.2% of its eddington luminosity , which is @xmath172 erg s@xmath1 for the black hole mass @xmath173m@xmath174 of ngc 4151 ( bentz et al .
2006 ) . adopting @xmath175 and @xmath176 for the efficiency of the accretion disk ( shakura & sunyaev 1973 )
, this implies that the central engine of ngc 4151 is accreting at @xmath177 yr@xmath1 .
since the measured outflow rate is 160 times the accretion rate necessary to feed the active nucleus , there must be either significant entrainment of the host galaxy s ism while the nuclear outflow is expanding , or part of the gas fueling the nucleus is ejected before radiating , or a high fraction ( over 99% ) of the x - ray gas does not take part in the outflow and is illuminated by the agn `` in situ '' .
however , the latter seems unlikely as the grating spectra strongly indicate that the emission lines have outflow velocities of 200300 km s@xmath1 ( e.g. , armentrout et al . 2007 ) .
significant mass loading of the outflow by the host ism may be most plausible and common , since previous estimated nlr outflow rates in seyferts ( @xmath50.110 @xmath6 yr@xmath1 ) generally exceed tenfold the accretion rate ( veilleux et al .
2005 ) .
nevertheless , even with such a high @xmath154 in ngc 4151 , the kinematic power @xmath178 suggests that only 0.3% of the available accretion power is extracted to drive the outflow , similar to the findings of @xcite and holt et al .
( 2006 , 2011 ) . for a sample of seyfert nlr outflows
( ngc 4151 not included ) studied in the [ siii]@xmath179 emission , barbosa et al .
( 2009 ) reported a lower @xmath180 .
as pointed out in mathur et al .
( 2009 ) , this poses a problem to the majority of quasar feedback models which require a higher fraction of the accretion power of the black hole ( @xmath55% of @xmath181 to unity ) to be thermally coupled to the host ism .
our @xmath182 is in rough agreement with a two - stage feedback model recently proposed by hopkins & elvis ( 2010 ) , which requires only 0.5% of the radiated energy to drive the initial hot outflow to efficiently shut down star formation in the host .
in this paper we present spectral analysis and emission line images from deep _ chandra _ observation of ngc 4151 , aiming to resolve and characterise the x - ray emission in the nlr .
the findings are summarized as follows : * the soft x - ray emission line images of ngc 4151 ( ovii , oviii , and neix ) are clearly extended and show remarkable morphological coincidence with the biconical nlr mapped by the [ oiii ] emission , which supports a common emission mechanism for the hot- and cool - phase of the nlr gas . * extended emission in the x - ray image is detected along the nw
se sectors , which is the direction of a putative torus .
this may explain the faint rogue clouds identified in previous hst studies that require ionization in this direction , indicating leakage of nuclear ionization instead of full blocking by a continuous obscuring torus . * spectral models involving smooth continua ( a bremsstrahlung plus a power law ) with emission lines provide good descriptions of the spectra .
the emission lines can not be uniquely identified with the present spectral resolution , but are consistent with the brighter lines seen in the _ chandra _ hetgs and xmm - newton rgs spectra below 2 kev .
the absorption corrected x - ray luminosity of the extended emission between @xmath183 pc and @xmath184 kpc is @xmath42 erg s@xmath1 . *
photoionization models successfully reproduce the soft x - ray emission , supporting the dominant role of nuclear photoionization .
there are considerable variations in ionization states across the circum - nuclear region .
a high ionization ( @xmath185 ) component is present in most regions .
a low ionization ( @xmath92 ) component is present along the bicone direction ( ne sw ) , whereas a lower ionization ( @xmath125 ) component is found in the nw
se direction , which is consistent with filtered nuclear emission by warm absorbers instead of a continuous absorbing torus .
the lowest ionization component ( @xmath127 ) appears to be associated with the dense gas in the host plane .
a thermal component may still be present at @xmath86 of the total soft emission , perhaps related to hot ism in the galactic disk or shocks associated with the outflow .
* the measured ratios of [ oiii]/soft x - ray flux are consistent with a constant ratio of @xmath515 for the 1.5 kpc radius spanned by these measurements .
this suggests a similar relative contributions from the low- and high - ionization gas phases at different radii .
if the [ oiii ] and x - ray emission arise from a single photoionized medium , this further implies an outflow with a wind - like density profile ( @xmath186 ) . *
the estimated mass outflow rate in ngc 4151 is @xmath52@xmath6 yr@xmath1 at 130 pc and the kinematic power of the ionized outflow is @xmath187 erg s@xmath1 , 0.3% @xmath181 .
this value is significantly lower than the expected efficiency in the majority of quasar feedback models , but comparable to the two - stage model described in @xcite . placing all our findings in the context of previous studies , we obtain a comprehensive view of the ngc 4151 circum - nuclear region at various spatial scales , from the inner - most @xmath188 pc ( as illustrated in the schematic drawing in figure [ cartoon]a ) to as far as @xmath189 kpc ( figure [ cartoon]b ) .
the key points are recapped here .
photoionization by the nucleus is important at all scales . in the nuclear region , except for a few clouds that show interaction with the radio jet ( wang et al .
2009 , 2011b ) , most of the x - ray nlr clouds are consistent with being part of a photoionized biconical outflow , with a @xmath190 density profile expected for a nuclear wind . in the enlr of ngc 4151 ,
photoionization is still the dominant ionization mechanism for the observed x - ray emission ( e.g. , ogle et al .
2000 ; kraemer et al . 2005 ) , but a wide range of ionization states is present .
this is likely related to the nature of the enlr .
the bi - cone geometry indicates that one side of the conical outflow exits the galactic plane at a small angle , partly intersecting the host disk ( e.g. , crenshaw et al .
2000 ; das et al . 2005
; storchi - bergmann et al . 2010 ) . optical studies ( e.g. , robinson et al .
1994 ; kaiser et al .
2000 ) have identified such a boundary between nlr and enlr at @xmath191 ( sw ) , and the ionized gas kinematics become consistent with the rotation velocity of the galactic disk at @xmath192 .
thus the origin of the enlr is best explained as the inhomogeneous ism in the galactic plane ionized by the agn .
part of the high density nuclear spirals illuminated in the path of the bicone can produce the curvy shape of x - ray enhancement and the associated low ionization .
the presence of ionized emission perpendicular to the bicone indicates leaked nuclear emission along the putative torus direction , likely filtered nuclear emission by warm absorbers . in the large 3 kpc - scale cavity of the h i material ,
we find faint soft diffuse x - ray emission that provides evidence for agn
host interaction in ngc 4151 ( wang et al .
2010a ) , originated from either hot gas heated by the nuclear outflow or photoionized gas from a recent nuclear outburst . as a concluding remark ,
our findings in ngc 4151 demonstrate that abundant information can be extracted from x - ray spectral imaging studies when supplemented with multiwavelength data .
many valuable new radio , optical and ir observations and modeling of other nearby seyfert galaxies are emerging ( crenshaw et al .
2010 ; riffel et al . 2010 ; fisher et al . 2011 ,
schnorr mller et al .
2011 ) , and progress in the x - rays are being made to further our understanding of agn feeding and feedback processes in these galaxies .
we thank the anonymous referees for providing us with detailed and constructive comments that have improved the clarity of this manuscript .
this work is supported by nasa grant go8 - 9101x and go1 - 12009x .
we acknowledge support from the cxc , which is operated by the smithsonian astrophysical observatory ( sao ) for and on behalf of nasa under contract nas8 - 03060 .
cgm acknowledges financial support from the royal society and research councils u.k .
gd was supported by dfg grants sch 536/4 - 1 and sch 536/4 - 2 as part of spp 1177 .
j. w. thanks t. kallman , g. ferland , s. bianchi , a. marinucci , and s. chakravorty for advice on photoionization modeling .
this research has made use of data obtained from the _ chandra _ data archive , and software provided by the cxc in the application packages ciao and sherpa . some of the data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
cccccc ovii & & 0.57 & ... & @xmath193 & 13% + ... & ovii f & 0.5617 & 0.5611 & @xmath194 & + ... & ovii i & 0.5691 & 0.5687 & @xmath195 & + ... & ovii r & 0.5745 & 0.5740 & @xmath196 & + + oviii & & 0.66 & ... & @xmath197 & 13% + ... & oviii ly@xmath31 & 0.6541 & 0.6537 & @xmath198 & + ... & nvii rrc & 0.67 & ... & @xmath199 & + + ovii & & 0.73 & ... & @xmath200 & 12% + ... & ovii rrc & 0.74 & ... & @xmath201 & + ... & oviii ly@xmath202 & 0.7749 & 0.7747 & @xmath203 & + + oviii & & 0.82 & ... & @xmath204 & 2% + ... & oviii ly@xmath205 & 0.8175 & 0.8171 & @xmath206 & + ... & oviii rrc & 0.86 & ... & @xmath206 & + + neix & & 0.91 & ... & @xmath207 & 21% + ... & neix f & 0.9064 & 0.9052 & @xmath208 & + ... & neix i & 0.9172 & 0.9151 & @xmath209 & + ... & neix r & 0.9229 & 0.9221 & @xmath210 & + + nex & & 1.03 & ... & @xmath204 & 17% + ... & nex ly@xmath31 & 1.0224 & 1.0219 & @xmath210 & + ... & neix @xmath211 & 1.0749 & 1.0738 & @xmath212 & + + nex & & 1.20 & ... & @xmath213 & 12% + ... & neix rrc & 1.20 & ... & @xmath214 & + ... & nex ly@xmath202 & 1.2118 & 1.2110 & @xmath215 & + + mgxi & & 1.30 & ... & @xmath213 & 9% + ... & mgxi f & 1.3542 & 1.3313 & @xmath216 & + ... & mgxi r & 1.3525 & 1.3524 & @xmath214 & + + sii & & 1.75 & ... & @xmath217 & 9% + ... & sii k@xmath31 & 1.7413 & 1.7391 & @xmath207 & + ... & mgxii ly@xmath202 & 1.7472 & 1.7450 & @xmath218 & + ne 1 & 2432 & 0.04@xmath219 & 0.86 & 19.7 & -0.22 & 19.7 & @xmath220 & 63/65 & @xmath221 + .... 2 & 1814 & 0.07@xmath222 & 1.05 & 19.8 & 0.06 & 20.1 & @xmath223 & 60/53 & @xmath224 + .... 3 & 913 & 0.03@xmath225 & 0.49 & 19.9 & -1.87 & 23.5 & @xmath226 & 26/29 & @xmath227 + sw 1 & 2096 & 0.02 & 0.83 & 20.0 & -0.38 & 19.4 & @xmath228 & 60/57 & @xmath229 + .... 2 & 1667 & 0.02 & 1.06 & 20.0 & -0.42 & 19.8 & @xmath230 & 58/45 & @xmath231 + .... 3 & 1560 & 0.02 & 1.00 & 19.6 & -0.50 & 19.6 & @xmath232 & 54/46 & @xmath233 + .... 4 & 966 & 0.02 & 0.34 & 19.0 & -1.28 & 23.5 & @xmath234 & 34/30 & @xmath235 + nw & 1050 & 0.22@xmath219 & 1.03 & 19.2 & -0.98 & 19.5 & @xmath236 & 35/35 & @xmath237 + se & 1081 & 0.16@xmath238 & 1.26 & 20.3 & -1.16 & 19.9 & @xmath239 & 28/33 & @xmath240 + ovii - rich & 2626 & 0.02@xmath219 & 0.14 & 20.3 & -1.22 & 20.3 & @xmath223 & 60/56 & @xmath241 + neix - rich & 2141 & 0.07@xmath219 & 0.6 & 19.6 & -0.21 & 20.4 & @xmath242 & 50/56 & @xmath243 + cccccc 1 & 2.6 & 169 & 29.8@xmath244 & 25.7@xmath245 & 11.6 + 2 & 2.7 & 175 & 21.1@xmath246 & 15.2@xmath247 & 13.8 + 3 & 3.2 & 208 & 2.6@xmath248 & 2.4@xmath249 & 10.7 + 4 & 4.8 & 312 & 7.9@xmath250 & 5.4@xmath251 & 14.5 + 5 & 3.5 & 228 & 1.2@xmath248 & 1.6@xmath252 & 7.5 + 6 & 4.7 & 306 & 1.9@xmath253 & 4.7@xmath254 & 4.1 + 7 & 6.1 & 397 & 10.2@xmath255 & 6.6@xmath256 & 15.2 + 8 & 10.7 & 696 & 3.9@xmath257 & 4.2@xmath258 & 9.6 + 9 & 17.9 & 1164 & 0.9@xmath259 & 0.2@xmath260 & 43.4 + 10 & 17.7 & 1151 & 1.8@xmath259 & 0.9@xmath261 & 20.3 + | this paper is the third in a series in which we present deep chandra acis - s imaging spectroscopy of the seyfert 1 galaxy ngc 4151 , devoted to study its complex circum - nuclear x - ray emission .
emission features in the soft x - ray spectrum of the bright extended emission ( @xmath0 erg s@xmath1 ) at @xmath2 pc ( 2 ) are consistent with blended brighter ovii , oviii , and neix lines seen in the _ chandra _ hetgs and xmm - newton rgs spectra below 2 kev .
we construct emission line images of these features and find good morphological correlations with the narrow line region clouds mapped in [ oiii]@xmath35007 .
self - consistent photoionization models provide good descriptions of the spectra of the large scale emission , as well as resolved structures , supporting the dominant role of nuclear photoionization , although displacement of optical and x - ray features implies a more complex medium .
collisionally ionized emission is estimated to be @xmath412% of the extended emission .
presence of both low and high ionization spectral components and extended emission in the x - ray image perpendicular to the bicone indicates leakage of nuclear ionization , likely filtered through warm absorbers , instead of being blocked by a continuous obscuring torus
. the ratios of [ oiii]/soft x - ray flux are approximately constant ( @xmath515 ) for the 1.5 kpc radius spanned by these measurements , indicating similar relative contributions from the low- and high - ionization gas phases at different radial distances from the nucleus .
if the [ oiii ] and x - ray emission arise from a single photoionized medium , this further implies an outflow with a wind - like density profile .
using spatially resolved x - ray features , we estimate that the mass outflow rate in ngc 4151 is @xmath52@xmath6 yr@xmath1 at 130 pc and the kinematic power of the ionized outflow is @xmath7 erg s@xmath1 , approximately @xmath8 of the bolometric luminosity of the active nucleus in ngc 4151 . |
in newtonian theory , there are two basic methods of analysing the motion of a test mass under the attraction of a central , spherically symmetric gravitational field : either hamiltonian mechanics is applied to solve the problem in a spherical polar coordinate system or , for bound orbits , one may invoke a canonical transformation to action - angle variables . in the latter case , solving the _ hamilton - jacobi _ differential equation yields the integrals of motion and , in particular , _
kepler s third law _ which for elliptical orbits , in geometric units , reads @xcite @xmath9 where @xmath10 is the angular frequency of orbital revolution , @xmath7 is the eccentricity of the orbit , @xmath11 its semi - major axis and @xmath12 is the mass which acts as the source of the central field .
a generalisation of the hamilton - jacobi method to the relativistic motion of a test mass in kerr spacetime , as proposed by carter @xcite , shows that the motion is still separable and there is a complete set of constants of motion : the energy @xmath0 as measured by an observer at spatial infinity , the axial angular momentum component @xmath1 and the carter constant @xmath2 .
whereas @xmath0 and @xmath1 reflect isometries of the spacetime geometry , @xmath2 arises from the separation of the radial and polar components of motion .
the resulting first - order equations of motion can be used to compute the particle s worldline and , in limiting cases such as circular and equatorial orbits , to calculate periods of motion .
however , no attempt has been made yet to formulate a relativistic generalisation of action - angle variables for geodesic motion in kerr spacetime and to calculate the dynamical frequencies of arbitrary bound non - plunging orbits .
exactly these frequencies would be relevant to the computation of gravitational radiation emitted by stellar - mass compact objects in the vicinity of super - massive black holes if a perturbation theory based on the adiabatic approximation is employed . in this approximation one
considers the motion as being nearly geodesic under the assumption that the time scale of radiation reaction , @xmath13 , is much larger than the dynamical time scales @xmath14 , @xmath15 and @xmath16 of the three components of orbital motion .
recently , hughes @xcite has raised the question what time scales should , in general , be chosen for @xmath14 , @xmath15 and @xmath16 , as there is no obvious way of calculating orbital periods from the integrals of motion if the orbit is neither circular nor equatorial . in the following
, it is shown that suitable action - angle variables can be formulated within a fully relativistic framework for general orbits in kerr spacetime and formulas for dynamical frequencies in terms of numerically solvable integrals are derived .
let us first review the procedure applied by carter @xcite in order to determine the constants of motion .
we define the relativistic hamiltonian for geodesic motion of a particle that possesses no spin in a spacetime with given metric in the absence of an electromagnetic field as @xcite @xmath17 in this equation , the metric components @xmath18 are considered to be functions of the coordinates @xmath19 and the quantities @xmath20 are the conjugate momenta of the particle associated with these coordinates . substituting the contravariant components @xmath18 of the kerr metric in the boyer - lindquist coordinate representation @xcite , the explicit form of the hamiltonian
is given by @xmath21 where @xmath22 and @xmath23 .
a complete set of constants of motion can be determined if a canonical transformation @xmath24 is found such that the hamiltonian becomes cyclic in all of the new generalised coordinates @xmath25 , i.e. , @xmath26 , and the transformed momenta @xmath27 are thus conserved along the worldline of the particle .
the generator of such a canonical transformation , @xmath28 , is called the _
characteristic function _ and is determined by the _ hamilton - jacobi _ differential equation @xmath29 solving this equation , yields three constants of integration , @xmath0 , @xmath1 and @xmath2 , apart from @xmath30 , the value of the hamiltonian evaluated along the worldline of the particle .
the transformed conjugate momenta @xmath27 can be chosen as any functions of the these constants , i.e. , @xmath31 , assuming that @xmath32 is bijective and @xmath33 , and the generalised coordinates @xmath34 are obtained from hamilton s equations of motion , @xmath35 where each @xmath36 is a constant .
the generalised coordinates are therefore given by @xmath37 if the hamiltonian is cyclic in all of these coordinates .
for geodesic motion in kerr spacetime , the constants of motion @xmath38 and @xmath39 are related to the isometries of the metric in the coordinates @xmath40 and @xmath41 .
@xmath0 and @xmath1 can be interpreted , respectively , as the particle s energy and axial angular momentum component as seen by an observer at spatial infinity .
the third constant of motion , @xmath2 , is obtained by separating radial and polar motion and is called _
carter s constant_. in terms of @xmath42 , @xmath0 , @xmath1 and @xmath2 , the characteristic function @xmath43 is given by @xmath44 where @xmath45 ^ 2 - \delta[\mu^2 r^2+(l_{z}-a e)^2+q ] , \\
\label{eq : capt_theta } \theta & = q - \left[(\mu^2-e^2)a^2 + \frac{l_{z}^2}{\sin^2\theta}\right]\cos^2\theta.\end{aligned}\ ] ] since @xmath46 , the following set of first - order equations of motion is obtained @xcite : @xmath47 where @xmath48 .
if the conjugate momenta are chosen such that the hamiltonian becomes identically equal to @xmath49 , i.e. , @xmath50 , then we obtain @xmath51 and @xmath52 . substituting the characteristic function into the identities @xmath53 and @xmath54 and adjusting the constants @xmath55 in a suitable way , we find the following well - known integrals of motion @xcite : @xmath56 and , finally , @xmath57 in general , there are distinct turning points of the radial and polar motions which are asynchronously passed by the particle because @xmath58 and @xmath59 are not periodic functions of time . in consequence ,
equation ( [ eq : intgr_coord_time ] ) evaluates to different intervals of coordinate time when integrating over various cycles of radial or polar motion and , for this reason , there is no obvious way of calculating the time scales @xmath14 , @xmath15 and @xmath16 from the above integrals .
even though it might seem that there is no useful notion of orbital frequencies if the coordinate functions representing the motion are not periodic , it is still possible to find a representation in which _ complete periodicity _ becomes manifest .
however , such a representation is not found by means of a coordinate transformation in the usual sense , but by virtue of the more general concept of a _ canonical transformation _ of both spacetime coordinates and conjugate momenta . by its definition , a canonical transformation acts on the cotangential bundle @xmath60 of kerr spacetime , as opposed to a pure coordinate transformation between charts on the basis manifold @xmath61 . in analogy to phase space trajectories in non - relativistic hamiltonian mechanics , let us consider the locus of all points in @xmath60 which are passed by a particle in the course of its motion along a particular orbit .
we shall denote the image of that set under the coordinate chart corresponding to boyer - lindquist coordinates @xmath19 and their conjugate momenta @xmath20 by @xmath62 for a given set of constants of motion .
the basic topological properties of @xmath62 in the case of a bound orbit of the first kind can be readily inferred from the equations determining the conjugate momenta , @xmath63 firstly , since the orbital revolution progresses continuously in time , @xmath62 is not compact in timelike directions .
secondly , the motion is bound both in the @xmath64- and in the @xmath65-domain . moreover , since @xmath66 is solely a function of @xmath64 and , likewise , @xmath67 is a function of @xmath65 only , both radial and polar motion are of compact support and of the libration type provided that the orbit is _ bound _ and _ stable_. otherwise , the orbit either plunges and , hence , the particle approaches the horizon towards future timelike infinity or , if the orbit is only marginally stable , it asymptotically approached the inner turning point .
thirdly , the azimuthal motion is a rotation and , of course , it is of compact support as well because @xmath68 is to be identified with @xmath69 .
according to a theorem by arnold @xcite , that was formulated in the framework of non - relativistic mechanics , the phase space trajectory of a _ fully integrable _ system of @xmath70 degrees of freedom is diffeomorphic to the @xmath70-torus @xmath71 whenever the motion of the system is of _ compact support_. it seems plausible , that this theorem is applicable to relativistic problems as well if we account for the distinct role of the timelike coordinate which is related to the non - compactness of the worldline of a particle . accordingly , as a manifestation of the weak equivalence principle , geodesic motion in kerr spacetime has only three _
physical _ degrees of freedom because orbits are the same regardless of the mass @xmath42 of the particle .
since @xmath62 is defined with respect to a particular coordinate system , there is a well defined notion of projecting this set in the directions of the pair of conjugate variables @xmath72 .
the resulting projected set , @xmath73 , resides in the six - dimensional space spanned by the spacelike conjugate pairs @xmath74 , @xmath75 and @xmath76 and corresponds to a phase space . indeed , from the topological considerations in the previous paragraph , we see that @xmath73 is diffeomorphic to the three - torus @xmath77 .
utilising this toroidal topology , let us now define for each spatial component a closed curve @xmath78 which circumscribes the @xmath79-th torus and can be contracted into a single point with respect to the other tori . if @xmath80 is the coordinate chart on the cotangential bundle producing the boyer - lindquist coordinates and its conjugate momenta , then @xmath81 is the pre - image of the @xmath79-th loop in @xmath60 . furthermore , if we note that the _
canonical one - form _ of hamiltonian mechanics , @xmath82 , is the one - form corresponding to the relativistic four - momentum @xmath83 , i.e. , @xmath84 then we can define an _ action variable _
@xmath85 for the @xmath79-th spatial component as @xmath86 it might appear that such an action variable ultimately depends on the choice of the closed curve @xmath87 which is defined with respect to a particular coordinate system and , thus , one would assume that it is not a geometric invariant . however , @xmath85 is independent of the shape of @xmath87 because @xmath88 is _ closed _ on @xmath89 , i.e. , @xmath90 .
the closedness of the canonical one - form is shown by substituting the expressions ( [ eq : conj_momenta ] ) for the conjugate momenta and applying the rules of exterior differentiation .
the generalised stokes theorem then yields the desired property of @xmath85 @xcite .
furthermore , since transformations between coordinate charts preserve topological features , we could have defined the curves @xmath87 with respect to any coordinate system and then found their pre - images in @xmath60 .
it is only the topological features of @xmath87 with respect to the three - torus that matters !
therefore we conclude that the action variables ( [ eq : action_var ] ) are , indeed , geometric invariants . for convenience ,
however , we shall denote these variables as @xmath91 , @xmath92 and @xmath93 as if they were specific for the boyer - lindquist coordinate system .
since motion in kerr spacetime is separable in the coordinates @xmath64 , @xmath65 and @xmath41 , we may choose each @xmath87 such that it is located completely within the @xmath94-plane .
then the action variables can be easily calculated from cyclic integrals over the spatial conjugate momenta in the boyer - lindquist coordinate representation : @xmath95 as can be seen from the above expressions , @xmath91 , @xmath92 and @xmath93 are functions of the mass @xmath42 and the three constants of motion @xmath0 , @xmath1 and @xmath2 .
the generator of the transformation to the action variables @xmath85 is given by @xmath96 .
it is not globally defined , but changes by an amount of @xmath97 when integrating over the path @xmath87 . as a consequence ,
if the spatial coordinate @xmath98 goes through @xmath70 complete cycles and the other coordinates are kept unchanged , then the generalised coordinate @xmath99 associated with @xmath85 changes by @xmath100 .
this result implies that @xmath101 is a phase or _
angle variable _ and , since @xmath102 , the constant @xmath103 can be interpreted as a frequency . moreover , as a consequence of the geometrically invariant definition of the action variables @xmath85 , the frequencies @xmath103 specify fundamental properties of the orbital motion independent of any particular coordinate representation .
therefore , they are called the _ fundamental frequencies _ of the system .
however , these frequencies are subject to a straightforward physical interpretation only with respect to the spatial coordinates @xmath64 , @xmath65 and @xmath41 , when the paths @xmath87 are correspondingly chosen as cycles of , respectively , radial , polar and azimuthal motion . accordingly
, we shall use the notation @xmath3 , @xmath4 and @xmath5 for the fundamental frequencies of an orbit in kerr spacetime .
the standard procedure of determining fundamental frequencies is to find the explicit form of the hamiltonian in the action - angle representation , @xmath104 , and to calculate the frequencies from the partial derivatives with respect to the action variables @xmath85 @xcite : @xmath105 unfortunately , the definition of the radial and polar action variables in terms of the non - trivial integrals ( [ eq : action_var_r ] ) and ( [ eq : action_var_theta ] ) does not admit an explicit inversion because none of these integrals can be solved analytically .
however , as is shown in [ sct : determ_freq ] , the derivatives @xmath106 and , hence , the frequencies @xmath103 can be found even without knowing the functional form of the hamiltonian @xmath104 if the theorem on implicit functions is employed .
this procedure results in equations ( [ eq : h_derv_jr])([eq : h_derv_jphi ] ) of [ sct : determ_freq ] and the fundamental frequencies are therefore given by @xmath107x(r_{1},r_{2 } ) + 2 y(r_{1},r_{2 } ) , \\
\frac{2\pi}{\mu\omega^{\theta } } & = \frac{4 a^{2 } z_{+}}{\beta}\left[k(k)-e(k)\right ] + \frac{4}{\beta z_{+}}\frac{x(r_{1},r_{2})}{y(r_{1},r_{2})}k(k ) , \\
\frac{2\pi}{\mu\omega^{\phi } } & = 2\pi\frac{a^{2}z_{+}^{2}[k(k)-e(k)]x(r_{1},r_{2 } ) + k(k)y(r_{1},r_{2 } ) } { l_{z}[\pi(z_{-}^{2},k)-k(k)]x(r_{1},r_{2 } ) + k(k)z(r_{1},r_{2})},\end{aligned}\ ] ] where @xmath108 and @xmath109 are the turning points of radial motion , @xmath110 are the two roots of the equation @xmath111 when substituting @xmath112 in @xmath113 , @xmath114 , @xmath115 , @xmath116 , @xmath117 and @xmath118 are , respectively , the complete elliptical integrals of the first , second and third kind , @xmath119 and @xmath120 , @xmath121 and @xmath122 are radial integrals defined by @xmath123}{\delta\sqrt{r}}\,\rmd r.\end{aligned}\ ] ] the above equations for the fundamental frequencies , however , have two deficiencies : firstly , they are suggestive of @xmath103 being dependent on the mass @xmath42 of the particle which is , of course , not the case .
for this reason , we will cast these equations into a dimensionless scale - invariant form , in which @xmath42 does not appear anymore .
secondly , the radial functions @xmath124 , @xmath125 and @xmath126 are not proper integrals because the integrated functions are divergent at the turning points @xmath108 and @xmath109 .
fortunately , they are transformed into well - behaved integrals by virtue of the substitution @xmath127 , where @xmath128 varies from @xmath129 to @xmath130 as @xmath64 goes through a complete cycle .
although this substitution originates from the pure kepler ellipse , it is actually still expedient even for relativistic orbits that may be very different in shape . in analogy to non - relativistic orbits , @xmath6 is called the _ semi - latus rectum _ and @xmath7 is the _ eccentricity _ of the orbit . working out this substitution ,
yields the following equations for the dimensionless fundamental frequencies @xmath131 : @xmath132 where the tilde on top of each symbol indicates the corresponding dimensionless quantity ( see [ sct : const_of_motion ] ) and the constant @xmath133 is defined by @xmath134 the dimensionless radial integrals in the @xmath128-representation are given by @xmath135 where the functions @xmath136 , @xmath137 and @xmath138 are defined by @xmath139\frac{1-e^{2}}{p^{2 } } \right\ } \nonumber\\ & \times ( 1+e\cos\chi)^{2 } , \\
\label{eq : def_h } h(\chi ) = & 1 - \frac{2}{p}(1+e\cos\chi ) + \frac{\tilde{a}^{2}}{p^{2}}(1+e\cos\chi)^{2 } , \\
\label{eq : def_g } g(\chi ) = & \tilde{l_{z } } - \frac{2(\tilde{l}_{z}-\tilde{a}\tilde{e})}{p}(1+e\cos\chi).\end{aligned}\ ] ] in the above equations , the constants of motion @xmath0 , @xmath1 and @xmath2 are understood as functions of the orbital parameters @xmath6 , @xmath7 and @xmath140 .
although analytical expressions for these functions can be derived , it is hardly feasible to substitute these expressions for @xmath0 , @xmath1 and @xmath2 . for given orbital parameters ,
it is much easier to calculate the constants of motion numerically , as explained in [ sct : const_of_motion ] , and to substitute the resulting values into @xmath141 , @xmath142 and @xmath143 .
apart from the three frequencies @xmath3 , @xmath4 and @xmath5 , a fourth constant of motion is obtained , which is associated with the timelike generalised coordinate @xmath144 in the action - angle variable representation . according to equation ( [ eq : h_derv_p0 ] ) of [ sct : determ_freq ] ,
this constant is given by @xmath145k(k)- a^{2}z_{+}^{2}e x(r_{1},r_{2})e(k ) } { \left[y(r_{1},r_{2})+a^{2}z_{+}^{2}x(r_{1},r_{2})\right]k(k)- a^{2}z_{+}^{2}x(r_{1},r_{2})e(k)},\ ] ] where the function @xmath146 is defined by @xmath147\,\rmd r } { \delta\sqrt{r}}.\ ] ] in dimensionless form , @xmath148 can be expressed as @xmath149,\ ] ] where @xmath150 and the function @xmath151 is defined by @xmath152 as for the physical interpretation of @xmath148 , we note that the conjugate momentum associated with the generalised coordinate @xmath153 is @xmath154 , i.e. , the same as in the boyer - lindquist representation .
for this reason , one would expect that @xmath153 is in some sense an interval of coordinate time .
however , it can not be equal to @xmath155 as equation ( [ eq : intgr_coord_time ] ) tells us that the relation between the elapse of proper time @xmath156 and the corresponding interval of coordinate time , @xmath155 , is not linear .
on the other hand , there are three particular scales of proper time , @xmath157 , @xmath158 and @xmath159 , associated with the three fundamental frequencies , @xmath3 , @xmath4 and @xmath5 . substituting each of these constants into the expression for @xmath153 , three scales of _
coordinate time _
are produced : @xmath160 this seems to be a natural definition of the coordinate - time scales associated with the fundamental frequencies .
as we shall see in following section , the above definition , indeed , reproduces the known periods of coordinate time in the limits of , respectively , circular and equatorial motion .
hence , we are led to the conjecture that @xmath148 can be interpreted as a _ gravitational lorentz factor _ which relates the three dynamical scales of proper time , @xmath157 , @xmath158 and @xmath159 , to the corresponding intervals of coordinate time , @xmath14 , @xmath15 and @xmath16 . in the general case , this conjecture has the following implication : the ratio of time scales associated with two particular components of motion , say , radial and polar motion , is the same regardless of the frame of reference in which these time scales are defined .
for example , if an observer at infinity saw the intervals of time @xmath14 and @xmath15 , he would find that the ratio @xmath161 is equal to @xmath162 .
of course , this is exactly what one would expect , provided that the fundamental frequencies truly arise from invariant properties of orbital motion . in conclusion
, the ratio of time scales should be independent of the frame of reference in use .
with respect to coordinate time @xmath40 , this is inherently ensured by the above definition of @xmath14 , @xmath15 and @xmath16 . ultimately , the significance of the fundamental frequencies lies in the possibility of expanding a dynamical quantity @xmath163 in a fourier series with harmonics @xmath164 of these frequencies @xcite : @xmath165 where @xmath166 and @xmath167 according to definition of the action variables @xmath168 , @xmath169 and @xmath170 . the fourier coefficients @xmath171 of the above expansion are given by @xmath172 therefore , dynamical properties of the particle are completely specified by discrete sets of numbers such as @xmath173 .
this is the very essence of bound geodesic orbital motion in a kerr spacetime being _ quasi - periodic _ rather than quasi - chaotic , as was proposed due to the ergodic properties of these orbits .
as an important test to support the validity of the fundamental frequencies derived in the previous section , we will consider two special cases of bound motion in kerr spacetime : firstly , the limit of vanishing eccentricity and , secondly , purely equatorial orbits .
in particular , we will see that previously derived equations for the periods and the corresponding azimuthal angles of circular orbits @xcite can be reproduced from the general formulas .
briefly summarised , the features of orbits in the two limiting cases are as follows : * circular orbits ( @xmath7 = 0 ) : in schwarzschild spacetime , kepler s third law ( [ eq : kepler ] ) holds exactly for the period of azimuthal revolution with respect to coordinate time , i.e. , @xmath174 . for circular orbits in kerr spacetime , there are two distinct frequencies depending on whether the particle co - revolves or counter - revolves with the rotation of the black hole . if the motion is not confined to the equatorial plane , then the particle changes periodically its altitude such that @xmath175 is twice the elapse of proper time between two subsequent passages of the particle through the equatorial plane .
the frequencies @xmath4 and @xmath5 are generally incommensurate . as a consequence ,
the azimuthal angle @xmath176 that is accumulated as the particle completes a cycle of polar motion is different from @xmath130 . * equatorial orbits ( @xmath177 ) : as a consequence of the different shape of the potential of radial motion as compared to the newtonian potential , non - circular orbits do not close on themselves as the particle revolves around the gravitational centre . the total azimuthal angle
@xmath178 swept out by the particle as it moves from , say , the apastron at radius @xmath179 to the periastron at radius @xmath180 and back to the apastron is given by @xmath181 .
actually , there is another limit , called polar orbits , for which @xmath182 . in this case
, the particle crosses the spin axis of the black hole and the inclination parameter is @xmath183 .
however , it seems unlikely that polar orbits are astrophysically relevant and we only refer to @xcite for a general discussion .
orbits of zero eccentricity , @xmath184 , are called circular . in this case , the particle moves on a spherical surface of constant radial coordinate , @xmath185 , within the bounds of the inclination @xmath65 given by @xmath186 .
since the perihelion and aphelion radii coincide , we have @xmath187 and @xmath188 . in consequence ,
the integrals given by equations ( [ eq : rad_integr_x])([eq : rad_integr_z ] ) and ( [ eq : rad_integr_w ] ) are undefined in the limit of vanishing eccentricity .
the corresponding quantities in the @xmath128-representation , however , have well - defined limits because the functions @xmath151 , @xmath136 , @xmath137 and @xmath138 are constant if @xmath184 .
hence , we define @xmath189 where the constants @xmath190 , @xmath191 , @xmath192 and @xmath193 are obtained by substituting @xmath184 in equations ( [ eq : def_j])([eq : def_g ] ) and ( [ eq : def_f ] ) .
the frequencies of the angular components of motion are therefore given by @xmath194 } , \\
\label{eq : freq_phi_circ } \tilde{\omega}_{\phi}^{(\mathrm{circ } ) } & = \frac{\tilde{a}(2\tilde{e}\tilde{r}_{0 } - \tilde{a}\tilde{l}_{z})\tilde{\delta}_{0}^{-1}k(k ) + \tilde{l}_{z}\pi(z_{-}^{2},k ) } { ( \tilde{r}_{0}^{2}+\tilde{a}^{2}z_{+}^{2})k(k ) - \tilde{a}^{2}z_{+}^{2}e(k)}.\end{aligned}\ ] ] the existence of a distinct radial frequency even in the case of circular motion indicates that the limit of the integrated azimuthal angle over a cycle of _ radial _ motion , @xmath178 , does not become @xmath130 as the eccentricity approaches zero .
of course , this is well known from equatorial orbits .
on the other hand , we can calculate the integrated azimuthal angle @xmath195 as the particle goes through one cycle of _ polar _ motion by combining the equations of motion ( [ eq : motion_theta ] ) and ( [ eq : motion_phi ] ) .
the result is @xmath196,\ ] ] where @xmath197 .
when comparing this expression for @xmath195 with the quotient of @xmath198 and @xmath199 , we see that the following well known equation holds : @xmath200 furthermore , calculating the period of coordinate time @xmath15 for a cycle of polar motion from equation ( [ eq : intgr_coord_time ] ) , yields @xmath201 \right\}.\ ] ] if the constant @xmath148 given by equation ( [ eq : gamma ] ) is also calculated in the limit @xmath184 , we find that @xmath202 which suggests that @xmath203 can be interpreted as the _ lorentz factor _ of circular motion in kerr spacetime .
if @xmath177 , then the equation for the turning points of polar motion , @xmath111 , admits the solutions @xmath204 and @xmath205 . as a consequence , the motion of the particle is confined to the equatorial plane , i.e. , @xmath206 . since @xmath207 ,
the elliptical integrals become @xmath208 .
the resulting expressions for the fundamental frequencies of equatorial orbits are thus given by @xmath209 as in the case of circular orbits , we still have three different frequencies . combining the equations of motion ( [ eq : motion_r ] ) and ( [ eq : motion_phi ] )
, we can calculate the integrated azimuthal angle @xmath178 between two successive passages through , say , the periastron and we find that @xmath210 according to equation ( [ eq : intgr_coord_time ] ) , the corresponding period of coordinate time is given by @xmath211 .
using @xmath212 , we obtain @xmath213 and one can see that @xmath214 acts as a lorentz factor which relates the elapse of proper time over one cycle of equatorial motion with the corresponding period of coordinate time . in the newtonian limit ,
the constants of motion asymptotically approach the values of keplerian orbits which are given by equations ( [ eq : const_of_motion_kepl ] ) in [ sct : const_of_motion ] . since the keplerian values of energy and angular momentum , naturally , are independent of the spin of the black hole , the problem reduces to that one of motion in schwarzschild spacetime and we can set @xmath215 .
thus , we obtain the the asymptotic expressions @xmath216 which lead to @xmath217 where @xmath10 is the keplerian frequency given by equation ( [ eq : kepler ] ) .
substitution of the above asymptotic expressions into equations ( [ eq : freq_r])([eq : freq_phi ] ) and ( [ eq : gamma ] ) results in @xmath218 the three components of motion therefore become commensurate in the newtonian limit and the corresponding frequencies asymptotically degenerate . in the nearly newtonian regime , where the three frequencies are close to @xmath10 yet slightly different from each other , we expect that there will be a minute perihelion drift ( due to the small difference between @xmath5 and @xmath3 ) as well as a slow precession of the orbital plane ( due to the difference between @xmath5 and @xmath4 ) . in order to calculate this precession , an approximation of the fundamental frequencies including terms of next lower order in @xmath6 would be required .
however , this is left for future work .
let us now consider bound orbits around a nearly extreme kerr black hole of spin @xmath219 .
such a black hole is particularly interesting because effects due to spacetime dragging caused by the high angular momentum of the black hole are strong and stable orbits exist even very close to the horizon ( which is located at @xmath220 ) .
subsequently , we will explore the parameter space in all three dimensions and analyse the behaviour of the fundamental frequencies as each of the orbital parameters is varied .
the numerical evaluation of equations ( [ eq : freq_r])([eq : freq_phi ] ) and a general algorithm to calculate the constants of motion @xmath0 , @xmath1 and @xmath2 for arbitrary orbital parameters @xmath6 , @xmath7 and @xmath8 was implemented in mathematica . to begin with , we choose @xmath221 ( in this case the aphelion radius is twice the perihelion radius ) and @xmath222 ( corresponding to @xmath223 of maximum altitude from the equatorial plane ) .
the fundamental frequencies of the prograde and retrograde orbits as functions of @xmath6 are shown in figure [ fg : freq_p ] . basically , one can see that the frequencies @xmath224 , @xmath225 and @xmath226 increasingly deviate from each other as the distance from the black hole becomes smaller . for increasing @xmath6 ,
on the other hand , the fundamental frequencies tend to degenerate and , in fact , they converge towards the keplerian frequency @xmath227 for @xmath221 . thus
, we call the keplerian power law for given eccentricity the _ asymptotic keplerian branch _ of the fundamental frequencies .
the values of @xmath6 , for which the frequencies of , respectively , prograde and retrograde orbits diverge , are located exactly where the respective orbits become _ marginally stable_. in this case , stable prograde orbits only exist if @xmath228 and stable retrograde orbits only if @xmath229 .
hence , the corresponding minimal perihelion radii of stable prograde and retrograde orbits with @xmath221 and @xmath222 are , respectively , @xmath230 and @xmath231 .
note that @xmath232 and @xmath233 as @xmath234 .
this is a consequence of the orbits becoming `` zoom whirls '' which means that the particle moves from the apastron inwards and then revolves many times along an almost circular spiral before it actually reaches the periastron . in the limit of a marginally stable orbit ,
the periastron is approached asymptotically .
the elapse of proper time as the particle moves from apastron to periastron , thus , ever increases as @xmath234 and the radial frequency approaches zero .
if the particle represents an astrophysical object , however , radiation reaction will finally act on a time scale that is comparable to the dynamical time scale of radial motion ( @xmath235 ) and , as a consequence , it will drive the particle within finite time over the stability threshold so that it plunges into the black hole .
in parameter space , the locus of all points corresponding to marginally stable orbits , @xmath236 , is called the _
separatrix_. a plot of the separatrix for @xmath219 is shown in figure [ fg : sptrx ] , where the perihelion radius @xmath180 is used as parameter in place of @xmath6 .
there are two sheets , one corresponding to prograde marginally stable orbits and the other one corresponding to retrograde marginally stable orbits .
the two sheets smoothly join at @xmath237 , i.e. , where the polar orbits with @xmath238 are located .
as @xmath140 becomes smaller , however , there is an increasing space between the two sheets , in which only prograde but no retrograde orbits reside . as the eccentricity increases , both prograde and retrograde orbits exist closer to the horizon due to the higher momentum of the particle at the periastron as compared to circular orbits of the same radius . in particular , there are prograde orbits of eccentricity @xmath239 with @xmath240 for which the periastron is very close to the horizon ( within @xmath241 ) .
figure [ fg : freq_e ] shows the change of the fundamental frequencies with eccentricity for various choices of @xmath8 and @xmath6 .
the main feature of these plots is the convergence of the frequencies to a single value as @xmath242 .
this simply reflects the fact that orbits of eccentricity very close to unity are almost unbound and the apastron is located far away from the region of strong gravity . in consequence
, the particle is mostly moving in the nearly newtonian regions of spacetime and the fundamental frequencies tend to degenerate . in the plot showing the frequencies of nearly polar orbits with @xmath243 and @xmath244 ,
the frequencies terminate just below @xmath245 as prograde orbits of higher eccentricity do not exist if @xmath246 ( note that the value of the perihelion radius depends on @xmath7 according to @xmath247 when comparing to the plot of the separatrix in figure [ fg : sptrx ] ) .
the dependence of the fundamental frequencies on the inclination parameter @xmath8 ( which is the smallest angle of inclination to the spin axis of the black hole that can be reached by the particle ) is shown in figure [ fg : freq_z ] for different values of @xmath7 and @xmath6 .
apparently , both the radial and polar frequencies of prograde and retrograde orbits converge towards each other as @xmath248 , i.e. , in the polar limit . on the other hand
, there is a crossing - over of the azimuthal frequencies at some finite value of @xmath8 . as a consequence ,
the azimuthal motion of nearly polar retrograde orbits is _ faster _ than that of the corresponding prograde orbits and , thus , it seems that nearly polar orbits are markedly different from nearly equatorial orbits .
we have to keep in mind , however , that the revolution of the particle around the black hole in the case @xmath249 is rather specified by the polar frequency @xmath4 , whereas @xmath5 determines an azimuthal precession around the spin axis of the black hole .
finally , in figure [ fg : time_p ] , the scales of coordinate time , @xmath14 , @xmath15 and @xmath16 , defined by equation [ eq : coord_time_scl ] are shown in physical units as functions of the semi - major axis @xmath250 for orbits of eccentricity @xmath221 and inclination parameter @xmath222 .
the mass of the black hole is chosen to be @xmath251 which is about the mass determined for the black hole in the centre of our galaxy @xcite .
it is interesting to note that for virtually all prograde orbits the time scale of polar motion @xmath16 is very close to the the keplerian orbital period @xmath252 .
however , this is only accidentally the case for the chosen set of parameters , as for lower eccentricity @xmath16 is above @xmath253 , whereas it tends to become significantly smaller than @xmath253 at higher eccentricity .
the investigation of bound geodesic orbits in kerr spacetime presented in this article clearly illustrates that the properties of these orbits in the regime of strong gravity are profoundly different from keplerian orbits in the newtonian regime .
in particular , there are three different frequencies of orbital motion , @xmath3 , @xmath4 and @xmath5 , which increasingly deviate from the keplerian frequency as the size of the orbit decreases and finally approaches the limit of a marginally stable orbit .
this is a consequence of the radial , polar and azimuthal components of motion being _ incommensurate _ in kerr spacetime . as a consequence ,
the shapes of orbits in the vicinity of a kerr black hole are exceedingly complex . the explicit formulas for the fundamental frequencies @xmath3 , @xmath4 and @xmath5 presented in section [ sct : fund_freq ] are exact in the test particle limit and useful approximations if radiation reaction is _ adiabatic _ ,
i.e. , if the time scales associated with the frequencies are small compared to the time scale of radiation reaction . based on the constant of motion @xmath148 which is associated with a timelike generalised coordinate , we have proposed a conversion of the invariant fundamental frequencies into coordinate - time scales , @xmath14 , @xmath16 and @xmath15 , which should be relevant to an observer at spatial infinity and , thus , gravitational wave astronomy . once these time scales are established , one can compute _ instantaneous _ gravitational wave forms in the adiabatic approximation . in this context , the term `` instantaneous '' means an interval of time , over which no significant changes of the constants of motion are accumulated due to radiation reaction . from the resulting spectrum , an estimation of the power emitted in the frequency band of a detector could be made , without making use of an advanced theory of radiation reaction .
i thank c. glampedakis for many discussions and making me aware of the problem concerning the dynamical time scales of geodesic orbits in kerr spacetime and also b. s. sathyaprakash and l. p. grishchuk for their support and useful criticism .
this work was funded by a _
wilhelm - macke mobility scheme award _ of the johannes - kepler - university linz , austria , as part of my postgraduate studies at cardiff university .
let @xmath254 be the momenta given by @xmath154 and @xmath255 , where the @xmath85 are the action variables defined by equations ( [ eq : action_var_r])([eq : action_var_phi ] ) .
if we denote the jacobian matrix of @xmath32 by @xmath256 , then , by the theorem on implicit functions , @xmath257 , provided that @xmath32 is non - zero and the jacobian does not vanish @xcite . since @xmath258 is the invariant value of the hamiltonian , we can substitute @xmath259 , where @xmath104 is the hamiltonian in the action - angle variable representation . for brevity ,
let us subsequently use the symbol @xmath137 to denote the hamiltonian in that representation .
moreover , two rows of the jacobian matrix are trivial due to the identities @xmath260 and @xmath261 .
thus , the equation @xmath262 reads @xmath263 the above matrix equation can be split into four non - trivial sets of linear equations in the eight unknowns @xmath264 , @xmath265 , @xmath266 and @xmath267 : @xmath268 \\ 2 a^{2 } e\int_{\theta_{-}}^{\pi/2 } \frac{\cos^{2}\theta\rmd \theta}{\sqrt{\theta } } \end{array}\right ) \\
\label{eq : syst_act_r } & \mathsfsl{a}\cdot\frac{\partial}{\partial j_{r } } \left(\begin{array}{llll } h \\
q \end{array}\right ) = \left(\begin{array}{llll } 2\pi \\ 0 \end{array}\right ) , \\ & \mathsfsl{a}\cdot\frac{\partial}{\partial j_{\theta } } \left(\begin{array}{llll } h \\
q \end{array}\right ) = \left(\begin{array}{llll } 0 \\ \frac{\pi}{2 } \end{array}\right ) , \\ & \mathsfsl{a}\cdot\frac{\partial}{\partial j_{\phi } } \left(\begin{array}{llll } h \\
q \end{array}\right ) = \left(\begin{array}{llll } 2 \int_{r_{1}}^{r_{2}}\frac{\rmd r}{\sqrt{r } } \left[\frac{a p}{\delta}+(l_{z}-a e)\right ] \\ 2 l_{z}\int_{\theta_{-}}^{\pi/2 } \frac{\cot^{2}\theta\rmd \theta}{\sqrt{\theta } } \end{array}\right)\end{aligned}\ ] ] where @xmath269 , and the coefficient matrix @xmath270 is given by @xmath271 using the definitions ( [ eq : rad_integr_x])([eq : rad_integr_z ] ) and ( [ eq : rad_integr_w ] ) for the radial integrals and the identities @xmath272 , \\
\label{eq : angl_intgr3 } \int_{\theta_{-}}^{\pi/2}\frac{\cot^{2}\theta}{\sqrt{\theta}}\,\rmd \theta & = \frac{z_{+}}{\beta}[\pi(z_{-}^{2},k)-k(k)],\end{aligned}\ ] ] where the elliptical integrals @xmath116 , @xmath117 and @xmath273 are defined by equations ( [ eq : ell_integr_k])([eq : ell_integr_pi ] ) , the solutions of the above systems of equations for @xmath274 and @xmath265 are given by @xmath275x(r_{1},r_{2 } ) } { k(k)y(r_{1},r_{2 } ) + a^{2}z_{+}^{2}\left[k(k)-e(k)\right]x(r_{1},r_{2 } ) } , \\
\label{eq : h_derv_jr } & \frac{\partial h}{\partial j_{r } } = \frac{\pi k(k)}{k(k)y(r_{1},r_{2 } ) + a^{2}z_{+}^{2}\left[k(k)-e(k)\right ] x(r_{1},r_{2 } ) } , \\
\label{eq : h_derv_jtheta } & \frac{\partial h}{\partial j_{\theta } } = \frac{\pi\beta z_{+}x(r_{1},r_{2})}{2\{k(k)y(r_{1},r_{2 } ) + a^{2}z_{+}^{2}\left[k(k)-e(k)\right]x(r_{1},r_{2})\ } } , \\
\label{eq : h_derv_jphi } & \frac{\partial h}{\partial j_{\phi } } = \frac{k(k)z(r_{1},r_{2 } ) + l_{z}[\pi(z_{-}^{2},k)-k(k)]x(r_{1},r_{2 } ) } { k(k)y(r_{1},r_{2 } ) + a^{2}z_{+}^{2}[k(k)-e(k)]x(r_{1},r_{2})}.\end{aligned}\ ] ] the validity of the above solution depends on the jacobian matrix @xmath256 being non - singular .
in fact , @xmath276 if and only if @xmath277 \left(\frac{\partial h}{\partial j_{r}}\right)^{2 } \neq 0,\ ] ] as is easily shown by using the second equation of system ( [ eq : syst_act_r ] ) .
the above condition is fulfilled whenever @xmath278 , i.e. , if the orbit is non - equatorial . since @xmath177 for equatorial orbits ,
this is immediately clear from the above systems of equations as well .
nevertheless , the frequencies of equatorial orbits can be calculated by taking the limit @xmath279 of the general expressions for @xmath278 and , as demonstrated in section [ sct : equat ] , one finds agreement with the results obtained from a direct calculation based on the equations of motion .
in order to find the constants of motion @xmath0 , @xmath1 and @xmath2 for given orbital parameters @xmath6 , @xmath7 and @xmath8 , one has to solve the following set of equations : @xmath280 where @xmath281 and @xmath282 are given by equations ( [ eq : capt_r ] ) and ( [ eq : capt_theta ] ) .
the roots of these equations correspond to the turning points of , respectively , radial and polar motion , i.e.,we can substitute @xmath283 or @xmath284 in the first equation and @xmath285 in the second equation .
thus , a system of three equations is obtained which can be solved for the three unknown constants of motion , provided that @xmath286 . in the case @xmath184 , on the other hand , we have got to make use of the supplemental constraint @xmath287 , in addition to @xmath188 and @xmath288 , because circular orbits occur if both @xmath281 and its radial gradient @xmath289 vanish at the same point , @xmath290 . introducing the dimensionless quantities @xmath291 we can actually set up a completely scale - invariant formulation of the problem .
hence , it is sufficient to calculate @xmath292 , @xmath293 and @xmath294 only once for given orbital parameters .
the corresponding physical values for a particular black hole of mass @xmath12 and a particle of mass @xmath42 can then be inferred from the above definitions . to begin with , we re - arrange @xmath288 in order to express carter s constant @xmath294 in terms of @xmath8 , @xmath292 and @xmath293 , @xmath295,\ ] ] where @xmath140 .
if this expression is substituted for @xmath294 in the equation @xmath296 , we obtain @xmath297 where @xmath298 , \\ g(r ) & = 2\tilde{a}\tilde{r } , \\ h(r ) & = \tilde{r}(\tilde{r}-2 ) + \frac{z_{-}^{2}}{1-z_{-}^{2}}\tilde{\delta } , \\
d(r ) & = ( \tilde{r}^{2 } + \tilde{a}^{2}z_{-}^{2})\tilde{\delta},\end{aligned}\ ] ] and @xmath299 .
substitution of @xmath294 in @xmath300 yields @xmath301 where @xmath302 , \\ g'(r ) & = 2\tilde{a } , \\ h'(r ) & = \frac{2(\tilde{r}-1)}{1-\tilde{z}_{-}^{2 } } , \\ d'(r ) & = 2\left(2\tilde{r}-3\right)\tilde{r}^{2 } + 2\tilde{a}^{2}\left[(1+z_{-}^{2})\tilde{r}-z_{-}^{2}\right].\end{aligned}\ ] ] thus , we shall define the following coefficients in terms of the orbital parameters : @xmath303 eliminating @xmath293 from the set of quadratic equations , @xmath304 a quadratic equation in @xmath305 is obtained , @xmath306 where @xmath307 , @xmath308 , @xmath309 , @xmath310 and @xmath311 are defined by the following @xmath312 determinants : @xmath313 the roots of equation ( [ eq : energy ] ) are given by @xmath314 actually , these roots can be calculated regardless of whether the corresponding orbits are of the first or the second kind , i.e. , if they are stable or plunging .
there is a simple criterion , however , which can be used to determine the kind of the orbit : let the _ potential _ of the radial component of motion be defined by @xmath315 according to equation ( [ eq : motion_r ] ) , @xmath316 and the radial domain of motion is subject to the constraint @xmath317 .
hence , the orbit is _ stable _ if @xmath318 if the gradient of the potential vanishes at @xmath319 , the potential has a local maximum at the perihelion radius and , in that case , the orbit is _ marginally stable_. substituting @xmath320 into the system ( [ eq : energy_angmomt ] ) and solving for @xmath1 , yields four roots which satisfy both equations .
therefore , we obtain _ four _ possible solutions for the constants of motion with given orbital parameters , namely , @xmath321 , @xmath322 , @xmath323 and @xmath324 @xmath325 , where @xmath326 is the _ lower _ and @xmath327 the _ higher _ energy given by equation ( [ eq : energy_soln ] ) .
numerical evaluation shows that @xmath328 . in the first case , the particle has higher binding energy and co - revolves with the rotation of the black hole .
such orbits are called _
prograde_. in the second case , the particle has lower binding energy and usually counter - revolves , except for some orbits close to the horizon of rapidly rotating black holes .
these orbits are called _
retrograde_. the solutions with negative energies correspond to the respective motions under time reversal .
finally , we consider the _
newtonian _ limits of the constants of motion which are asymptotically approached as @xmath329 . using the formulas for the potential energy and angular momentum of keplerian orbits in geometric units @xcite , @xmath330 and noting that @xmath331
, we obtain the following asymptotic expressions for the constants of motion : and @xmath332 are fulfilled asymptotically in the highest power for non - circular orbits , and @xmath333 is satisfied down to @xmath334 if @xmath184 .
] @xmath335 since @xmath336 , carter s constant in the newtonian limit is the squared angular momentum component parallel to the equatorial plane . | the dynamical parameters conventionally used to specify the orbit of a test particle in kerr spacetime are the energy @xmath0 , the axial component of the angular momentum , @xmath1 , and carter s constant @xmath2 .
these parameters are obtained by solving the hamilton - jacobi equation for the dynamical problem of geodesic motion . employing the action - angle variable formalism ,
on the other hand , yields a different set of constants of motion , namely , the fundamental frequencies @xmath3 , @xmath4 and @xmath5 associated with the radial , polar and azimuthal components of orbital motion .
these frequencies , naturally , determine the time scales of orbital motion and , furthermore , the instantaneous gravitational wave spectrum in the adiabatic approximation . in this article , it is shown that the fundamental frequencies are geometric invariants and explicit formulas in terms of quadratures are derived .
the numerical evaluation of these formulas in the case of a rapidly rotating black hole illustrates the behaviour of the fundamental frequencies as orbital parameters such as the semi - latus rectum @xmath6 , the eccentricity @xmath7 or the inclination parameter @xmath8 are varied .
the limiting cases of circular , equatorial and keplerian motion are investigated as well and it is shown that known results are recovered from the general formulas . |
there have been several studies of oort cloud formation in a star cluster environment . in 1999
soenke eggers published his ph.d .
thesis in which he analysed the effects of a star cluster on the formation and evolution of the oort cloud ( oc ) .
he used a monte carlo method and two star clusters , in which the stellar encounters occurred at constant time intervals and their effects on the comets were computed analytically .
the first cluster had an effective number density of 625 stars pc@xmath0 and the other had an effective number density two orders of magnitude lower .
for both clusters eggers assumed a velocity dispersion of 1 km s@xmath1 .
his model did not include a tidal field caused by the cluster potential .
eggers defined a comet to be in the oort cloud if it attained @xmath2 au and simultaneously @xmath3 au . with these definitions
, he obtained efficiencies of 1.7% and 4.8% for the low and high density clusters respectively , with comets on orbits with a typical semi - major axis between 3000 au and 6000 au .
the clouds were also found to be mostly isotropic .
+ a parallel study of the formation of the inner oort cloud in a denser , clusteresque environment has been performed by fernndez & brunni ( 2000 ) .
they placed comets on eccentric orbits with semi - major axes @xmath4 au and included an approximate model of the tidal field of the gas and passing stars from the cluster .
the cluster had a maximum stellar number density of 100 pc@xmath0 and the maximum mass density of the core of the molecular cloud gas was 5000 @xmath5 pc@xmath0 .
their simulations formed a dense inner oort cloud where the comets had semi - major axes of a few hundred to a few thousand astronomical units .
the outer edge of this cloud was dependent on the density of gas and stars in the cluster .
the most interesting part of their study was their ability to successfully deposit a fair amount of material that was scattered by jupiter and saturn , which were the main contributors to forming the inner oort cloud .
in the current environment , on the other hand , the contribution to the oort cloud from jupiter and saturn is lower than that from uranus and neptune ( e.g. dones et al . , 2004 ) .
however , fernndez & brunni ( 2000 ) pointed out that if the sun remained in this dense environment for long , the passing stars could strip a significant fraction of the comets away from the sun and the trapping efficiency might end up being low .
this low trapping efficiency was partially a result of their cluster lifetimes being too long .
+ the interest in the formation of the inner oort cloud in a cluster environment gained renewed interest with the discovery of ( 90377 ) sedna , a dwarf planet with semi - major axis 500 au and perihelion of 76 au , so that its orbit is detached from that of neptune ( brown et al . , 2004 ) .
gladman et al . (
2002 ) has shown that the object 2000 cr@xmath6 , having an orbit with @xmath7 au and @xmath8 au , could not be reproduced via chaotic diffusion , which ceases beyond @xmath9 au .
thus another mechanism had to be responsible for placing both 2000 cr@xmath6 and sedna on their current orbits .
morbidelli & levison ( 2004 ) , together with kenyon & bromley ( 2004 ) , successfully demonstrated that the most viable way to reproduce the orbits of sedna and 2000 cr@xmath6 is via a slow , close passage of a relatively heavy star .
morbidelli & levison ( 2004 ) argued that the encounter had to happen early in order to still have a reasonably populated oort cloud afterwards .
however , the low velocity of the encounter is difficult to obtain in the current galactic environment , and they suggested that this passage occurred while the sun was in its birth cluster .
+ the above results led to brasser et al .
( 2006 ) - henceforth bdl6 - to investigate the formation of the inner oort cloud in an embedded cluster environment , in which the gas from the molecular cloud is still present ( lada & lada , 2003 ) . inspired by fernndez & brunni ( 2000 )
they attempted to constrain the environment that was needed to save comets under the dynamical control of jupiter .
they employed a plummer model ( plummer , 1911 ) to construct a series of clusters with varying central density but with a more or less fixed number of stars , because most stars form in clusters of a few hundred stars ( lada & lada , 2003 ) .
bdl6 used a simple leapfrog integrator for the cluster in the plummer potential and recorded the positions and velocities of stars , including time , as they came within a user - specified distance of the sun .
the encounter data were then used in swift rmvs3 ( levison & duncan , 1994 ) .
the latter was modified to include the effects of the stars , as in dones et al .
( 2004 ) , and the gravitational force from the cluster gas on the comets and the planets .
the sun was assumed to be on a fixed orbit in the plummer potential from which the tidal torque of the gas on the comets was computed .
the typical efficiency for the formation of the inner oort cloud was 10% .
they showed that sedna s orbit could be reproduced when the central density of the cluster exceeded 10000 @xmath5 pc@xmath0 in gas and stars .
for these clusters sedna was found to be at the inner edge of the inner oort cloud . in order to reproduce the orbit of 2000 cr@xmath6 , an even higher ( central )
density was needed . however
, the orbital distribution of the inner oort cloud in these very high density clusters is found to be inconsistent with the current observations of the outer solar system ( schwamb et al . , 2010 ) , so that the clusters where sedna is at the inner edge are preferred .
+ in a similar study , kaib & quinn ( 2008 ) studied inner oort cloud formation in an open cluster with stellar number densities ranging from 10 pc@xmath0 to 100 pc@xmath0 .
kaib & quinn ( 2008 ) modelled the effect of the stars using the same approach as dones et al .
( 2004 ) and bdl6 .
the maximum cluster life time was set to 100 myr and the density of the stars in the cluster decayed linear with time , to account for mass loss by mutual scattering of the stars .
when the cluster had completely disappeared kaib & quinn ( 2008 ) continued their oort cloud simulations until the age of the solar system , something bdl6 did not do .
quantitatively their results were similar to bdl6 and they were able to reproduce sedna when the stellar number density exceeded 30 pc@xmath0 .
+ in a recent publication attempting to solve some of the outstanding problems associated with the oort cloud as a whole , levison et al .
( 2010 ) investigated the capture by the sun of comets from other stars .
they simulated embedded clusters ranging from 30 to 300 stars with a star formation efficiency of 10% to 30% .
they placed a disc of comets around each star with random orientation and orbits with @xmath10 au and semi - major axes ranging from 1000 to 5000 au .
the whole system of stars and comets was simulated until the median spacing between the stars became 500000 au . from these numerical simulations
levison et al . ( 2010 ) concluded that the capture efficiency is high enough to obtain the current population of the oort cloud provided that most of the stars in the cluster contained a similar number of comets to the sun .
at least 90% of the comets in the oort cloud could be extrasolar in origin .
+ unfortunately , apart from the levison et al .
( 2010 ) study , which relied on extrasolar comets rather than indigenous comets to populate the oort cloud , all of the above works suffer from the limitation that the stars are treated as hyperbolic encounters and the sun is on a fixed orbit in the cluster
. both of these assumptions are wrong : the stars motion with respect to the sun reverse direction when their distance to the sun is of the same order as the size of the current oort cloud .
if the stellar number density in a cloud is @xmath11 pc@xmath0 , then their average nearest - neighbour distance is @xmath12 pc , and taking a typical density of @xmath13 pc@xmath0 yields @xmath14 pc or 42000 au .
secondly , mutual scattering among the stars changes their orbits and some end up on highly elliptical orbits on their way to being ejected .
hence the assumption of a static orbit is no longer valid .
a third issue is gas removal .
bdl6 stopped their simulations after 3 myr by assuming that at this time the sun left the cluster .
kaib & quinn ( 2008 ) decreased the density of their fictitious open cluster linearly and it was gone after 100 myr .
levison et al . (
2010 ) made the gas go away exponentially with an e - folding time of 10000 yr .
fourth , the bdl6 study relied on very high central gas densities in order to torque comets under the dynamical control of jupiter into the inner oort cloud before they were ejected . while bdl6 argued that the densities that were chosen are in agreement with the peak densities observed in some embedded clusters ( gutermuth et al .
, 2005 ) , their initial conditions can be improved by using more recent observational data and better models for the embedded clusters .
any reasonable model of the formation of the inner oort cloud in a star cluster environment has to take the above issues into account .
the aim of this paper , therefore , is to re - investigate the formation of the inner oort cloud in a cluster environment using ( i ) a better model for the embedded star cluster which best matches the current observations , in particular the initial conditions for the stars , gas and the gas dispersal , and ( ii ) a computer code that can handle stars , planets and comets at the same time so that there is no need to rely on the assumption of hyperbolic stellar encounters .
+ levison et al .
( 2010 ) used a computer code , based on symba ( duncan et al . , 1998 ) , that was able to integrate both the comets and stars symplectically and self - consistently without relying on assumptions of hyperbolic flybys .
in this study we shall use their code .
this paper is divided as follows . in section 2
we summarise some of the basic properties of embedded star clusters that we need for our simulations as inferred from observations .
section 3 deals with the initial conditions and methods of our numerical simulations . in section 4
we present the properties of the inner oort cloud resulting from the numerical simulations . in section 5
we compare these results with recent observations of the outer solar system .
in section 6 we present our discussion , followed by the conclusions in section 7 .
in this section we discuss the models and parameters that we employed for the simulation of the embedded star clusters .
we use the data of embedded star clusters within 2 kpc of the sun from lada & lada ( 2003 ) and gutermuth et al .
. adams et al .
( 2006 ) use the lada & lada ( 2003 ) data and find that most embedded clusters have between 100 to 1000 members .
the cumulative distribution of the number of stars , @xmath15 , has a best fit @xmath16 i.e the distribution increases logarithmically with @xmath15 ( adams , 2010 ) .
+ ccccccc name & @xmath17 [ pc ] & @xmath18 [ pc ] & @xmath19 [ pc ] & @xmath20 & @xmath21 & @xmath22 + + mon r2 & 2.88 & 2.01 & 1.85 & 235 & 132 & 371 + ic 348 & 2.1 & 0.55 & 1.0 & 160 & 56 & 300 + ngc 1333 & 1.19 & 0.51 & 0.49 & 133 & 96 & 143 + ggd 1215 & 2.12 & 0.64 & 1.13 & 119 & 78 & 134 + s 106 & 4.16 & 1.26 & 0.3 & 79 & 36 & 160 + mwc 297 & 0.92 & 0.39 & 0.5 & 23 & 10 & 37 for most embedded clusters , the observed surface density is a constant ( allen et al . , 2007 )
i.e. @xmath23 is constant , where @xmath24 is the radius of the cluster .
this relation is in agreement with the observed density structure of giant molecular clouds ( blitz et al . , 2007 ) and theoretical modelling of cloud collapse ( larson , 1985 ) .
the observed constant surface density implies that the size of the cluster scales with the number of stars as @xmath25 , where @xmath26 is a scaling parameter . from the catalogue of lada & lada ( 2003 ) ,
adams et al .
( 2006 ) find a best fit @xmath27 pc where @xmath28 pc .
for their subsequent simulations adams et al .
( 2006 ) use @xmath29 pc to maximise dynamical interactions among the stars
. however , the radii of embedded clusters presented in lada & lada ( 2003 ) might only be valid for the cores of the clusters that are listed .
recently gutermuth et al .
( 2009 ) performed a _
study of a large sample of embedded clusters , some of which are also listed in lada & lada ( 2003 ) .
gutermuth et al .
( 2009 ) characterise each cluster by a core and an extended halo. using nearest - neighbour distance counts and assigning a radius to the cluster as being half the distance between the two farthest stars , a best fit through their data for the halos yields @xmath30 where @xmath31 pc and the exponent @xmath32 . for the cores @xmath33 with @xmath34 pc and @xmath35 .
the fitting parameters and their error bars are displayed in fig .
the common clusters of gutermuth et al .
( 2009 ) and lada & lada ( 2003 ) for which radii are available are listed in table [ gll ] .
the columns are : name , total radius from gutermuth et al .
( 2009 ) , core radius from gutermuth et al .
( 2009 ) , radius from lada & lada ( 2003 ) , total number of stars from gutermuth et al .
( 2009 ) , number of stars in the core from gutermuth et al .
( 2009 ) and number of stars listed in lada & lada ( 2003 ) .
clusters which are common to both catalogues but have no radius listed in lada & lada ( 2003 ) are afgl 490 , ic 5136 , cep c , lh h@xmath36 101 , serpens , cep a , l988-e and r cra . as can be seen from table [ gll ] , most of the sizes listed in lada & lada ( 2003 ) are close to the size of the cores listed in gutermuth et al .
( 2009 ) , or are intermediate between the core and the halo . in any case , the cluster sizes in lada & lada ( 2003 ) are systematically smaller than in gutermuth et al .
( 2009 ) , and the fit through the core data of gutermuth et al .
( 2009 ) is compatible with the fit through the data of lada & lada ( 2003 ) , but the halos are not . + the best fits seem to indicate that @xmath37 with @xmath38 $ ] . however , the @xmath39 relation is based on the assumption that the column or surface density of stars is more or less constant and is an artefact of the way the stars are counted .
most star identification algorithms rely on the density - weighted nearest - neighbour method of casertano & hut ( 1985 ) , or the minimum spanning tree method ( e.g. graham & hell , 1985 ) .
both are often employed to identify star clusters ( bastian et al . , 2007 , 2009 ; cartwright & whitworth , 2004 ; gutermuth et al . , 2009 ; schmeja & klessen , 2006 ) .
however , in all cases the cluster radius is defined as the radius of a circle with the same area as the projected cluster ( schmeja , 2011 ) .
all methods truncate the size of the cluster when the projected distance between two neighbouring stars is larger than some threshold value , which is equivalent to assuming that the surface density inside said circle is more or less constant .
thus , the size of the cluster then obviously scales as @xmath39 .
however , the @xmath39 relation appears in disagreement with another observation , and that is that the average stellar number density in these clusters is more or less constant ( carpenter , 2000 ; lada & lada , 2003 ; proszkow & adams , 2009 ) , with a median number density of @xmath40 pc@xmath0 . from the clusters listed in lada & lada ( 2003 ) and gutermuth et al .
( 2009 ) , we compute the median value for the halos of gutermuth et al .
( 2009 ) to be @xmath41 pc@xmath0 while for the cores it is @xmath42 pc@xmath0 , comparable to the value listed earlier .
however , the average stellar number density from one cluster to the next can vary by approximately an order of magnitude .
this true for both cores and halos .
thus the values quoted above should be interpreted as indicative only . + it is easy to verify whether the observation of more or less constant stellar number density is consistent with the relation between the cluster s size and the number of stars .
the total number of stars in the cluster , apart from a constant , is @xmath43 , where @xmath44 is the average stellar number density .
substituting @xmath45 we have @xmath46 , which is a constant only if @xmath47 .
thus it seems sensible to adopt a cluster size that scales as @xmath48 . in order to determine
whether or not this scaling makes sense , we turn to observations of open and globular star clusters for guidance .
+ king ( 1962 ) proposed that the size of a star cluster , whether open or globular , is given by its tidal radius , @xmath49 . at this distance from the centre ,
the tidal effects of the milky way galaxy start to dominate over the self gravity of the cluster .
king ( 1962 ) states that @xmath50 where @xmath51 is the gravitational constant , @xmath52 is the total mass of the cluster and @xmath53 and @xmath54 are the oort constants ( e.g. binney & tremaine , 1987 ) . the tidal radius scales as @xmath48 because @xmath55 .
when adopting a flat galactic rotation curve with angular velocity @xmath56 km s@xmath1 kpc@xmath1 ( macmillan & binney , 2010 ) , we have @xmath57 km s@xmath1 kpc@xmath1 and so @xmath58 pc , which is much larger than the halo sizes of lada & lada ( 2003 ) and gutermuth et al .
this discrepancy is most likely caused by the fact that for embedded clusters the background density is not that of the galactic disc , but rather of the surrounding molecular cloud .
typical densities of molecular clouds are some @xmath59 m@xmath60 pc@xmath0 , so that the tidal radius in equation ( [ rt ] ) above should be divided by @xmath61 i.e. @xmath62 pc .
this result is similar to and compatible with the halo distances obtained by gutermuth et al .
the shape of the zero - velocity curves of the cluster reduce the tidal radius even further ( innanen et al . , 1983 ) to @xmath63 pc .
given the uncertainties in the observed properties of the clusters and in the size versus number of stars , we shall anchor the value of @xmath49 for @xmath64 to the best - fit halo value of gutermuth et al .
( 2009 ) and thus use @xmath65 pc for the size of the cluster .
+ open and globular star clusters usually show a dense core with more or less constant surface brightness and then an extended halo where the surface density falls off ( king , 1962 ) , typically as @xmath66 , where @xmath67 ( king , 1962 ; elson et al . , 1987 ) .
observers generally define the cluster core radius , @xmath68 , as the distance from the cluster centre at which the surface brightness drops by a factor of two from the central value .
theorists , however , often use a definition based on the central density , @xmath69 , and central velocity dispersion , @xmath70 .
the core radius is then given by ( king , 1966 ) @xmath71 which we shall use here .
for most cluster models the core radius corresponds roughly to where the stellar volume density has decreased by a factor of 3 .
a second method computes a local density using a star s nearest neighbours ( casertano & hut 1985 ) , in which the core radius becomes a density - weighted quantity obtained from the root - mean - square stellar distances .
we refer to portegies zwart et al .
( 2010 ) for a more in - depth discussion on how the core radius is defined and measured .
king ( 1962 ) defines the concentration ratio of the cluster as @xmath72 , where the log is with base ten ; we shall use the non - logarithmic form @xmath73 here . for young open clusters ( piskunov et al . ,
2008 ) the value of @xmath74 is typically 3 to 6 .
the embedded cluster table of gutermuth et al .
( 2009 ) yields similar values . for the young open , non - relaxed cluster
ngc 6611 , whose estimated age is 1.3 myr , @xmath75 ( bonatto et al . , 2006 ) .
all of these observations suggest that @xmath74 ranges from approximately 3 to 10 , and thus these clusters have fairly shallow profiles and central potential wells
. we shall use @xmath76 and @xmath77 in our simulations .
+ inside the cluster the volume density scales as @xmath78 where @xmath79 ( schmeja & klessen , 2006 ; schmeja et al . , 2008 ; andr et al . , 2007 ) , with values in the range 0 to 1 being the most common .
it is well known that the density of most globular clusters are best fitted with king profiles ( king , 1966 ) , which have a well - defined core and halo . inside the core
the density is more or less constant while outside the core the density falls off quickly ( portegies zwart et al . , 2010 ) .
however , it is unclear if the king profiles are suitable for young / embedded clusters ( portegies zwart et al . ,
since the potential for the king models can not be written in closed form as a function of distance from the centre , which we need in our computer code , we prefer not to use these models .
there exist various alternatives in the literature to compute the density and potential . for spherical galaxies the models by dehnen ( 1993 ) and tremaine et al .
( 1994 ) are often used .
the volume density of these profiles are @xmath80 where @xmath52 is the total mass in gas and stars , @xmath81 is a parameter radius and @xmath82 measures the density concentration at the centre .
the density profiles of jaffe ( 1983 ) and hernquist ( 1990 ) are the cases with @xmath83 and @xmath84 respectively . in their cluster simulations
adams et al .
( 2006 ) use the density profile of hernquist ( 1990 ) , for which the density is given by @xmath85 where @xmath86 is the hernquist radius .
the corresponding potential is @xmath87 adams et al .
( 2006 ) set @xmath88 , with @xmath24 the size of the clusters obtained from lada & lada ( 2003 ) .
the total mass can be converted to a central density through @xmath89 .
the total mass inside @xmath86 is @xmath90 and here @xmath91 . in order to model clusters where @xmath92 close to the centre
, one could use a dehnen profile with @xmath93 .
however , we decided to settle for the density profile of plummer ( 1911 ) , which is widely used in star cluster simulations because of its simplicity ( e.g. aarseth et al . 1974 ; baumgardt & kroupa , 2007 ; kroupa et al . , 2001 ) .
its density profile is given by @xmath94 where the central density is @xmath95 and the density at the centre scales as @xmath92 . here
@xmath96 is the plummer radius .
the potential is given by @xmath97 here we shall use both the hernquist and plummer distributions only for their simplicity and ability to adequately reproduce the observed density structure in the centre of the cluster . + an additional quantity to address is the magnitude of the velocity dispersion within the clusters .
observations indicate that in the youngest embedded clusters the velocities of starless clumps and young stellar objects are a fraction of the virial value .
thus , the orbits of stars are mostly radial .
hydrodynamical simulations of cluster formation from dynamically hot gas results in the formation of young stellar objects that have speeds comparable to the sound speed , which are much lower than the virial value ( bate et al . , 2003 ) .
an example of a cluster with sub - virial speeds is l1688 , part of the @xmath98 oph complex , where the velocities of the stars are approximately 30% of the virial value ( andr et al .
similar results are found in @xmath98 oph a ( di francesco et al . , 2004 ) at 50% of the virial value , while ngc 2264 ( peretto et al . , 2006 ) and ngc 1333 ( walsh et al . , 2007 )
have even lower values .
indeed , some of the pre - stellar cumps and some young stellar objects in these clusters appear to exhibit a collapse because of the low speeds .
the deviation from virial equilibrium of a cluster is quantified by the parameter @xmath99 , which is the ratio between the total kinetic energy and total potential energy ; virial systems have @xmath100 . for our purpose
we shall consider a velocity dispersion with a value between 0.3 and 0.5 of the virial value i.e. @xmath101 $ ] .
+ the last issue we discuss is mass segregation .
massive stars are believed to sink to the cluster center over a relatively long time scale , given approximately by @xmath102 , where @xmath103 is the dynamical relaxation time , and @xmath104 is the mass of the star in solar masses ( portegies zwart , 2009 ) .
of course , the massive stars can also be formed at the cluster centres , and some observational evidence ( testi et al .
2000 ; peretto et al . , 2006 ) and theoretical considerations ( bonnell & davies 1998 ; mckee & tan , 2003 ) support this point of view . however , allison et al .
( 2009 ) report very rapid mass segregation if the initial structure of the cluster is sufficiently fractal and the velocities are highly subvirial , typical for these young clusters . in their study of these highly fractal clusters with subvirial velocities ,
mass segregation was achieved for stars heavier than 5 @xmath5 and completed in approximately 1 myr .
this rapid mass segregation appears consistent with observations of the orion nebula cluster ( hillenbrand & hartmann , 1998 ; moeckel & bonnell , 2009 ) . in this study
we consider clusters which are already mass segregated for stars heavier than 5 @xmath5 .
it should be noted that the rapid mass segregation coincides with a violent relaxation phase after the gravitational collapse , both of which are a result of its subvirial velocities , as the cluster tries to reach equipartition . as it does so it shrinks in size by approximately a factor two or more ( allison et al . , 2009 ) .
in this study we do not concern ourselves with the mechanism behind mass segregation , whether it is through dynamics or by formation , but assume it has already happened for stars heavier than 5 @xmath5 .
+ now that we have discussed most of the properties of embedded clusters based on the latest observations , and constrained some of the key parameters that will be used , we next describe our numerical methods .
in this section we describe the initial conditions and methods employed for our numerical simulations .
we generate the stars in each cluster as follows .
first , the desired number of stars was chosen .
the mass of each star was then calculated randomly according to the initial mass function formulation of kroupa et al .
( 1993 ) , with the functional form @xmath105 here @xmath106 is a number chosen randomly on the interval @xmath107 and @xmath108 is the mass of the star in solar masses .
the average stellar mass is then @xmath109 @xmath5 .
the total mass of the stars is approximately @xmath110 , and is valid for large @xmath15 .
no primordial binaries were included . in order to model the solar system we generated one star with a mass of exactly 1 @xmath5 and considered it to be the sun .
the tidal radius of the cluster was computed as @xmath111 pc , and the core radius is either @xmath112 or @xmath113 . for the plummer profile the plummer radius , @xmath96 ,
is then computed from @xmath68 by using @xmath114 and solving for @xmath96 , resulting in @xmath115 . for the hernquist profile ,
the central velocity dispersion is 0 , so we use @xmath116 and solve for @xmath86 to find @xmath117 .
+ the stars in the cluster are subjected to three forces . the first is their mutual gravitational interaction .
the second is caused by the galactic tide and bulge , and is modelled according to the formulation of levison et al .
( 2001 ) but with the oort constants set at @xmath57 km s@xmath1 kpc@xmath1 respectively ( macmillan & binney , 2010 ) .
the third force is caused by the gas that is present in the cluster , whose density profile is also modelled either by the hernquist or plummer distribution , with the same values of @xmath86 or @xmath96 used for the stars and gas . the mass in gas
is related to the total mass in stars , @xmath104 and the star formation efficiency ( sfe , @xmath118 ) by @xmath119 .
the sfe was set either to 0.1 or to 0.25 , which mostly brackets the observed range ( lada & lada , 2003 ) .
+ the magnitude of the position and velocity vectors of the stars are generated from the isotropic energy distribution functions of either the hernquist or plummer profiles using a von neumann rejection technique ( press et al . , 1992 ) .
for the plummer model we followed aarseth et al .
they solve the equation @xmath120 for @xmath121 for each star , with @xmath122 and @xmath123 .
this results in successive values of @xmath124 being evenly spaced , which appears to be fine for the heavy stars that have settled in the centre , but is artificial for the other stars .
thus we decided to adopt this method for the heavy , segregated stars but replaced @xmath106 by a random number on the interval between @xmath125 for the other stars . here
@xmath126 is determined by @xmath127 . for the hernquist model
we followed the same procedure .
the singularity of the distribution function of the hernquist model at energy @xmath128 is avoided by noting that @xmath129 , where @xmath130 is the hernquist distribution function and @xmath131 is the velocity of a star .
we use the latter formulation with the von neumann technique ( d. c. heggie , personal communication ) .
the position and velocity vectors were calculated with random orientation . in order to avoid the sun leaving the cluster immediately ,
we rigged the system by requiring that the sun moves inwards if it is farther then @xmath132 from the centre when the simulation is started .
this was accomplished by requiring that for the sun @xmath133 if @xmath134 . + after generating the velocities and positions from the distribution function the kinetic energy of the stars was reduced .
a single value for the virial parameter @xmath99 for each simulation was randomly computed on the interval @xmath101 $ ] , and the kinetic energy of each star was then reduced accordingly .
the maximum speed of the stars is @xmath135 where @xmath136 is the local escape speed and @xmath137 is the gravitational potential of the hernquist or plummer sphere .
the stars low kinetic energy implies that the stars are initially on nearly radial orbits and exhibit almost a free fall ( which justifies having the sun move inwards at first ) .
the resulting set of stellar positions and velocities was then used as the starting conditions in our full simulations .
+ in the above procedure , all stars were assumed to have formed at the same time .
there is an ongoing debate about whether or not this is true , but based on observations of ngc 2264 peretto et al .
( 2006 ) favour the turbulent formation mechanism of mckee & tan ( 2003 ) , in which most stars form within @xmath138 yr , with the heaviest stars in the centre before the initial collapse of the cluster .
palla & stahler ( 1999 ) report a similar conclusion based on the orion cluster while murray ( 2011 ) states that most stars form within the free - fall time of the cluster .
these results suggest that most stars in the cluster form within a short time of each other , justifying our procedure .
+ the next thing to model is the decay of the gas .
the early work by lada et al . ( 1984 ) removed the gas either instantaneously or on a time scale ranging from three to four crossing times .
the crossing time is defined as the time it takes for a star to cross the whole cluster i.e. @xmath139 , where @xmath140 is the velocity dispersion . for our clusters @xmath141 myr .
adams et al .
( 2006 ) kept the gas density constant and then removed it instantaneously after 5 myr , independent of the properties of the cluster .
baumgardt & kroupa ( 2007 ) removed the gas on a time scale from zero to ten crossing times .
proszkow & adams ( 2009 ) removed the gas instantaneously at times ranging from 1 myr to 7 myr , to account for the spread in observed embedded cluster lifetimes .
levison et al . ( 2010 ) kept the gas density constant for 3 myr and removed it exponentially with an e - folding time of 10000 yr .
thus , previous studies show great variability in their choice of the time of gas removal and its decay rate .
an initially bound cluster responds to external perturbations and changes in the potential in approximately a crossing time and thus this time scale serves as a bench mark .
kroupa ( 2000 ) reports that typically the gas in an embedded cluster is removed on a crossing time
. however , for our purposes we do not want to end up with a bound system of which the sun is a potential member , and thus we would like to take the gas away quickly to ensure that the sun escapes from the system .
even though proszkow & adams ( 2009 ) took the gas away instantaneously , they frequently found that bound systems remained .
the strongest dependency appears to be on the star formation efficiency , @xmath118 , but even when setting @xmath142 , they reported that some 15% of systems remained bound after the gas went away .
ultimately , the gas in the cluster is removed because stellar winds from heavy stars create h@xmath36 bubbles and supernovae .
stellar winds cause rapid outflow of h@xmath36 bubbles , typically with velocities of @xmath143 km s@xmath1 ( whitmore et al . , 1999 ) .
this leads to a removal time scale of @xmath144 yr , for clusters with @xmath145 $ ] .
given this rapid time scale , we have decided to proceed as follows : the gas is kept at its initial density for a time of either 2 myr or 4 myr . given that the typical lifetime of an embedded cluster is some 5 myr ( lada & lada , 2003 ) with a maximum of 10 myr , and that circumstellar discs have a median lifetime of 3 myr ( currie et al .
, 2008 ) , and that jupiter and saturn had to form in this time ( lissauer & stevenson , 2007 ) , a typical cluster lifetime after the formation of jupiter and saturn of 1 myr to 5 myr is reasonable .
after the constant density phase , the density of the gas decays exponentially with an e - folding time of @xmath146 yr .
+ star clusters are not well - suited to the symplectic integration methods typically used to follow the orbits of comets about stars ( spurzem et al . , 2009 ) .
these methods split the hamiltonian into the sum of a part representing keplerian motion around a fixed star and a part representing the non - keplerian perturbations from other stars ( wisdom & holman , 1991 ) .
however , we find that the orbits of the stars and other bodies are efficiently computed using the standard ` leap - frog ' integrator , which is a second - order symplectic integrator where the hamiltonian is split into the sum of a part representing the kinetic energy and one representing the gravitational potential energy . as described in duncan et al .
( 1998 ) , the key idea is to incorporate a multiple time step symplectic method such that whenever two or more stars , or a comet and star(s ) , or comet and planet suffer close encounters , the time step for the relevant bodies is recursively divided to whatever level is required to resolve the encounter . in other words ,
the code splits the encounter into a series of shells which duncan et al .
( 1998 ) place apart as @xmath147 and the time step is divided by 3 when crossing into the next shell . since these requirements are already implemented in the symba integration package ( duncan et al . ,
1998 ) , we modified it to use the leap - frog scheme . from now on this new code will be referred to as symbac . for a description of tests of the code without planets
we refer to levison et al .
+ including the jovian planets orbiting the sun in symbac proved to be tricky .
the leap - frog integration scheme is not very accurate for kepler orbits unless one takes of the order of 1000 steps per orbit ( kokubo et al . , 1998 ) .
however , even with such a small time step it introduces a secular drift in the argument of pericentre , @xmath148 , of the planets ( kokubo et al . , 1998 ) .
this drifting in @xmath148 changes the secular properties of the planets and it is easy to grow the eccentricities of jupiter and saturn to crossing values .
one solution to this problem is to invent a new symplectic method to handle both kepler orbits and stellar encounters , or to search for a set of parameters that keep the planets stable .
we chose to do the latter .
it turned out that it is imperative to keep jupiter and saturn on the same shell , and that the shells are scaled as @xmath149 and the time step between shells is divided by 6 .
the first shell around the sun is set to 2916 au and the time step is 100 yr , and is used for all of the calculations that follow . since the typical spacing between the stars is much larger than 3000 au , encounters between stars are relatively rare and the code does not often need to resort to the recursive encounter routines apart from in the beginning when there are many comets close to the sun .
hence the simulations we performed were finished in mere hours rather than days or weeks .
as in bdl6 we do not include uranus and neptune , because their formation mechanism is still not well understood and the time scale of their formation is not well constrained ( goldreich et al . , 2004 ) .
+ comets were added only around the sun as bodies with masses thirteen orders of magnitude smaller than the stars and planets .
the stars and planets mutually interact , but the comets are only affected by the stars and the planets , and not by each other .
the comets were started on orbits between 4.5 au and 12 au with eccentricities and inclinations between 0 and 0.01 ( radians ) .
a total of 4000 comets were used in each simulation . in order to hasten the evolution , we removed comets from the simulation when they were farther than 1.5@xmath49 from the cluster centre and unbound . in order to keep the calculations traceable , and because we are only interested in the formation of the inner oort cloud with indigenous comets , no comets were added around other stars , unlike in levison et al .
the simulations were stopped either after some maximum time ( 15 myr ) or when the sun was farther than 1.5@xmath49 from the cluster centre after the gas began to evaporate .
+ the last parameter to choose is the number of stars in the cluster . we have chosen seven individual values which bracket the approximate observed range of embedded clusters : 50 , 100 , 250 , 350 , 500 , 750 and 1000 .
a total of 40 simulations were performed for each cluster , with 10 for each combination of the time the gas density remained constant , @xmath150 , and the concentration @xmath74 . together with the two values of the sfe and the two potential / density pairs , this results in a total of 1120 simulations , which were all performed on the crimson beowulf cluster at the observatoire de la cte dazur . in this subsection
we give an example of the evolution of one of the many star clusters that we simulated .
the cluster is of the hernquist type with @xmath151 , @xmath152 , @xmath153 myr and the concentration parameter @xmath77 .
figures [ cl1 ] and [ cl2 ] show snapshots of the stars in the @xmath154-plane at times indicated in the top - right corners of the panels .
the sun has been coloured in green to distinguish it from the other stars , and the size of the dots correlates to the mass of the star as @xmath155 .
+ the top - left panel of fig .
[ cl1 ] depicts the initial positions of the stars , projected onto the @xmath154-plane .
these were generated according to the methods described above .
since the stars are all sub - virial , the cluster undergoes a state of collapse , through a process called violent relaxation ( lynden - bell , 1967 ) .
some embedded clusters are observed to be in this state of collapse ( peretto et al . , 2006 ; andr et al . , 2007 ) .
as one can see in the top - right panel of fig .
[ cl1 ] , the cluster has shrunk significantly and the density in the core is much higher than it was initially . after about 1.4 myr
, the cluster has reached its smallest size and maximum core density .
the cluster has a size of approximately 1 pc .
after this maximum compression of the core , the cluster expands again ( bottom - right panel ) and some of the stars are either unbound or on very elongated orbits . continuing on to fig .
[ cl2 ] we see that the cluster continues to expand and reaches some sort of a steady - state . at 4
myr the gas goes away and in the bottom two panels of fig .
[ cl2 ] we see the stars escaping from the system .
this sequence of events , in which the cluster collapses and then expands again , occurs in all our simulations .
+ now that we have given a brief overview of the dynamical evolution of the star clusters , we turn to the results of our simulations of stars and comets .
in this section we present the results of our numerical simulations .
many properties of the inner oort cloud that are formed in this way are similar to earlier results presented in fernndez & brunni ( 2000 ) , bdl6 and kaib & quinn ( 2008 ) , such as the the cloud being nearly isotropic apart from the inner 10% or so .
we shall not repeat all of these here but instead only focus on the key aspects and how these scale with the cluster properties : the distribution in semi - major axis , inclination and perihelion , the location of the inner edge and the formation efficiency . during this investigation it became clear that the parameter space is very large ,
so we shall try to reduce it first .
when examining the output from the simulations , it turned out that the initial virial ratio , @xmath99 , does not impact the results beyond the statistical noise one expects from one simulation to another .
this result is surprising , because the value of @xmath99 determines how much the cluster shrinks during the initial collapse and violent relaxation .
thus , in what follows we consider the results to be averaged over the value of @xmath99 .
this leaves us with the number of stars , @xmath15 , the star - formation efficiency , @xmath118 , the time the gas was removed , @xmath150 , the concentration , @xmath74 , and the density profile ( plummer or hernquist ) . for reference ,
a comet is defined to be in the cloud if it satisfies both @xmath156 au and @xmath157 au . + a simple way to depict the size and distance of the inner oort cloud to the sun is to plot the cumulative semi - major axis distribution of comets that are in the cloud .
we have plotted this distribution for hernquist clusters with 50 , 250 , 500 and 1000 stars ( fig . [ cumah ] ) and plummer ( fig . [ cumap ] ) distributions respectively .
both figures correspond to clusters with an sfe of 10% .
the red lines are for clusters with @xmath158 myr and @xmath76 , green lines have @xmath153 myr and @xmath76 , the blue lines correspond to the parameters @xmath158 myr with @xmath77 , while the magenta lines depict cases with @xmath153 myr and @xmath77 .
we will use these @xmath159-colour correlations throughout the paper , unless specified otherwise .
the plots show a few interesting features that require further discussion .
+ the first is that , generally but not exclusively , the magenta lines precede the blue lines , which precede the green lines , which in turn precede the red lines . in other words ,
the inner oort clouds formed above become more centrally condensed when both @xmath150 and @xmath74 increase .
this is easy to imagine .
as the sun resides in the cluster for longer times , the outermost comets are stripped away by encounters with the other stars .
only the tightest - bound comets survive these encounters over long times because stars need to come ever closer to destabilise the close - in comets .
in addition , as the sun spends more time in the cluster , the tidal forces from the gas and stars have more time to torque the comets perihelia away from the planets . since this time scale is @xmath160 ( duncan et al . , 1987 ) ,
spending more time in the cluster will torque comets at smaller semi - major axis .
the same argument holds for the torquing of the comets perihelia by stellar encounters . secondly , ignoring the red curve in the upper - right panel of fig .
[ cumah ] , where the sun suffered a very close encounter with another star early on , the range of the curves of the same colour are similar among all the panels of each model .
however , when comparing the hernquist and plummer clouds , for the hernquist model the range of the clouds is from approximately 200 au to 200000 au ( where we truncated it ) , while for plummer the clouds are farther away , starting at about 800 au .
this makes sense : the central density in the hernquist model is higher than that in the plummer model and thus as the sun flies through the centre of the cluster the torquing by the gas and the encounters with the other stars in the hernquist model are more violent than in the plummer model .
a third interesting feature is that the cumulative distributions have a similar shape and appear just different in their median values .
it appears as if both stochastic effects from one simulation to the next and the number of stars in the cluster affects the final distribution .
kaib & quinn ( 2008 ) also reported that their results suffered from strong stochastic effects from one simulation to the next .
we examined the cumulative semi - major axis distributions for clusters with an sfe of 0.25 and conclude that they are very similar to the figures shown above . + in order to characterise the oort clouds better , in fig . [ ocfig ]
we have plotted a few sample oort clouds in semi - major axis - pericentre space .
the data are accumulated over a series of clusters with different @xmath15 .
the top two panels pertain to hernquist models while the bottom two panels are for plummer clusters .
note that the plummer clusters are less centrally concentrated than the hernquist clusters .
note also that unlike the traditional oort cloud , which has an inner edge at approximately 2000 au ( e.g. dones et al .
, 2004 , kaib & quinn , 2008 ) the oort cloud formed in the cluster environment can extend all the way down to @xmath161200 au so that the cluster environment produces an oort cloud that is much more centrally condensed . +
the differences among the cumulative distributions are a measure of the concentration of the cloud .
unfortunately there is no definite way to define this from the cumulative distributions of the semi - major axis .
however , we might turn to the theory of star clusters for guidance .
in analogy with star clusters , a useful way to characterise the central concentration of the inner oort cloud is by considering the number density of the comets as a function of semi - major axis .
this is similar to counting the stars in a cluster as a function of their projected distance to the centre .
the number density of the comets as a function of the semi - major axis , @xmath162 , can be well approximated by a power law when far from the sun , and the slope is usually -@xmath163 or @xmath164 ( duncan et al .
, 1987 ; dones et al . , 2004 ;
bdl6 ; kaib & quinn , 2008 ) , while close to the sun the density is usually flat ( e.g. brasser et al . ,
this suggests a best fit through the number density profile of the form @xmath165 , where @xmath166 ( dehnen , 1993 ; tremaine et al . , 1994 ) measures the central number density of the cloud , @xmath167 is the total number of comets in the cloud and @xmath168 is a parameter that measures the central concentration .
the lower its value , the more centrally condensed the cloud is , and probably the more centrally condensed or long - lived the sun s birth cluster was .
the number of comets as a function of semi - major axis is then @xmath169 , which is a reasonable approximation to the curves presented in figs .
[ cumah ] and [ cumap ] . the half - mass radius @xmath170 .
in addition @xmath171 , so that most of the comets are in the halo rather than in the core .
+ we have performed a series of fits to the data to characterise the range and typical value of @xmath168 .
while performing the fits , we encountered cases where a better expression was @xmath172 , where @xmath173 ( elson et al . , 1987 ) and @xmath174 .
this distribution turns quicker from @xmath175 to @xmath176 around @xmath168 than the dehnen distribution .
however , it is slightly more compressed , because the half - mass radius is at @xmath177 , though @xmath178 . as an example , for the red curve in the top - right panel in fig .
[ cumah ] , in which the inner oort cloud is rather concentrated towards the centre , we have @xmath179 au , but for the red curve in the upper - left panel @xmath180 au . for the same curves using plummer profiles , @xmath181 au and @xmath182 au respectively .
the density profiles for these examples , and their best fits , are given in fig .
we find large variations for @xmath168 as a function of @xmath15 for one set of values of @xmath150 and @xmath74 .
the value of @xmath168 depends more sensitively on @xmath74 than on @xmath150 , and decreases as either of these increase .
we found no correlation between the value of @xmath168 and @xmath99 , suggesting that the previous two parameters have a stronger effect on the final structure of the inner oort cloud .
for reference , for the classical oort cloud that was formed in the current galactic environment ( dones et al . , 2004 ;
kaib & quinn , 2008 ) @xmath183 au .
+ we have plotted the values of @xmath168 , averaged over the various combinations of @xmath150 and @xmath74 as a function of @xmath184 in the top two panels of fig .
[ seff ] for an sfe of 10% , and in fig .
[ seff2 ] for clusters with sfe of 25% .
error bars denote the maximum and minimum values that we obtained from our data for each value of @xmath15 and are a proxy for the values of @xmath150 , @xmath74 ( and @xmath99 ) and how stochastic the dynamics is .
the error bars indicate that our previous assessment of the variations the cumulative semi - major axis distribution being due to stochastic effects was essentially correct .
the left panels refer to hernquist clusters while the right panels refer to plummer models .
as one can see , there is a trend for @xmath168 to increase with @xmath15 .
best fits through the data yield a steeper slope for plummer than for hernquist in both cases .
even though the fitting errors are large , there _ is _ a trend for @xmath168 to increase with @xmath15 .
this trend appears to be systematic and we get back to it later .
+ in bdl6 we concluded that the median distance of the inner oort cloud to the sun scales with the cluster density as @xmath185 .
in bdl6 we said that this combination of @xmath186 and @xmath98 is found in the torquing time of the pericentre , @xmath187 , and thus if the product @xmath188 is constant , then @xmath187 must be a constant . from bdl6
we find for the inner edge @xmath189 au .
thus , to get ( 90377 ) sedna , an inner oort cloud dwarf planet with @xmath190 au ( brown et al . , 2004 ) , we need a density of @xmath191 @xmath5 pc@xmath0 .
do we see this occur in our clusters and does it occur for long enough to torque objects onto sedna - like orbits ?
+ in the bottom panels of fig .
[ seff ] we plot the value of the cumulative semi - major axis distribution at sedna s location ( @xmath192 au ) , @xmath193 , as a function of @xmath184 .
the bottom - left panel once again refers to hernquist clusters while the bottom - right panel are the plummer models .
the data is averaged over @xmath150 , @xmath74 and @xmath99 , which are incorporated into the error bars .
once again these denote the maximum and minimum values that we observed . for the hernquist case
the data can be fit as @xmath194 , with errors of 17% and 25% on the coefficients . for plummer and @xmath152 a sedna
is found in only 70% of the cases , but then only typically at the 1% level of the cloud , with no trend in @xmath15 . when averaging over all @xmath15 , the average values of @xmath193 for hernquist are 4.8% for @xmath195 , 7.7% for ( 2,6 ) , 2.7% for ( 4,3 ) and 7.6% for ( 4,6 ) , so that @xmath193 depends more sensitively on @xmath74 than on @xmath150 .
we report that in most cases @xmath193 increases as either @xmath74 or @xmath150 increases , but we found no systematic correlation between @xmath193 and @xmath99 .
+ the trend of @xmath193 decreasing and @xmath168 increasing with @xmath15 can be explained as follows . in order to obtain a sedna , the sun needs to ( i ) temporarily pass through an environment with very high density , ( ii ) pass through that part of the cluster where the torquing on the comet is at a maximum , or ( iii ) experience a close stellar passage . from bdl6
we know that in a plummer cluster the torquing on the comet is a maximum when @xmath196 .
unfortunately , this does not coincide with the maximum density , and indeed the torque on a comet vanishes at the centre of the cluster .
thus , to produce sedna in a plummer cluster requires a close stellar passage .
in contrast , for the hernquist model both the density and the torque are a maximum at the centre of the cluster , so that the only requirement is that the sun passes close to the centre . however , the possible trajectories that the sun can have to get close to the centre and experience the spike in density and torque that are needed , decrease with increasing @xmath15 because the size of the cluster itself increases . in other words ,
the orbit of the sun needs to be more and more radial with increasing @xmath15 and thus the initial conditions for these orbits occupy a smaller region of phase space for the larger clusters than for smaller ones .
the figures above indicate that the location and existence of sedna are only mildly dependent on the cluster parameters .
+ figure [ cumi ] shows the cumulative value of the cosine of the orbital inclination in the ecliptic plane for various inner oort clouds from hernquist clusters .
as one can see , there is little variation among the distributions with either @xmath15 , @xmath150 or @xmath74 .
all clouds are predominantly prograde since the median value of @xmath197 is larger than 0 .
if the inclination distribution were isotropic , then a cumulative distribution of @xmath197 would be a straight line from 0 to 1 as @xmath197 goes from 1 to @xmath198 .
as one can see , this is not the case for most distributions and thus the inner oort cloud is not ( yet ) isotropic .
that said , the median inclination is lower in the inner parts of the cloud and higher in the outermost parts ( not shown ) .
however , even in the innermost part ( less than 10% in cumulative semi - major axis ) the median value of the inclination is typically between 45@xmath199 and 55@xmath199 .
sedna , with it s orbital inclination of 12@xmath199 , is mostly in the bottom 5% of the inclination distribution .
figure [ cumq ] plots the cumulative perihelion distribution for the same clusters .
once again the different curves are similar in each plot .
the cumulative distribution in @xmath200 scales fairly well with @xmath201 , so that @xmath202 and the differential distribution is flat in @xmath203 .
this implies that most comets are found with small perihelion relative to the semi - major axis .
this is no surprise for two reasons .
first the galactic tide has not had time to randomise the @xmath200 distribution .
second , the innermost part of the oort cloud i.e. comets with @xmath204 au , are barely affected by the galactic tides even on gyr time scales ( dones et al . , 2004 ; bdl6 )
, so we expect both the perihelion and inclination distributions to be preserved in this region .
+ when examining the inclination and perihelion distributions for the plummer clusters , they turn out to be very similar to those for the hernquist clusters .
this is to be expected because these distributions have not been evolved to the present age and thus the galactic tide has not had time to mix the inclinations and eccentricities of the comets . regarding the cumulative eccentricity distribution , which is not shown , we find the cumulative distribution scales as @xmath205 , with @xmath206 .
a thermal distribution has the cumulative @xmath207 .
the spearman rank correlation coefficient between the semi - major axis and eccentricity distributions varies from -0.07 to 0.05 , suggesting no trend of the eccentricity to either increase or decrease with increasing semi - major axis . between the semi - major axis and inclination distributions , the spearman rank correlation coefficient is approximately 0.25 , suggesting a weak trend for the inclination to increase with increasing semi - major axis .
we verified this by noting that the inner portion of the oort cloud has a lower median inclination than the outer part .
in bdl6 we argued that comets with semi - major axis @xmath204 au are barely affected by the galactic tide because the average rotation rate of their apses is governed by planetary perturbations rather than the galactic tide .
therefore , it is likely that this population has preserved its original orbital structure since its formation during the time the sun was in a cluster .
we call this population the fossilised inner oort cloud , and its most prominent member is sedna , with 2000 cr@xmath6 @xmath208 and 2004 vn@xmath209 @xmath210 being potential other members . the high - inclination objects ( 136199 ) eris and 2004 xr@xmath211 ( buffy )
are excluded because their existence can be explained by other dynamical mechanisms ( gomes , 2011 ) .
here we are interested in the orbital distribution of objects in this fossilised inner oort cloud in order to directly compare it with observations .
however , before proceeding there are a few issues that we have to consider .
+ recently schwamb et al .
( 2010 ) attempted to constrain some of the properties of the sun s birth cluster by comparing recent deep - field observations with palomar of trans - neptunian objects with the simulations presented in bdl6 .
schwamb et al .
( 2010 ) concluded that the current environment of the outer solar system is incompatible with the two densest clusters from bdl6 with a confidence of 95% or better ; they were unable to reject the next - densest clusters , with central density @xmath212 @xmath5 pc@xmath0 , for which sedna was always at the inner edge .
this should help us put some crude constraints on the cluster models employed in the current study that are compatible with the data from schwamb et al .
( 2010 ) and which are not .
re - examining the data from bdl6 we find that for the @xmath213 @xmath5 pc@xmath0 clusters sedna is located at the 3% level in cumulative semi - major axis , and the innermost object has a semi - major axis of 218 au . for the denser clusters the inner edge is closer in . however , how can we distinguish between objects that were placed on their current orbits by a stellar encounter and by planetary actions , in particular during the late epoch of planetary instability where neptune most likely temporarily resided on a highly eccentric orbit ( tsiganis et al . , 2005 )
? gomes et al .
( 2005 ) and gomes ( 2011 ) have demonstrated that the combined effect of mean - motion resonances with neptune and the kozai mechanism ( kozai , 1962 ) can place objects from the scattered disc on orbits with a high inclination ( up to 50@xmath199 ) and large pericentre distance ( more than 40 au ) , mimicking a fossilised inner oort cloud object that was perturbed by a star .
in addition , as neptune migrated and its eccentricity decreased through dynamical friction , some objects detached from is resonances or sphere of influence and now permanently reside on high - perihelion , high - inclination orbits ( gomes et al . , 2005 ; gomes , 2011 ) . however , the combined effects of mean - motion resonances and the kozai mechanism breaks down when the mean - motion resonances with neptune begin to overlap , which occurs for semi - major axes larger than 200 to 250 au ( gomes et al . , 2005 ; lykawka & mukai , 2007 ) .
thus any object with a semi - major axis longer than this value , and a pericentre distance farther than 38 au beyond which chaotic diffusion stops ( gladman et al . ,
2002 ) could be a fossilised inner oort cloud object .
therefore , for our purpose , we eliminate objects with semi - major axis shorter than 250 au and perihelia lower than 38 au .
+ the cumulative distributions of the inclination and pericentre distance for the fossilised inner oort cloud are very similar to those depicted in figs .
[ cumi ] and [ cumq]above , apart that the distribution in @xmath200 rolls over earlier than above .
we decided instead to combine all the data from the plummer simulations with sfe of 10% into one plot . in fig .
[ fossilp ] we plot , in the top - left panel , a histogram of the semi - major axis distribution for all the objects in the fossilised inner oort cloud .
the stochastic behaviour of the semi - major axis distribution is caused by the chaotic nature of the simulations and the peaks and troughs should not be considered as definitive , although the increasing trend is a systematic effect .
the top - right panel depicts a histogram of the @xmath200 distribution , and matches the @xmath203 profile discussed earlier i.e. most objects are trapped with a low value of @xmath200 , with the median value being @xmath214 au , but @xmath4 au for the hernquist clusters .
the bottom - left panel displays the inclination distribution clearly indicating the prograde bias , while the bottom - right panel plots the semi - major axis vs pericentre distance .
here we report the formation efficiency of the inner oort cloud from our numerical simulations .
we have plotted the formation efficiency as a function of the number of stars in the cluster in fig .
each filled square denotes the average efficiency for a certain combination of @xmath150 and @xmath74 and the error bars denote the maximum and minimum .
the combinations of @xmath150 and @xmath74 are , from left to right , ( 2,3 ) , ( 2,6 ) , ( 4,3 ) and ( 4,6 ) .
the left panels are for hernquist clusters with an sfe of 10% ( top ) and 25% ( bottom ) .
the right panels are for plummer clusters with the top having an sfe of 10% and the bottom 25% .
as is clear , the typical efficiency does not depend strongly on the properties of the cluster but remains at approximately 1.5% .
bdl6 reported typical efficiencies of 10% although kaib & quinn ( 2008 ) obtained efficiencies of typically 3% during their open cluster phase , which are in better agreement with our current value .
in addition , the results of kaib & quinn ( 2008 ) do not strongly depend on their cluster properties either .
we can not pinpoint the exact reason for the significant difference in the trapping efficiency between our current simulations and those presented in bdl6 although we can present a plausible argument .
+ kaib & quinn ( 2008 ) argued that their resulting inner oort clouds and efficiencies were very strongly dependent on the closest stellar passage through their oort clouds .
if this passage happened late then there would be little material left in the oort cloud or the scattered disc to refill the cloud ( levison et al . ,
in addition , in bdl6 the orbit of the sun remained fixed in the plummer potential and were obtained from a cluster in virial equilibrium , so that most orbits had a small radial excursion . since the sun stayed near the plummer radius the fluctuations in density and flux of stellar encounters that the sun experienced were limited .
here this is not the case : the sun s orbit is nearly radial so that it passes close to the cluster centre before receding back into the halo after the initial phase of violent relaxation .
thus the sun does not stay close to the plummer or hernquist radius and , especially in the latter clusters , experiences a change in background density of at least an order of magnitude .
the passages through the centre of cluster occur after 1 to 3 myr , by which time the number of comets that are being scattered by jupiter and saturn has decreased by as much as 80% .
the close encounters with massive stars near the centre of the cluster and the decrease in tidal radius around the sun strip many of its comets when it passes through the centre of the cluster . by
then there are not many comets left to resupply the oort cloud and thus the corresponding efficiency is low . this efficiency
could possibly be increased by adding uranus and neptune and having them scatter the comets in their vicinity .
however their formation mechanism and time scale is not well understood ( goldreich et al . , 2004 ) .
thus we have reason to believe that the efficiencies that we obtained here are in agreement with the expected dynamics . + is the existence of sedna in agreement with a typical trapping efficiency of 1.5% ?
levison et al . ( 2008 ) and morbidelli et al . ( 2009 ) estimate that there were approximately 1000 pluto - sized objects in the trans - neptunian disc .
since the mass in this disc is probably comparable to that in the jupiter - saturn region , we also estimate that there were some 1000 pluto - sized objects in this region . taking the size distribution for large kuiper belt objects from bernstein et al . (
2004 ) , with a cumulative slope of 3.5 , we estimate there were approximately 4000 sedna - sized objects shortly after the formation of jupiter and saturn .
if only 1.5% of these ended up in the oort cloud , this implies there are some 60 sedna - sized objects in the oort cloud , and thus we can expect to see one of these in the innermost 2% of the cloud . thus , our low formation efficiency is compatible with the detection of one sedna - like object .
however , this estimate is uncertain by factors of a few because of the lack of knowledge of the primordial number of sedna - like objects in the jupiter - saturn region .
+ in the next section we better compare the compatibility of our cluster models with a single detection of sedna .
in this section we compare the orbital distribution of sedna - like bodies produced in the above cluster environments to the observational constraints the wide - field survey of schwamb et al .
( 2010 ) ( hereafter s10 ) place on the sedna population .
s10 searched 11786 deg@xmath215 down to a mean limiting @xmath24 magnitude of @xmath16121.3 , within @xmath21630@xmath199 of the ecliptic . with the exception of sedna , no new sedna - like bodies with perihelia beyond 45
au were found despite a sensitivity out to distances of @xmath1611000 au .
for further details about the survey , we refer the reader to s10 . here
we apply the same technique and survey simulator , developed in s10 , to test whether the new cluster orbital distributions can serve as the source of the sedna population and are consistent with the single re - detection of sedna . for each cluster environment
, we compare the orbital distributions of single detections produced by the survey simulator to sedna .
we employ a modified 3-dimensional kolmogorov - smirnov ( 3-d ks ) test , detailed in s10 , which simultaneously compares the semi - major axis , inclination , and eccentricity ( @xmath186 , @xmath217 , @xmath218 ) distributions of single detections produced by each cluster environment to sedna s orbital parameters ( @xmath186=519 au , @xmath217=0.853 , @xmath218=11.9@xmath199 ) .
+ from each cluster - created orbital distribution , we randomly generate 3 million orbits from the final inner oort clouds from our simulations , obtaining the semi - major axis , inclination , and eccentricity for each inner oort cloud candidate object .
the s10 observations probe the present - day inner oort cloud population , in particular the sedna region , but the orbital distributions presented here are valid only shortly after the cluster dissipates and the sun exits the cluster .
although sedna s orbit and other objects in the fossilised inner oort cloud are dynamically protected from the effects of passing stars and galactic tides in the current solar environment , we must account for the 4.6 gyr of subsequent evolution the solar system has undergone and account for those effects that may have sculpted other objects in the sedna region since emplacement . to avoid confusion with scattered disc and detached kuiper belt objects ,
we conservatively exclude objects with perihelia smaller than 50 au and semi - major axes shorter than 250 au because these could have dynamically interacted with neptune .
we also exclude orbits with semi - major axes greater than 3000 au because these objects are capable of becoming long period comets ( kaib & quinn 2009 ) .
bdl06 obtain a value of @xmath161 2 gyr for the precession frequency of sedna ; therefore we assume that the orbits of other sedna - like objects have been randomised due to planetary effects , and thus randomly choose all other orbital angles .
+ to create our sample of single detections , we randomly assign absolute magnitudes to each of our simulated inner oort cloud populations .
due to the large uncertainties in the albedos of such a distant population , we choose to assign absolute magnitudes rather than sizes .
we assume a single power - law brightness distribution where the number of objects brighter than or equal to a given absolute magnitude , @xmath219 , is described by : @xmath220 the brightness distribution is scaled to @xmath221 , the number of bodies with an absolute magnitude brighter than or equal to sedna ( @xmath222 ) .
+ for both the hernquist and plummer cluster models we performed the 3-d ks test for a possible range of values for @xmath36 ( 0.2 - 0.82 ) , including @xmath223 and @xmath224 , the best - fit values for the hot ( @xmath225 ) and cold ( @xmath226 ) kbos respectively ( fraser et al . , 2010 ) .
each single instance of the brightness distribution can be thought of as a separate survey , and we continued sampling the brightness distribution until we created 10000 synthetic single detections of inner oort cloud objects for every cluster environment and each value of @xmath36 .
we found that for any value of @xmath36 , the 3-d k - s probability of rejection is a relatively flat value as @xmath221 increases .
therefore we restrict our analysis and discussion to @xmath221 = 1 .
+ table [ tab : cluster ] lists the results from the k - s tests for each cluster environment .
we have sampled the clusters according to sfe and number of stars , thereby crudely averaging over the concentration and @xmath150 .
for the first column the first four letters refer to the potential that was used , either hern or plum , e01 or e25 refers to the sfe being either 10% of 25% , and n50 to n1000 refers to the number of stars in the cluster . except for the plume01n100 model , which marginally produced sedna , and the plume25n100 model , _ all _ the orbital distributions are consistent with the s10 observations of a single detection of a sedna - like body , and
can not be rejected at greater than the 90@xmath227 confidence level for the entire range of @xmath36 that was tested .
the majority of the simulated cluster environments reproduced sedna s orbit at the inner edge of the distribution , emplacing the bulk of the population onto orbits with semi - major axes and perihelia greater than sedna s .
unfortunately we find no direct correlation between cluster size and the model fit , and the majority of the plummer and hernquist potential clusters are consistent with the s10 observations . though the synthetic hernquist potential populations provide a slightly better fit to the data , we are unable to reject the majority of the plummer potential clusters .
the hernquist distribution has a wider range of eccentricities at lower semi - major axes for the single produced single detections than for the plummer model .
+ given that all models appear almost equally viable in the production of sedna we must conclude that the formation of a sedna population is a natural outcome of the cluster birth scenario independent of the initial conditions and properties of the cluster , and that the small variations are just the result of stochastic effects in the cluster .
until this work the 10@xmath228 @xmath5 pc@xmath0 cluster of bdl06 was the only orbital distribution consistent with the detection of a single sedna - like body in the s10 survey . for @xmath223 and @xmath224 respectively , s10 find the best - fit values for the number of objects brighter than or equal to sedna are 595@xmath229 and 112@xmath230 for the 10@xmath228 @xmath5 pc@xmath0 cluster of bdl06 .
the plummer and hernquist distributions in this study resemble the 10@xmath228 @xmath5 pc@xmath0 cluster orbital distribution , with sedna located at the inner edge of the orbital distribution with many more objects having semi - major axes and eccentricities larger than sedna .
we expect these clusters would produce similar population estimates to the 10@xmath228 @xmath5 pc@xmath0 clusters of bdl06 . to estimate the size of the sedna population , we selected a few of the orbital distributions from each potential to examine .
for each given value of @xmath36 , we randomly assigned an absolute magnitude 50000 times to each synthetic object our survey simulator created for every value of @xmath221 . for
each @xmath221 tested , accounting for survey efficiency , the number of simulations in which , like the real survey , one object on a sedna - like orbit is detected , are tallied . using the same slopes for the brightness distribution ( @xmath223 and @xmath224 ) ,
we find similar results than s10 . in other words
, a cluster environment could emplace on the order of a hundred to a thousand planetoids brighter than or equal to sedna beyond the kuiper belt , roughly consistent with the earlier estimate of approximately 100 objects based on the efficiency of inner oort cloud formation and a crude estimate as to the number of available objects .
an order of magnitude or two more mass may reside in the sedna region than exists in the present kuiper belt . .3d ks test results for the cluster that produced single detections compared to sedna s orbit .
we report the confidence level we can reject the two distributions as drawn from the same parent population . [ cols="<,^,^,^,^,^,^ " , ]
the fact that many clusters seem to fit the currently - known structure of the outer solar system requires further elaboration . from the above simulations that we performed and the present location of sedna
, we can not constrain the number of stars in the sun s birth cluster . the only conclusion we can draw
is that almost all the clusters are consistent with the detection of one sedna - like object at the inner edge of the oort cloud , and thus the production of sedna at its current location is a generic outcome .
the observational data marginally favours the hernquist model over the plummer clusters , even though the former produces an inner oort cloud that is more centrally condensed than the latter . in order to determine which model best corresponds with the reality deeper surveys , such as lsst ( large synoptic survey telescope ) , which can probe the sedna region directly , are needed .
+ unfortunately , our simulations remain inconclusive as to what value of @xmath15 best fits the current observational data .
several previous studies favour a large value of @xmath15 , typically 1000 to 10000 , for reasons such as supernova enrichment , gas disc truncation through photo - evaporation , gas disc truncation through close stellar passages and the survival of the planets in a stellar environment ( adams , 2010 ) . on the other hand ,
gounelle & meibom ( 2008 ) argue that it is unlikely that the solar system formed in an environment similar to that of the orion nebula cluster ( onc ) , and argued instead that the sun is a second - generation star that formed after some of the stars in the first generation had already gone supernova .
after a few myr of evolution , star formation in onc - like settings occurs mainly in photo - dissociation regions where the hii region created by the central star and the surrounding molecular gas meet , which is at a few parsecs from the massive star . the fate of the onc is well illustrated by the 2 - 3 myr old cluster ngc 2244 , whose most massive star has the same spectral type as @xmath231 c ori . in ngc 2244 ,
star formation is occurring in the outskirts of the cluster , at distances 5 - 10 pc from the central o6 star .
gounelle & meibom ( 2008 ) argue that the sun formed in such a secondary environment and that some of the short - lived radio nuclei that are found in meteorites were inherited from the interstellar medium rather than supernova enrichment , or by irradiation .
the first generation cluster had to be large in order for it to have undergone several supernovae , but constraints on the second - generation are weaker .
unfortunately , at this stage we are unable to model the second - generation cluster from which the sun could have formed , and instead rely on the first generation , from which we can not make a definitive conclusion about the size and membership of the sun s cluster .
we can not rule out that the sun formed in a cluster with @xmath232 stars as advocated by some studies , but at the same time we can not rule out clusters with @xmath233 either .
most stars appear to form in clusters with @xmath234 stars ( lada & lada , 2003 ; adams et al . , 2006 ) , all of which are dynamically consistent with the observed orbital distribution of the outer solar system .
+ given the results above the natural question to ask is whether or not we could have done things differently .
we performed a comprehensive search of the literature to determine what are the best initial conditions and properties for the embedded clusters and the gas expulsion .
the only thing that we can think of is that our models do not update the mass distribution of the gas as the cluster evolves , and neither does it account for a central cavity in the gas distribution around the heaviest star(s ) .
both of these could affect the dynamics of the cluster .
to our knowledge no n - body simulations of star clusters containing gas take into account the motion of and the related changes in the gas .
doing so would most likely require a hybrid study of n - body for the stars , planets and comets , and hydrodynamics for the gas .
this would be a completely new study in star cluster dynamics far beyond the scope of this paper .
+ a second issue concerns the initial conditions of the cluster . for this study we chose to artificially have
the stars reach their final masses at the same time , while in reality this occurs over a certain time span .
the sequential star formation could prevent the early violent relaxation phase of the cluster and thus the dynamics would be milder .
adams et al .
( 2006 ) incorporated the sequential star formation by adding the stars one by one to his simulations over a time span of 1 myr .
levison et al . ( 2010 ) randomised the phases of the stars by integrating the whole cluster in the hernquist potential for 1 myr .
but , as we argued earlier , there is no definitive piece of evidence that supports either sequential star formation or having most of the stars in the cluster form at more or less the same time .
further study is needed to support either mechanism and how it would impact the outer parts of the solar system .
+ one additional issue that we need to address is the problem of when is @xmath235 ? in our simulations , @xmath235 corresponds to when jupiter and saturn have fully formed and begin to scatter the comets in their vicinity .
here we assume that their time scale of formation is short compared to the free fall time of the sun in the cluster .
the free fall time is given by @xmath236 myr .
a typical number density in the centre of the clusters is @xmath237 pc@xmath0 ( carpenter , 2000 ) , which implies @xmath238 @xmath5 pc@xmath0 . taking the sfe to be 0.1 we have @xmath239 myr , which is more or less the time when the cluster reaches it maximum density in fig .
the formation of jupiter and saturn is thought to have taken between 1 myr and 3 myr ( lissauer & stevenson , 2007 ) and observations indicate that circumstellar gas discs have a typical lifetime of 3 myr ( currie et al . , 2008 ) .
thus , it is possible that the sun had already passed through the centre of the cluster at least once before the formation of jupiter and saturn had been completed . would including jupiter , saturn and the comets after this initial passage dramatically change the outcome of our simulations ?
we do not think so because the sun will still have a radial orbit and continue to experience a few passages through the cluster centre before the gas is blown away by stellar winds .
given that the first passage deposits most of the comets in the cloud and is also the most damaging , we believe that the outcome will be similar to what we have presented here . + the last issue we would like to address is whether or not our model predicts more sednas .
as we discussed in section 5 , the existence of sedna can not rule out the current cluster model . to our knowledge
, the lsst will be able to place constraints on the model . by extrapolating expected detections from our models to a survey such as lsst with a limiting magnitude of 24 , and covering @xmath240 from the ecliptic
allows us to constrain the slope of the size distribution of objects in the sedna region .
if there are more objects in this region brighter than or equal to sedna , then we expect there to be of the order of several to a few hundred more bodies of similar size , similar to our predictions from section 5 .
however , the final outcome depends on the slope of the size distribution .
if the slope is steep then we expect there to be many more similar objects in this region , but if the size distribution is rather shallow then maybe only a single or handful of objects detectable .
the next - generation surveys such as lsst could hopefully shed some light on this issue .
we have performed a series of numerical simulations of the formation of the inner oort cloud in an embedded cluster environment .
the initial conditions and other relevant parameters of the clusters are chosen according to the most recent observations . specifically , we use two potential / density pairs for the cluster , those of plummer ( 1911 ) and hernquist ( 1990 ) , two values for the central concentration of the cluster , two values for the star formation efficiency and the time the gas is removed , and a range of values of the initial subvirial kinetic energy of the stars .
the number of stars in the clusters are 50 , 100 , 250 , 350 , 500 , 750 and 1000 .
the formation of the inner oort cloud is modelled by having jupiter and saturn scatter comets in their nearby vicinity .
uranus and neptune were not included at this stage .
we conclude the following : * the inner edge of the oort cloud ranges from approximately 100 au to a couple hundred au , while the outer edge is beyond 10000 au .
both are virtually independent on the number of stars in the cluster . *
the central concentration of the oort cloud is measured by the concentration radius , @xmath168 .
half of the mass resides within @xmath241 . for hernquist clusters @xmath168 ranges from approximately 600 au to 1500 au , while for plummer clusters it ranges from 1500 au to 4000 au .
this difference is no surprise because the hernquist clusters are more centrally condensed . for the current oort cloud which formed in the current galactic environment
the concentration radius is approximately 5000 au .
* the concentration radius increases with increasing number of stars in the cluster , so that more populated clusters form less compact inner oort clouds .
the reason for this is the smaller volume of phase space that the sun can occupy to pass close enough to the centre to experience the necessary high density and torquing to deposit the comets in the cloud . *
the location of the dwarf planet sedna at 500 au is usually within the inner 5% of the cloud for hernquist clusters , and within the innermost 2% of the cloud for plummer clusters , in agreement with the idea that it is at the inner edge of the inner oort cloud . *
almost all the orbital distributions of the clusters are consistent with a single detection of a sedna - like body , and can not be rejected at greater than the 90@xmath227 confidence level for the entire range of the size distribution slopes that were tested . in other words , from the position of sedna and a comparison with the current structure of the outer solar system we can not constrain the number of stars in the cluster . *
the inner oort clouds formed in these clusters are not isotropic , but show a slight prograde preference with a median inclination of approximately 50@xmath199 .
the inner part is significantly more prograde while the outer part is slightly retrograde . *
the perihelion distribution is flat in @xmath203 , so that most inner oort cloud objects have small perihelia compared to their semi - major axis .
* we examine the ` fossilised inner oort cloud ' , which is the region inside approximately 2000 au where the galactic tide has barely altered the orbital elements of the comets .
we find that the distributions in inclination , perihelion and semi - major axis are similar than for the whole cloud .
in this region we expect these distributions to be preserved because the galactic tide does not randomise the inclinations and eccentricities .
the median perihelion distance is @xmath242 au for plummer and @xmath4 au for hernquist . *
the typical formation efficiency of the oort cloud is 1.5% , lower than previous estimates , but consistent with a single detection of sedna .
rb gratefully thanks germany s helmholtz alliance and their planetary evolution and life programme for financial support .
mjd acknowledges funding from canada s nserc .
hfl gratefully acknowledges funding from nasa through their origin of solar systems programme .
mes and meb acknowledge receiving funding from nasa origins of solar systems program grant nng05gi02 g .
mes is supported by a nasa earth and space science fellowship and is supported by an nsf astronomy and astrophysics postdoctoral fellowship under award ast-1003258 .
aarseth , s. j. , henon , m. , wielen , r. 1974 . a comparison of numerical methods for the study of star cluster dynamics .
astronomy and astrophysics 37 , 183 - 187 .
+ adams , f. c. , proszkow , e. m. , fatuzzo , m. , myers , p.c . , 2006 . early evolution of stellar groups and clusters : environmental effects on forming planetary systems .
j. 641 , 504 - 525 .
+ adams , f. c. 2010 .
the birth environment of the solar system . annual review of astronomy and astrophysics 48 , 47 - 85
. + allen , l. , megeath , s. t. , gutermuth , r. , myers , p. c. , wolk , s. , adams , f. c. , muzerolle , j. , young , e. , pipher , j. l. 2007 . the structure and evolution of young stellar clusters .
protostars and planets v 361 - 376 .
+ allison , r. j. , goodwin , s. p. , parker , r. j. , de grijs , r. , portegies zwart , s. f. , kouwenhoven , m. b. n. 2009 . dynamical mass segregation on a very short timescale . the astrophysical journal 700 , l99-l103 .
+ andr , p. , belloche , a. , motte , f. , peretto , n. 2007 .
the initial conditions of star formation in the ophiuchus main cloud : kinematics of the protocluster condensations .
astronomy and astrophysics 472 , 519 - 535 .
+ bastian , n. , ercolano , b. , gieles , m. , rosolowsky , e. , scheepmaker , r. a. , gutermuth , r. , efremov , y. 2007 .
hierarchical star formation in m33 : fundamental properties of the star - forming regions .
monthly notices of the royal astronomical society 379 , 1302 - 1312 .
+ bastian , n. , gieles , m. , ercolano , b. , gutermuth , r. 2009 . the spatial evolution of stellar structures in the large magellanic cloud .
monthly notices of the royal astronomical society 392 , 868 - 878 .
+ bate , m. r. , bonnell , i. a. , bromm , v. 2003 . the formation of a star cluster : predicting the properties of stars and brown dwarfs.monthly notices of the royal astronomical society 339 , 577 - 599 .
+ baumgardt , h. , kroupa , p. 2007 . a comprehensive set of simulations studying the influence of gas expulsion on star cluster evolution .
monthly notices of the royal astronomical society 380 , 1589 - 1598 .
+ bernstein , g. m. , trilling , d. e. , allen , r. l. , brown , m. e. , holman , m. , malhotra , r. 2004 .
the size distribution of trans - neptunian bodies . the astronomical journal 128 , 1364 - 1390 .
+ binney , j. , tremaine , s. , 1987 .
galactic dynamics .
princeton univ . press ,
princeton , nj , usa . + blitz , l. , fukui , y. , kawamura , a. , leroy , a. , mizuno , n. , rosolowsky , e. 2007 .
giant molecular clouds in local group galaxies.protostars and planets v 81 - 96 .
+ bonatto , c. , santos , j. f. c. , jr . , bica , e. 2006 .
mass functions and structure of the young open cluster ngc 6611 .
astronomy and astrophysics 445 , 567 - 577 .
+ bonnell , i. a. , davies , m. b. 1998 . mass segregation in young stellar clusters .
monthly notices of the royal astronomical society 295 , 691 - 698 .
+ brasser , r. , duncan , m. j. , levison , h. f. 2006 .
embedded star clusters and the formation of the oort cloud .
icarus 184 , 59 - 82 .
+ brasser , r. , higuchi , a. , kaib , n. 2010 . oort cloud formation at various galactic distances . astronomy and astrophysics 516 , a72-a84 .
+ brown , m.e . ,
trujillo , c. , rabinowitz , d. , 2004 .
discovery of a candidate inner oort cloud planetoid .
j. 617 , 645 - 649 .
+ carpenter , j. m. 2000 .
2mass observations of the perseus , orion a , orion b , and monoceros r2 molecular clouds .
the astronomical journal 120 , 3139 - 3161 .
+ cartwright , a. , whitworth , a. p. 2004 .
the statistical analysis of star clusters .
monthly notices of the royal astronomical society 348 , 589 - 598 .
+ casertano , s. , hut , p. 1985 . core radius and density measurements in n - body experiments connections with theoretical and observational definitions . the astrophysical journal 298 , 80 - 94 .
+ currie , t. , plavchan , p. , kenyon , s. j. 2008 .
a spitzer study of debris disks in the young nearby cluster ngc 2232 : icy planets are common around 1.5 - 3 @xmath5 stars . the astrophysical journal 688 , 597 - 615 .
+ dehnen , w. 1993 .
a family of potential - density pairs for spherical galaxies and bulges . monthly notices of the royal astronomical society 265 , 250 - 256 .
+ dones , l. , weissman , p. r. , levison , h. f. , duncan , m. j. , 2004 .
oort cloud formation and dynamics . in : johnstone , d. , adams ,
lin , d.n.c . , neufeld , d.a . ,
ostriker , e.c .
( eds . ) , aspc , vol .
asp , san francisco , usa .
+ duncan , m. , quinn , t. , tremaine , s. , 1987 .
the formation and extent of the solar system comet cloud .
j. 94 , 1330 - 1338 .
+ duncan , m. j. , levison , h. f. , lee , m. h. 1998 .
a multiple time step symplectic algorithm for integrating close encounters .
the astronomical journal 116 , 2067 - 2077 .
+ eggers , s. , 1999 .
cometary dynamics during the formation of the solar system , ph.d .
thesis , mathematisch - naturwissenschaftlichen fakultten der georg - august - universitt , gttingen , germany .
+ elson , r. a. w. , fall , s. m. , freeman , k. c. 1987 .
the structure of young star clusters in the large magellanic cloud . the astrophysical journal 323 , 54 - 78 .
+ fernndez , j. a. , brunni , a. , 2000 .
the buildup of a tightly bound comet cloud around an early sun immersed in a dense galactic environment : numerical experiments .
icarus 145 , 580 - 590 .
+ di francesco , j. , andr , p. , myers , p. c. 2004 . quiescent dense gas in protostellar clusters : the ophiuchus a core . the astrophysical journal 617 , 425 - 438 .
+ fraser , w. c. , brown , m. e. , schwamb , m. e. 2010 .
the luminosity function of the hot and cold kuiper belt populations .
icarus 210 , 944 - 955 .
+ gladman , b. , holman , m. , grav , t. , kavelaars , j. , nicholson , p. , aksnes , k. , petit , j .-
m . , 2002 .
evidence of an extended scattered disk .
icarus 157 , 269279 .
+ goldreich , p. , lithwick , y. , sari , r. 2004 .
planet formation by coagulation : a focus on uranus and neptune . annual review of astronomy and astrophysics 42 , 549 - 601 .
+ gomes , r. s. , gallardo , t. , fernndez , j. a. , brunni , a. , 2005 . on the origin of the high - perihelion scattered disk : the role of the kozai mechanism and mean - motion resonances .
91 , 109 - 129 .
+ gounelle , m. , meibom , a. 2008 . the origin of short - lived radionuclides and the astrophysical environment of solar system formation.the astrophysical journal 680 , 781 - 792 .
+ graham , r. l. and hell , p. , 1985 . on the history of the minimum spanning tree problem .
comput . 7 , 43 - 57 .
+ guthermuth , r. a. , megeath , s. t. , pipher , j. l. , williams , j. p. , allen , l. e. , myers , p. c. , raines , s. n. , 2005 .
the initial conguration of young stellar clusters : a k - band number counts analysis of the surface density of stars .
j. 632 , 397 - 420 .
+ gutermuth , r. a. , megeath , s. t. , myers , p. c. , allen , l. e. , pipher , j. l. , fazio , g. g. 2009 . a spitzer survey of young stellar clusters within one kiloparsec of the sun : cluster core extraction and basic structural analysis . the astrophysical journal supplement series 184 , 18 - 83 .
+ hernquist , l. 1990 . an analytical model for spherical galaxies and bulges .
the astrophysical journal 356 , 359 - 364 .
+ hillenbrand , l. a. , hartmann , l. w. 1998 . a preliminary study of the orion nebula cluster structure and dynamics . the astrophysical journal 492 , 540 - 553 .
+ innanen , k. a. , harris , w. e. , webbink , r. f. 1983 .
globular cluster orbits and the galactic mass distribution . the astronomical journal 88 , 338 - 360 .
+ jaffe , w. 1983 . a simple model for the distribution of light in spherical galaxies . monthly notices of the royal astronomical society 202 , 995 - 999 .
+ kaib , n. a. , quinn , t. 2008 .
the formation of the oort cloud in open cluster environments .
icarus 197 , 221 - 238 .
+ kenyon , s. j. , bromley , b. c. 2004 .
stellar encounters as the origin of distant solar system objects in highly eccentric orbits.nature 432 , 598 - 602 .
+ king , i. 1962 .
the structure of star clusters .
i. an empirical density law . the astronomical journal 67 , 471 - 485 .
+ king , i. r. 1966 .
the structure of star clusters .
some simple dynamical models . the astronomical journal 71 , 64 - 75 .
+ kokubo , e. , yoshinaga , k. , makino , j. 1998 . on a time - symmetric hermite integrator for planetary n - body simulation .
monthly notices of the royal astronomical society 297 , 1067 - 1072 .
+ kozai , y. , 1962 .
secular perturbations of asteroids with high inclination and eccentricity .
j. 67 , 591598 .
+ kroupa , p. , 2000 .
constraints on stellar - dynamical models of the orion nebula cluster .
new astron . 4 , 615 - 624 .
+ kroupa , p. , tout , c. a. , gilmore , g. , 1993 .
the distribution of low - mass stars in the galactic disk .
mon . not .
r. astron .
262 , 545 - 587 .
+ kroupa , p. , aarseth , s. , hurley , j. , 2001 .
the formation of a bound star clus- ter : from the orion nebula cluster to the pleiades .
mon . not .
r. astron .
321 , 699 - 712 .
+ lada , c. j. , lada , e. a. , 2003 . embedded clusters in molecular clouds .
41 , 57 - 115 .
+ lada , c. j. , margulis , m. , dearborn , d. , 1984 .
the formation and early dynamical evolution of bound stellar systems .
j. 285 , 141 - 152 .
+ larson , r. b. 1985 .
cloud fragmentation and stellar masses .
monthly notices of the royal astronomical society 214 , 379 - 398 .
+ levison , h. f. , duncan , m. j. , 1994 .
the long - term dynamical behavior of shortperiod comets .
icarus 108 , 18 - 36 .
+ levison , h. f. , dones , l. , duncan , m. j. 2001 .
the origin of halley - type comets : probing the inner oort cloud . the astronomical journal 121 , 2253 - 2267 .
+ levison , h. f. , morbidelli , a. , dones , l. , 2004 .
sculpting the kuiper belt by a stellar encounter : constraints from the oort cloud and scattered disk .
j. 128 , 2553 - 2563 .
+ levison , h. f. , morbidelli , a. , vokrouhlick , d. , bottke , w. f. 2008 . on a scattered - disk origin for the 2003 el@xmath243 collisional family - an example of the importance of collisions on the dynamics of small bodies .
the astronomical journal 136 , 1079 - 1088 .
+ levison , h. f. , duncan , m. j. , brasser , r. , kaufmann , d. e. 2010b .
capture of the sun s oort cloud from stars in its birth cluster.science 329 , 187 - 190 .
+ lissauer , j. j. , stevenson , d. j. 2007 .
formation of giant planets .
protostars and planets v 591 - 606 .
+ lykawka , p. s. , mukai , t. 2007 .
resonance sticking in the scattered disk .
icarus 192 , 238 - 247 .
+ lynden - bell , d. 1967 .
statistical mechanics of violent relaxation in stellar systems .
monthly notices of the royal astronomical society 136 , 101 .
+ mckee , c. f. , tan , j. c. 2003 .
the formation of massive stars from turbulent cores .
the astrophysical journal 585 , 850 - 871 .
+ mcmillan , p. j. , binney , j. j. 2010 .
the uncertainty in galactic parameters .
monthly notices of the royal astronomical society 402 , 934 - 940 .
+ moeckel , n. , bonnell , i. a. 2009 .
limits on initial mass segregation in young clusters .
monthly notices of the royal astronomical society 396 , 1864 - 1874 .
+ morbidelli , a. , levison , h. f. , 2004 .
scenarios for the origin of the orbits of the trans - neptunian objects 2000 cr105 and 2003 vb12 ( sedna ) .
j. 128 , 2564 - 2576 .
+ morbidelli , a. , levison , h. f. , bottke , w. f. , dones , l. , nesvorn , d. 2009 .
considerations on the magnitude distributions of the kuiper belt and of the jupiter trojans .
icarus 202 , 310 - 315 .
+ murray , n. 2011 .
star formation efficiencies and lifetimes of giant molecular clouds in the milky way .
the astrophysical journal 729 , 133 - 147 .
+ oort , j.h . , 1950 .
the structure of the cloud of comets surrounding the solar system and a hypothesis concerning its origin .
inst . neth .
11 , 91 - 110 .
+ palla , f. , stahler , s. w. 1999 .
star formation in the orion nebula cluster .
the astrophysical journal 525 , 772 - 783 .
+ peretto , n. , andr , p. , belloche , a. 2006 . probing the formation of intermediate- to high - mass stars in protoclusters . a detailed millimeter study of the ngc 2264 clumps .
astronomy and astrophysics 445 , 979 - 998 .
+ piskunov , a. e. , schilbach , e. , kharchenko , n. v. , rser , s. , scholz , r .- d .
tidal radii and masses of open clusters.astronomy and astrophysics 477 , 165 - 172 .
+ plummer , h. c. 1911 . on the problem of distribution in globular star clusters .
monthly notices of the royal astronomical society 71 , 460 - 470 .
+ portegies zwart , s. f. 2009 .
the lost siblings of the sun . the astrophysical journal 696 , l13-l16 .
+ portegies zwart , s. f. , mcmillan , s. l. w. , gieles , m. 2010 .
young massive star clusters . annual review of astronomy and astrophysics 48 , 431 - 493 .
+ press , w.h . ,
teukolsky , s.a . ,
vetterling , w.t . , flannery , b.p .
numerical recipes in fortran : the art of scientic computing , second ed .
cambridge univ . press , cambridge ,
+ proszkow , e .-
m . , adams , f. c. 2009 . dynamical evolution of young embedded clusters : a parameter space survey . the astrophysical journal supplement series 185 , 486 - 510 .
+ schmeja , s. , klessen , r. s. 2006 . evolving structures of star - forming clusters . astronomy and astrophysics 449 , 151 - 159 .
+ schmeja , s. , kumar , m. s. n. , ferreira , b. 2008 . the structures of embedded clusters in the perseus , serpens and ophiuchus molecular clouds .
monthly notices of the royal astronomical society 389 , 1209 - 1217 .
+ schmeja , s. 2011 . identifying star clusters in a field : a comparison of different algorithms .
astronomische nachrichten 332 , 172 - 184 .
+ schwamb , m. e. , brown , m. e. , rabinowitz , d. l. , ragozzine , d. 2010 .
properties of the distant kuiper belt : results from the palomar distant solar system survey .
the astrophysical journal 720 , 1691 - 1707 .
+ spurzem , r. , giersz , m. , heggie , d. c. , lin , d. n. c. 2009 .
dynamics of planetary systems in star clusters . the astrophysical journal 697 , 458 - 482 .
+ testi , l. , sargent , a. i. , olmi , l. , onello , j. s. 2000 .
star formation in clusters : early subclustering in the serpens core .
the astrophysical journal 540 , l53-l56 .
+ tremaine , s. , richstone , d. o. , byun , y .-
i . , dressler , a. , faber , s. m. , grillmair , c. , kormendy , j. , lauer , t. r. 1994 .
a family of models for spherical stellar systems . the astronomical journal 107 , 634 - 644 .
+ tsiganis , k. , gomes , r. , morbidelli , a. , levison , h. f. 2005 .
origin of the orbital architecture of the giant planets of the solar system .
nature 435 , 459 - 461 .
+ walsh , a. j. , myers , p. c. , di francesco , j. , mohanty , s. , bourke , t. l. , gutermuth , r. , wilner , d. 2007 . a large - scale survey of ngc 1333 . the astrophysical journal 655 , 958 - 972 .
+ whitmore , b. c. , zhang , q. , leitherer , c. , fall , s. m. , schweizer , f. , miller , b. w. 1999 .
the luminosity function of young star clusters in `` the antennae '' galaxies ( ngc 4038 - 4039 ) . the astronomical journal 118 , 1551 - 1576 .
+ wisdom , j. , holman , m. , 1991 .
symplectic maps for the @xmath244-body problem .
j. 102 , 1528 - 1538 . | we re - examine the formation of the inner oort comet cloud while the sun was in its birth cluster with the aid of numerical simulations .
this work is a continuation of an earlier study ( brasser et al . ,
2006 ) with several substantial modifications . first , the system consisting of stars , planets and comets is treated self - consistently in our n - body simulations , rather than approximating the stellar encounters with the outer solar system as hyperbolic fly - bys .
second , we have included the expulsion of the cluster gas , a feature that was absent previously .
third , we have used several models for the initial conditions and density profile of the cluster either a hernquist or plummer potential and chose other parameters based on the latest observations of embedded clusters from the literature .
these other parameters result in the stars being on radial orbits and the cluster collapses .
similar to previous studies , in our simulations the inner oort cloud is formed from comets being scattered by jupiter and saturn and having their pericentres decoupled from the planets by perturbations from the cluster gas and other stars .
we find that all inner oort clouds formed in these clusters have an inner edge ranging from 100 au to a few hundred au , and an outer edge at over 100000 au , with little variation in these values for all clusters .
all inner oort clouds formed are consistent with the existence of ( 90377 ) sedna , an inner oort cloud dwarf planetoid , at the inner edge of the cloud : sedna tends to be at the innermost 2% for plummer models , while it is 5% for hernquist models .
we emphasise that the existence of sedna is a generic outcome .
we define a ` concentration radius ' for the inner oort cloud and find that its value increases with increasing number of stars in the cluster , ranging from 600 au to 1500 au for hernquist clusters and from 1500 au to 4000 au for plummer clusters .
the increasing trend implies that small star clusters form more compact inner oort clouds than large clusters .
we are unable to constrain the number of stars that resided in the cluster since most clusters yield inner oort clouds that could be compatible with the current structure of the outer solar system .
the typical formation efficiency of the inner oort cloud is 1.5% , significantly lower than previous estimates .
we attribute this to the more violent dynamics that the sun experiences as it rushes through the centre of the cluster during the latter s initial phase of violent relaxation .
origin , solar system ; comets , dynamics ; planetary dynamics |
ever since the 1920s , every wireless system has been required to have an exclusive license from the government in order not to interfere with other users of the radio spectrum . today ,
with the emergence of new technologies which enable new wireless services , virtually all usable radio frequencies are already licensed to commercial operators and government entities . according to former u.s .
federal communications commission ( fcc ) chair william kennard , we are facing with a spectrum drought " @xcite . on the other hand ,
not every channel in every band is in use all the time ; even for premium frequencies below 3 ghz in dense , revenue - rich urban areas , most bands are quiet most of the time .
the fcc in the united states and the ofcom in the united kingdom , as well as regulatory bodies in other countries , have found that most of the precious , licensed radio frequency spectrum resources are inefficiently utilized @xcite . in order to increase the efficiency of spectrum utilization ,
diverse types of technologies have been deployed .
cognitive radio is one of those that leads to the greatest technological gain in wireless capacity . through the detection and utilization of the spectra that are assigned to the licensed users but standing idle at certain times , cognitive radio acts as a key enabler for spectrum sharing .
spectrum sensing , aiming at detecting spectrum holes ( i.e. , channels not used by any primary users ) , is the precondition for the implementation of cognitive radio . the cognitive radio ( cr )
nodes must constantly sense the spectrum in order to detect the presence of the primary radio ( pr ) nodes and use the spectrum holes without causing harmful interference to the prs .
hence , sensing the spectrum in a reliable manner is of vital importance and constitutes a major challenge in cr networks .
however , detection is compromised when a user experiences shadowing or fading effects or fails in an unknown way . to get a better understanding of the problem ,
consider the following example : a typical digital tv receiver operating in a 6 mhz band must be able to decode a signal level of at least -83 dbm without significant errors @xcite .
the typical thermal noise in such bands is -106 dbm .
hence a cr which is 30 dbm more sensitive has to detect a signal level of -113 dbm , which is below the noise floor @xcite . in such cases
, one cr user can not distinguish between an unused band and a deep fade . in order to combat such effects
, recent studies suggest collaboration among multiple cr nodes for improving spectrum sensing performance .
collaborative spectrum sensing ( css ) techniques are introduced to improve the performance of spectrum sensing . by allowing different secondary users to collaborate and share their information ,
pr detection probability can be greatly increased .
css can be classified into two categories .
the first category involves multiple users exchanging information @xcite , and the second category uses relay transmission @xcite . some recent studies on collaborative spectrum sensing
include cooperative scheme design guided by game theory @xcite and random matrix theory @xcite , cluster - based cooperative css @xcite , and distributed rule - regulated css @xcite ; studies concentrating on css performance improvement include @xcite introducing spatial diversity techniques to combat the error probability due to fading on the reporting channel between the cr nodes and the central fusion center .
there are also studies concerning other interesting aspects of css performance under different constraints @xcite .
very recently , there are emerging applications of the compressive sensing concept for css @xcite
. existing literature mostly focuses on the css performance examination when the centralized fusion center receives and combines _ all _ cr reports . in an @xmath1 channel cognitive radio network with @xmath2 cr nodes
, the fusion center has to deal with @xmath3 reports and combine them wisely to form a channel sensing result .
however , it is known that wireless channels are subject to fading and shadowing . when secondary users experience multi - path fading or happen to be shadowed , the reports transmitted by cr users are subject to transmission loss . as a result , in practice ,
no entire report data set is available at the fusion center .
besides , due to the fact that each cognitive radio can only sense a small proportion of the spectrum with limited hardware , each cr user gathers only very limited information about the entire spectrum .
* contributions : * we seek to release crs from sending , and the central control unit from gathering , an excessively large number of reports , also target at the situations where there are only a few cr nodes in a large network and thus unable to gather enough sensing information for the traditional css .
we propose to equip each cognitive radio node with a frequency selective filter , which linearly combines multiple channel information .
the linear combinations are sent as reports to the fusion center , where the occupied channels are decoded from the reports by compressive sensing algorithms . as a consequence ,
the amount of channel sensing at crs and the number of reports sent from the crs to the fusion center are both significantly reduced .
following our previous work @xcite on compressive sensing , we propose two approaches to collaborative spectral sensing .
the first approach is based on solving a matrix completion problem @xcite , which seeks to efficiently reconstruct a matrix ( typically low - rank ) from a relatively small number of revealed entries . in this approach ,
the entries of the underlying matrix are linear combinations of channel powers .
each cr node takes its local spectrum measurements , but instead of directly recording channel powers , it uses its frequency - selective filters to take @xmath4 _ linear combinations _ of channel powers and reports them to the fusion center .
the total @xmath5 linear combinations taken by @xmath2 crs form a @xmath5 matrix at the fusion center . considering transmission loss , we allow the the matrix to be incomplete . we show that this matrix is low - rank and has the properties enabling its reconstruction from only a small number of its entries , and therefore , information about the complete spectrum usage can be recovered from a small number of reports from the cr nodes .
this approach significantly reduces the amount of sensing and communication workload .
the second approach is based on joint sparsity recovery @xcite , which is motivated by the observation that the spectrum usage information the cr nodes collect has a common sparsity pattern : each of the few occupied channels is typically observed by multiple crs .
we develop a novel algorithm for joint sparsity signal recovery , which is more effective than existing algorithms in the compressive sensing literature since it can accommodate a large dynamic range of channel gains . in both approaches
, every cr senses all channels ( by taking random linear projections of the powers of all channels ) , and the crs do not communicate . while they work independently , their measurements are analyzed jointly by the detection algorithms running at the fusion center .
therefore , our approaches are very different from the existing collaborative spectrum sensing schemes in which different crs are assigned to different channels .
our approaches move from collaborative sensing to collaborative " computation and shift coordination from the sensing phase to the post - sensing phase .
our work is among the first that applies matrix completion or joint sparsity recovery to collaborative spectrum sensing in cognitive radio networks .
matrix completion and joint sparsity recovery are both being intensively studied in the compressive sensing community .
we present them both because it is too early at this time to make a verdict of an eventual winner .
the rest of this paper is organized as follows : in section [ sec : model ] , the system model is given .
the matrix completion - based algorithm for collaborative sensing is described in section [ sec : algorithm1 ] , and the joint sparsity based algorithm is described in section [ sec : algorithm2 ] .
after that , in section [ sec : discussion ] we compare the two proposed approaches , discuss their computational complexity as well as filter design and dynamic update .
simulation results are presented in section [ sec : simulation ] , and conclusions are drawn in section [ sec : conclusion ] .
we consider a cognitive radio network with @xmath2 cr nodes that locally monitor a subset of @xmath1 channels .
a channel is either occupied by a pr or unoccupied , corresponding to the states @xmath6 and @xmath7 , respectively .
we assume that the number @xmath8 of occupied channels is much smaller than @xmath1 .
the goal is to recover the occupied channels from the cr nodes observations .
since each cr node can only sense limited spectrum at a time , it is impossible for limited @xmath2 crs to observe @xmath1 channels simultaneously . to overcome this problem
, we propose the scheme depicted in fig .
[ f : system_model ] .
instead of scanning all channels and sending each channel s status to the fusion center , using its frequency - selective filters , a cr takes a small number of measurements that are linear combinations of multiple channels .
the filter coefficients can be designed and implemented easily . in order to mix the different channel sensing information ,
the filter coefficients are designed to be random numbers .
then , these filter outputs are sent to the fusion center .
suppose that there are @xmath4 frequency selective filters in each cr node sending out @xmath4 reports regarding the @xmath1 channels .
for the non - ideal cases , where we have relatively less measurements @xmath9 , i.e. , the number of reports sent from all crs is less than the total number of channels .
the sensing process at each cr can be represented by a @xmath10 filter coefficient matrix @xmath11 .
let an @xmath12 _ diagonal _ matrix @xmath13 represent the states of all the channel sources using @xmath7 and @xmath6 as diagonal entries , indicating the unoccupied or occupied states , respectively .
there are @xmath8 nonzero entries in @xmath14 .
in addition , channel gains between the crs and channels are described in an @xmath15 channel gain matrix @xmath16 given by @xcite : @xmath17 where @xmath18 is the @xmath19 primary user s transmitted power , @xmath20 is the distance between the primary transmitter using @xmath21 channel and the @xmath19 cr node , @xmath22 is the propagation loss factor , and @xmath23 is the channel fading gain . for awgn channel , @xmath24 ; for rayleigh channel , @xmath25 follows independent rayleigh distribution ; and for shadowing fading , @xmath25 follows log - normal distribution @xcite . without loss of generality , we assume that all prs use unit transmit power ( otherwise , we can compensate by altering the corresponding channel gains ) . the measurement reports sent to the fusion center can be written as a @xmath5 matrix @xmath26 note that due to loss or errors , some of the entries of @xmath27 are possibly missing .
the binary numbers on the diagonal of @xmath13 are the @xmath1channel states that we shall estimate from the available entries of @xmath27 .
it is typically difficult for the fusion center to acquire all entries of @xmath27 due to transmission failure , which means that our observation is a subset @xmath28\times[m]$ ] of @xmath27 .
however , it is possible to recover the missing entries in @xmath27 since it holds the following two important properties @xcite required for matrix completion : 1 . *
low rank * : @xmath29 equals to @xmath8 , which is the number of prime users in the network and is usually very small .
incoherent property * : generate @xmath11 randomly ( subject to hardware limitation ) . from and the fact that @xmath13 has only @xmath8 nonzeros on the diagonal , @xmath27 s svd factors @xmath30 , @xmath31 , and @xmath32 satisfy the _ incoherence condition _ @xcite . * there exists a constant @xmath33 such that for all @xmath34 $ ] , @xmath35 $ ] , we have @xmath36 , @xmath37 @xmath38 .
* there exists @xmath39 such that @xmath40 .
@xmath27 is in general incomplete because of transmission failure .
moreover , each cr might only be able to collect a random ( up to @xmath4 ) number of reports due to the hardware limitation .
therefore , the fusion certain receives a subset @xmath28\times[m]$ ] of @xmath27 s entries .
we assume that the received entries are uniformly distributed with high probability .
hence , we work with a model in which each entry shows up in @xmath41 identically and independently with probability @xmath42 .
given @xmath43 , the partial observation of @xmath27 is defined as a @xmath5 matrix given by @xmath44 we shall first recover the unobserved elements of @xmath27 from @xmath45 .
then , we reconstruct @xmath46 from the given @xmath11 and @xmath27 using the fact that all but @xmath8 rows of @xmath46 are zero .
these nonzero rows correspond to the occupied channels .
since @xmath4 and @xmath2 are much smaller than @xmath1 , our approach requires a much less amount of sensing and transmission , compared to traditional spectrum sensing in which each channel is monitored separatively . in previous research on matrix completion @xcite , it was proved that under some suitable conditions , a low - rank matrix can be recovered from a random , yet small subset of its entries by nuclear norm minimization : @xmath47 where @xmath48 denotes the nuclear norm of matrix @xmath27 and @xmath49 is a parameter discussed in section [ sec : stop ] below . for notational simplicity ,
we introduce the linear operator @xmath50 that selects the components @xmath41 out of a @xmath10 matrix and form them into a vector such that @xmath51 .
the adjoint of @xmath50 is denoted by @xmath52 .
recent algorithms developed for ( [ mtx_cmp ] ) include , but not limited to , the singular value thresholding ( svt ) algorithm @xcite and the fixed - point continuation iterative algorithm ( fpca ) @xcite for fast completion of large - scale matrices ( e.g. , more than @xmath53@xmath54@xmath53 ) , a special trimming step introduced by keshavan et al . in @xcite . for our problem , we adopt fpca , which appears to run very well for our small dimensional tests . in the following subsections ,
we describe this algorithm and the steps we take for nuclear norm minimization .
also , we study how to use the approximate singular value decomposition ( svd)-based iterative algorithm introduced in @xcite for fast execution .
we further discuss the stopping criteria for iterations to acquire optimal recovery .
finally we show how to obtain @xmath13 from the estimation @xmath55 of @xmath27 .
fpca is based on the following fixed
point iteration : @xmath56 where @xmath57 is step size and @xmath58 is the matrix shrinkage operator defined as follows : [ def : shrinkage ] * matrix shrinkage operator @xmath59 * : assume @xmath60 and its svd is given by @xmath61 , where @xmath62 , @xmath63 , and @xmath64 . given @xmath65 ,
@xmath59 is defined as @xmath66 with the vector @xmath67 defined as : @xmath68 simply speaking , @xmath69 reduces every singular values ( which is nonnegative ) of @xmath27 by @xmath49 ; if one is smaller than @xmath22 , it is reduced to zero . in addition
, @xmath70 is the solution of @xmath71 where @xmath72 is the frobenius norm .
to understand , observe that the first step of is a gradient - descent applied to the second term in and thus reduces its value . because the previous gradient - descent generally increases the nuclear norm ,
the second step of involves solving to reduce the nuclear norm of @xmath73 .
iterations based on converge when the step sizes @xmath57 are properly chosen ( e.g. , less than 2 , or select by line search ) so that the first step of is not `` expansive '' ( the other step is always non - expansive ) .
as stated in @xcite , the second step of requires computing the svd decomposition of @xmath73 , which is the main computational cost of .
however , if one can predetermine the rank of the matrix @xmath27 , or have the knowledge of the approximate range of its rank , a full svd can be simplified to computing only a rank-@xmath74 approximation to @xmath73 .
combined with the above fixed point iteration , the resulting algorithm is called fixed - point continuation algorithm with approximate svd ( fpca ) .
specifically , the approximate svd is computed by a fast monte carlo algorithm developed by drineas et al.@xcite . for a given matrix @xmath75 and parameters @xmath76
, this algorithm returns an approximations to the largest @xmath76 singular values corresponding left singular vectors of the matrix @xmath77 in a linear time .
we tune the parameters in fpca for a better overall performance .
continuation is adopted by fpca , which solves a sequence of instances of , easy to difficult , corresponding to a sequence of large to small values of @xmath49 .
the final @xmath49 is the given one but solving the easier instances of gives intermediate solutions that warm start the more difficult ones so that the entire solution time is reduced .
solving each instance of requires proper stopping . because our ultimate goal is to recover 0/1 values on the diagonal of @xmath13 , accurate solutions of are not required .
therefore , we use the criterion : @xmath78 where @xmath79 is a small positive scalar .
experiments shows that @xmath80 is good enough for obtaining optimal @xmath13 .
since @xmath11 has more columns than rows , directly solving @xmath81 in from given @xmath27 is under - determined .
however , each row @xmath82 of @xmath83 corresponds to the occupancy status of channel @xmath84 . ignoring noise in @xmath27 for now
, @xmath82 contains a positive entry if and only if channel @xmath84 is used .
hence , most rows of @xmath83 are completely zero , so every column @xmath85 of @xmath83 is sparse and all @xmath85 s are jointly sparse
. such sparsity allows us to reconstruct @xmath83 from and identify the occupied channels , which are the nonzero rows of @xmath83 .
since the channel fading decays fast , the entries of @xmath83 have a large dynamic range , which none of the existing algorithms can deal with well enough . hence , we develop a novel joint - sparsity algorithm briefly described as follows .
the algorithm is much faster than matrix completion and typically needs 1 - 5 iterations . at each iteration
, every column @xmath85 of @xmath83 is independently reconstructed using the model @xmath86 , where @xmath87 is the @xmath88th column of @xmath27 .
for noisy @xmath27 , we instead use the constraint @xmath89 .
the same set of weights @xmath90 is shared by all @xmath88 at each iteration .
@xmath90 is set to 1 uniformly at iteration 1 .
after channel @xmath84 is detected in an iteration , @xmath90 is set to 0 . through @xmath90 ,
joint sparsity information is passed to all @xmath88 .
channel detection is performed on the reconstructed @xmath85 s at each iteration .
it is possible that some reconstructed @xmath85 is wrong , so we let larger and sparser @xmath85 s have more say . if there is a relatively large @xmath91 in a sparse @xmath85 , then @xmath84 is detected .
we have found this algorithm to be very reliable .
the detection accuracy is determined by the accuracy of @xmath27 provided .
in this section , we describe a new , highly effective algorithm for recovering @xmath92 and thus @xmath13 by thresholding @xmath83 . the algorithm allows but does not require the same @xmath11 for all crs , i.e.
, each cr can use a different sensing matrix @xmath11 .
the design of @xmath11 is discussed in section [ sec : discussion_design ] below . in @xmath83
, each column ( denoted by @xmath85 ) corresponds to the channel occupancy status received by cr @xmath88 , and each row @xmath93 corresponds to the occupancy status of channel @xmath84 . _
ignoring noise _ for now , a row has a positive value ( i.e. , @xmath94 ) if and only if channel @xmath84 is used .
since there are only a small number of used channels , @xmath83 is sparse in terms of the number of rows containing nonzero . in each nonzero row @xmath93
, there is typically more than one nonzero entry ; in other words , if @xmath95 , other entries in the same row are likely nonzero . therefore , @xmath83 is _
jointly sparse_. in the case that the true @xmath83 contains noise , it is approximately , rather than exactly , jointly sparse .
joint sparsity is utilized in our algorithm to recover @xmath83 .
while there are existing algorithms for recovering jointly sparse signals in the literature ( e.g. , in @xcite ) , our algorithm is very different and more effective for our underlying problem .
none of the existing algorithms works well to recover @xmath83 because the entries of @xmath83 have a very large dynamic range because , in any channel fading model , channel gains decay rapidly with distance between crs and prs .
most existing algorithms are based on minimizing @xmath96 for @xmath97 and @xmath98 .
if @xmath99 , it is the same as minimizing the 1-norm of each column independently , so joint sparsity is not used for recovery . if @xmath100 or @xmath98 , joint sparsity is considered , but it penalizes a large dynamic range since the large values in a nonzero row of @xmath83 contribute superlinearly , more than the small values in that row , to the minimizing objective . in short , @xmath4 close 1 loses joint sparsity and @xmath4 bigger than 1 penalizes large dynamic ranges .
our new algorithm not only utilizes joint sparsity but also takes advantages of the large dynamic range of @xmath83 .
the large dynamic range has its pros and cons in cs recovery .
it makes it easy to recover the locations of large entries , which can be achieved even without recovering the locations of smaller ones . on the other hand
, it makes difficult to recover both the locations and values of the smaller entries .
this difficulty has been studied in our previous work @xcite , where we proposed a fast and accurate algorithm for recovering 1d signals @xmath101 by solving several ( about 5 - 10 ) subproblems in the form of @xmath102 where the index set @xmath103 is formed iteratively as @xmath104 excluding the identified locations of large entries of @xmath101 . with techniques such as early detections and warm starts ,
it achieves both the state of the art speed and least requirement on the number of measurements .
we integrate the idea of this algorithm with joint sparsity into the new algorithm below .
@xmath105 @xmath106 for every cr @xmath88 with enough measurements ( in presence of measurement noise , @xmath107 is replaced by @xmath108 ) report @xmath83 , and @xmath13 by thresholding @xmath83 the framework of the proposed algorithm is shown in table [ t : algorithm ] . at each iteration , every channel is first subject to independent recovery . unlike minimizing @xmath96 , which ties all crs together
, independent recovery allows large entries of @xmath83 to be quickly recovered .
joint sparsity information is passed among the crs through a shared index set @xmath103 , which is updated iteratively to exclude the used channels that are already discovered .
below , we describe each step of the above algorithm in more details . in the * independence recovery * step , for every qualified cr , a constrained problem in the form of with constraints @xmath109 in the noiseless case , or @xmath108 in the noisy case ,
is considered , where @xmath110 is an estimated noise level . as
problem dimensions are small in our application , solvers are easily chosen : matlab s ` linprog ' for noiseless cases and mosek @xcite for noisy cases .
both of these solvers run in polynomial times .
this step dominates the total running time of algorithm [ t : algorithm ] , but up to @xmath2 optimization problems can be solved in parallel .
parallelization is simple for the joint - sparsity approach . at each outer iteration ,
all lps are solved independently , and they have small scales relative to today s lp solvers , like gurobi @xcite and its matlab interface gurobi mex @xcite , where gurobi automatically detects and uses all cpu and cores for solving lps .
crs without enough measurements ( e.g. , most of their reports are missing due to transmission losses or errors ) are not qualified for independent recovery because cs recovery is known unstable in such a case .
specifically , we require the number of the available measurements from each qualified cr to exceed twice as many as used channels or @xmath111 .
when measurements are ample , the first iteration will yield exact or nearly exact @xmath85 s . otherwise , insufficient measurements can cause a completely wrong @xmath85 that misleads channel detection ; neither the locations nor the values of the nonzero entries are correct . the algorithm , therefore , filters trusted @xmath85 s that must be either sparse or compressible .
large entries in such @xmath85 s likely indicate correct locations .
a theoretical explanation of this argument based on stability analysis for is given in @xcite .
used channels are detected among the set of trusted @xmath85 s . to further reduce the risk of false detections , we compute a percentage for every channel in a way that those channels corresponding to larger values in @xmath83 and whose values
are located in relatively sparser @xmath85 s are given higher percentages . here
, relative sparsity is defined proportionally to the number of measurements ; for fixed number of non - zeros or degree of compressibility , the more the measurements , the higher the relative sparsity .
hence , @xmath85 corresponding to more reported cr @xmath88 also tends to have a higher percentage . in short , larger and sparse solutions
have more say .
the channels receiving higher percentages are detected as used channels .
the index set @xmath103 is set as @xmath112 excluding the used channels that are already detected .
obviously , @xmath103 needs to change from one iteration to the next ; otherwise , two iterations will result in an identical @xmath83 and thus the stagnation of algorithm .
therefore , if the last iteration posts no change in the set of used channels yet the stopping criterion ( see next paragraph ) is not met , the channels @xmath84 corresponding to the larger @xmath113 are also excluded from @xmath103 , and such exclusion becomes more aggressive as iteration number increases .
this is not an ad hoc but a rigorous treatment .
it is shown in @xcite that larger entries in an inexact cs recovery tend to be the true nonzero entries , and furthermore , as long as the new @xmath103 excludes more true than false nonzero entries by a certain fraction , will yield a better solution in terms of a certain norm . in short ,
used channels leave @xmath103 , and in case of no leaves , channels with larger joint values @xmath113 leave @xmath103 .
finally , the iteration is terminated when the tail of @xmath83 is small enough .
one way to define the tail size of @xmath83 is the fraction @xmath114 , i.e. , the thought
unused divided by the thought used .
suppose that @xmath103 precisely contains the unused channels and measurements are noiseless , then every recovered @xmath85 in channel detection is exact , so the fraction is zero ; with noise , the fraction depends on noise magnitude and is small as long as noise is small .
if @xmath103 includes any used channel , the numerator will be large whether or not @xmath85 s are ( nearly ) exact . in a sense , the tail size measures how well @xmath83 and @xmath103 match the measurements @xmath115 and expected sparseness .
unless the true number of used channels is known , the tail size appears to be an effective stopping indicator .
in the worst case , algorithm [ t : algorithm ] reduces the cardinality of @xmath103 by 1 per iteration , corresponding to recovering at least 1 additional used channel . therefore , the number of iterations can not exceed the number of total channels .
however , the first couple of iterations typically recover most of the used channels . at each iteration
, the independence recovery step solves up to @xmath2 optimization problems , which can be independently solved in parallel , so the complexity equals a linear program ( or second - order cone program ) whose size is no more than @xmath1 .
the worst case complexity is @xmath116 but it is almost never observed in sparse optimization thanks to solution sparsity .
the two other steps are based on basic arithmetic and logical operations , and they run in @xmath117 . in practice ,
algorithm [ t : algorithm ] is implemented and run on a workstation at the fusion center .
computational complexity will not be a bottleneck of the system . as to the matrix completion algorithm
, according to @xcite , fpca can recover @xmath118 matrices of rank 50 with a relative error of @xmath119 in about 3 minutes by sampling only 20 percent of the elements .
the matrix completion ( section [ sec : algorithm1 ] ) and joint sparsity recovery ( section [ sec : algorithm2 ] ) approaches both take linear channel measurements as input and both return the estimates of used channels . on the other hand
, the joint sparsity approach takes the full advantage of @xmath11 , so it is expected to work with smaller numbers of measurements . in addition , even though only one matrix completion problem needs to be solved in the matrix completion approach , it still takes much longer than running the entire joint sparsity recovery , and it is not easy to parallelize any of the existing matrix completion algorithms .
however , in the small - scale networks , in cases where too much sensing information is lost during transmission or there are too many active prs in the network , which increase the signal sparsity level , joint sparsity recovery algorithm with our current settings will experience degradation in performance . we , however , can not verdict an eventual winner between the two approaches as they are both being studied and improved in the literature .
for example , if a much faster matrix completion algorithm is developed which takes advantage of @xmath11 , the disadvantages of the approach may no longer exist .
the proposed method senses the channels , not by measuring the responses of individual channels one by one , but rather measures a few incoherent linear combinations of all channels responses through onboard frequency - selective filter set .
the filter coefficients which perform as the sensing matrix should have entries independently sampled from a sub - gaussian distribution , since this is known to be best for compressive sensing in terms of the number of measurements ( given in order of magnitude ) required for exact recovery . in other words ,
up to a constant , which is independent of problem dimensions , no other type of matrix is yet known to perform consistently better . however , other types of matrices ( such as partial fourier / dct matrices @xcite and other random circulant matrices @xcite ) have been either theoretically and/or numerically demonstrated to work as effectively in many cases .
these latter sensing matrices are often easier to realize physically or electrically .
for example , applying a random circulant matrix performs sub - sampled convolution with a random vector . frequency - selective surfaces ( fsss ) can be used to realize frequency filtering .
this can be done by designing a planar periodic structure with a unit element size around half wavelength of the frequency of interests .
both the metallic and dielectric materials can be used . to deal with the bandwidth , unit elements in different shapes will be tested .
channel occupancy evolves over time as prs start and stop using their channels .
channel gains can also change when the prs move .
however , the cs research has so far focused on static signal sensing except the very recent path following algorithms in @xcite . in the future work
, we can investigate recovery methods for a dynamic wireless environment where based on existing channel occupancy information , an insignificant change of channel states can be quickly and reliably discovered . given existing channel occupancy @xmath83 , each new report , which is an entry @xmath120 of @xmath27 ,
is compared with @xmath121 .
if a significant number of such comparisons show differences , then there is a change in the true @xmath83 . since @xmath122 , either @xmath13 or @xmath16 , or both , have changed .
a change in @xmath13 means new channel occupation or release . if @xmath13 is unchanged , then those channel gains in @xmath16 corresponding to occupied channels have changed .
it is easy to deal with the latter case ( i.e. , @xmath16 changed , but @xmath13 did nt ) and update the gains of occupied channels because it boils down to solving a small linear system .
let @xmath123 and @xmath124 denote the sub - matrices of @xmath11 and @xmath83 , respectively , formed by their columns and rows corresponding to the occupied channels .
then , the new gains are given in the least - squares solution of @xmath125 , where @xmath27 shall include new reports arrived after the previous recovery / update but may still have missing entries . this system is easy to solve since the number of occupied channels is small . in a similar way it is easy
to discover released channels as long as there is no introduction of new occupied channels .
the release of channel @xmath84 means row @xmath82 of @xmath83 turns into 0 , or small numbers .
therefore , one can solve the system @xmath125 and find the released channels , which correspond to the rows of @xmath124 with all zero ( or small ) entries .
when the system @xmath125 is inconsistent , it means that the received reports can not be explained by the previously occupied channels , so there must be new channel occupation . discovering new channel occupation is more difficult since it is to find changes in the previously unoccupied ones , which are much more than the occupied channels .
however , it is computationally much easier than starting from scratch .
let @xmath126 and @xmath83 denote the previous and current channel information , respectively .
arguably , @xmath127 is highly sparse in the joint sense because only its rows corresponding to newly occupied or released channels can have large nonzero entries .
hence , @xmath83 can be quickly recovered by performing joint sparsity recovery on @xmath127 over the constraints @xmath128 ( or a relax version in the noisy case ) , a task that can be done by the algorithms for stationary recovery .
the probability of detection ( pod ) and false alarm rate ( far ) are the two most important indices associated with spectrum sensing .
we also consider the miss detection rate ( mdr ) of the proposed system .
the higher the pod , the less interference will the crs bring to the prs , while from the crs perspective , lower far will increase their chance of transmission .
there is a tradeoff between pod and far .
while designing the algorithms , we try to balance the cr nodes capability of transmission and their interferences to the pr nodes .
performance is evaluated in terms of pod , far and mdr defined as follows : @xmath129 where _ no . false _ is the number of false alarms , _ no . miss _ is the number of miss detections , _ no .
hit _ is the number of successful detections of primary users , and _ no .
correct _ is the number of correct reports of no appearance of pr . we define sampling rate as @xmath130 where @xmath131 is the amount of total sensing workload in traditional spectrum sensing . + according to fcc and defense advance research projects agency ( darpa )
reports @xcite data , we chose to test the proposed matrix completion recovery algorithm for spectrum utilization efficiency over a range from 3% to 12% , which is large enough in practice .
specifically , the number of active primary users is 1 to 4 on a given set of 35 channels with 20 cr nodes .
[ f : far_mdr_sr ] shows the false alarm and miss detection rates at different sampling rates for different numbers of pr nodes . among all cases ,
the highest miss detection rate is no more than 5% , and this is from only 20% samples which are supposed to be gathered from the cr nodes regarding all the channels .
when the sampling rate is increased to 50% and even when the channel occupancy is relatively high , i.e. , 12% of the channels are occupied by the prs , the miss detection rates can be as low as no more than 2% . from our simulation results , with a moderate channel occupancy at 9% ,
the false alarm rates are around 3% to 5% .
[ f : pod_sr ] shows the probability of detection at different sampling rates .
when the spectrum is lightly occupied by the licensed user at 3% channels being occupied , only 20% samples offer a pod close to 100% , and when there is a slightly raise in sampling rate , pod can reach 100% . in the worst case of 12% spectrum occupancy , 20% sampling rate still can offer a pod of higher than 95% , and as the sampling rate reaches 50% , pod can reach 98% .
joint sparsity recovery is designed for large scale application , and simulations carried out for a larger dimensional applications with the following settings : we consider a @xmath132-node cognitive radio network within a @xmath133 meter square area centered at the fusion center .
the @xmath132 cr nodes are uniformly randomly located .
these cognitive radio nodes collaboratively sense the existence of primary users within a @xmath118 meter square area on @xmath134 channels , which are centered also at the fusion center .
we chose to test the proposed algorithm for the number of active pr nodes ranging from @xmath6 to @xmath135 on the given set of 500 channels .
since the fading environments of the cognitive radio networks vary , we evaluate the algorithm performance under three typical channel fading models : awgn channel , rayleigh fading channel , and lognormal shadowing channel .
+ + we first evaluate the pod , far , and mdr performance of the proposed joint sparsity recovery performance in the noiseless environment .
[ f : nn_t0_5 ] , fig .
[ f : nn_t1_5 ] , and fig .
[ f : nn_t2_5 ] show the pod , far and mdr performance at different sampling rate , for awgn channel , rayleigh fading channel , and lognormal shadowing channel , respectively , when small number of cr nodes sense the spectrum collaboratively . fig .
[ f : nn_t0_10 ] , fig .
[ f : nn_t1_10 ] , and fig .
[ f : nn_t2_10 ] show the pod , far and mdr performance at different sampling rate , for the aforementioned three types of channel models , when there are more cr nodes involved in the collaborative sensing of the spectrum .
we observe that , log - normal shadowing channel model shows the best pod , far , and mdr performance no matter how many cr nodes are involved in the spectrum sensing . while the agwn channel model shows the worst pod , far , and mdr performance .
with respect to pod , the performance gap between these two models is at most 10% , which happens when the sampling rate is extremely low . for the rayleigh fading channel model , when the number of samples is @xmath136 of the total number of channels , for all tested cases we achieve @xmath137 pod .
if there are less active pr nodes in the network , smaller number of samples are required for exact detection .
in essence , the proposed ccs system is robust to severe or poorly modeled fading environments .
cooperation among the cr nodes and robust recovery algorithm allow us to achieve this robustness without imposing stringent requirements on individual radios .
+ we then evaluate the pod , far , and mdr performance of the proposed joint sparsity recovery performance in noisy environments .
for all the simulations considering noise , we adopt the rayleigh fading channel model .
[ f : n_performance ] and fig .
[ f : n_per_snr ] show the corresponding results .
we observe that noise does degrade the performance .
however , as shown in fig .
[ f : n_performance ] , when the number of active prs is small enough ( e.g. , no . of pr = 1 ) , even with signal to noise ratio as low as 15 db , we still can achieve @xmath137 pod with a sampling rate of merely @xmath0 . then with an increase in the signal to noise ratio
, lower sampling rate enables more pr nodes to be detected exactly .
[ f : n_per_snr ] shows the pod , far and mdr performance vs. sampling rate at different noise level , each curve for a specific noise level is relatively flat ( i.e. , performance varies a little as sampling rate changes ) .
this shows that the noise level has greater impact on the spectrum sensing performance rather than the sampling rate . at low noise level ,
e.g. , snr = 45 db , @xmath138 sampling rate enables @xmath137 pod for 4 pr nodes .
as snr reduces to 15 db , no more than @xmath139 pod will be achieved even when the number of samples equals to the number of channels in the network . + for comparison , we applied joint sparsity recovery algorithm on a small - scale network with the same settings as we have used to test the matrix completion recovery . instead of using a 500-channel network
, we use a network with only 35 channels .
simulation results show that joint sparsity recovery algorithm performs better than the matrix completion algorithm in the following aspects : 1 .
faster computation due to lower computational complexity ; 2 . higher pod for the spectrum utilization rate between 3% and 12% in the noise free simulations ; to conclude , matrix completion algorithm is good for small - scale networks , with relatively high spectrum utilization , while joint sparsity recovery algorithm has the advantage of low computational complexity which enables fast computation in large - scale networks .
+ + +
in order to reduce the amount of sensing and transmission overhead of cognitive radio ( cr ) nodes , we have applied compressive sensing for collaborative spectrum detection in cognitive radio networks . we propose to equip each cr node with a frequency - selective filter , which linearly combines multiple channel information , and let it send a small number of such linear combinations to the fusion center , where the channel occupancy information is then decoded .
consequently , the amount of channel sensing at the crs and the number of reports sent from the crs to the fusion center reduce significantly .
two novel decoding approaches have been proposed
one based on matrix completion and the other based on joint sparsity recovery .
the novel matrix completion approach recovers the complete cr to center reports from a small number of valid reports and then reconstructs the channel occupancy information .
the joint sparsity approach , on the other hand , skips recovering the reports and directly reconstructs channel occupancy information by exploiting the fact that each occupied channel is observable by multiple cr nodes .
our algorithm enables faster recovery for large - scale cognitive radio networks .
the primary user detection performance of the proposed approaches has been evaluated by simulations .
the results of random tests show that , in noiseless cases , the number of samples required are no more than 50% of the number of channels in the network to guarantee exact primary user detection for both approaches ; while in noisy environments , at low channel occupancy rate , we can still have high probability of detection .
the work of w. yin was supported in part by nsf career award dms-07 - 48839 , onr grant n00014 - 08 - 1 - 1101 , and an alfred p. sloan research fellowship .
the work of h. li was supported in part by nsf grants 0831451 and 0901425 .
the work of z. han was supported in part by nsf cns-0910461 , cns-0901425 , nsf career award cns-0953377 , and air force office of scientific research .
the work of e. hossain was supported by the nserc , canada , strategic project grant stpgp 380888 . c. h. lee and w. wolf , energy efficient techniques for cooperative spectrum sensing in cognitive radios , " in _ proc .
ieee consumer communications and networking conference ( ccnc08 ) _ , las vegas , nevada , january 2008 .
a. ghasemi and e. sousa , collaborative spectrum sensing for opportunistic access in fading environments , " in _ proc .
ieee international symposium on new frontiers in dynamic spectrum access networks ( dyspan05 ) _ , baltimore , md , november 2005 .
w. saad , z. han , m. debbah , a. hjrungnes , and t. basar , coalitional games for distributed collaborative spectrum sensing in cognitive radio networks , " in _ proc .
ieee conference on computer communications ( infocom09 ) _ , rio de janeiro , brazil , april 2009 . l. s. cardoso , m. debbah , p. bianchi , and j. najim , cooperative spectrum sensing using random matrix theory , " in _ proc . international symposium on wireless pervasive computing ( iswpc08 ) _ , santorini , greece , may 2008 . c. sun , w. zhang , and k. b. letaief ,
cluster - based cooperative spectrum sensing in cognitive radio systems , " in _ proc .
ieee international conference on communications ( icc07 ) _ , glasgow , scottland , june 2007 .
l. cao and h. zheng , distributed rule - regulated spectrum sharing , " _ ieee journal on selected areas in communications : special issue on cognitive radio : theoryand applications _ , vol .
1 , pp . 130145 , january 2008 . c. sun , w. zhang , and k. b. letaief , cooperative spectrum sensing for cognitive radios under bandwidth constraint , " in _ proc .
ieee wireless communications and networking conference ( wcnc07 ) _ , hong kong , march 2007 .
j. meng , j. ahmadi - shokouh , h. li , z. han , s. noghanian , and e. hossain , sampling rate reduction for 60 ghz uwb communication using compressive sensing , " in _ proc .
asilomar conference on signals , systems @xmath140 computers _ ,
monterey , ca , november 2009 .
j. meng , h. li , and z. han , sparse event detection in wireless sensor networks using compressive sensing , " in _ proc .
43rd annual conference on information sciences and systems ( ciss ) _ , baltimore , md , march 2009 .
e. j. cands and b. recht , exact low - rank matrix completion via convex optimization , " in _ proc .
communication , control , and computing , 46th annual allerton conference _ , monticello , il , september 2008 .
m. f. duarte , s. sarvotham , m. b. wakin , d. baron , and r. g. baraniuk , joint sparsity models for distributed compressed sensing , " in _ proc .
workshop on signal processing with adaptive sparse structured representations _ ,
rennes , france , november 2005 .
j. tropp , a. c. gilbert , and m. j. strauss , simulataneous sparse approximation via greedy pursuit , " in _ proc .
ieee 2005 international conference on acoustics , speech , and signal processing ( icassp ) _ , philadelphia , pa , march 2005 .
j. meng , w. yin , h. li , e. hossain , and z. han , collaborative spectrum sensing from sparse observations using matrix completion for cognitive radio networks , " in _ proc .
ieee 2010 international conference on acoustics , speech , and signal processing ( icassp ) _ , march 2010 .
e. j. cands , j. romberg , and t. tao , robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , " _ ieee trans . on information theory _ , 52(2 ) , pp . 489509 , february 2006 .
w. yin , s. p. morgan , j. yang , and y. zhang , fast sensing and signal reconstruction : practical compressive sensing with toeplitz and circulant matrices , " _ rice university caam technical report tr10 - 01_. _ submitted to vcip 2010 _ , 2010 . | spectrum sensing , which aims at detecting spectrum holes , is the precondition for the implementation of cognitive radio ( cr ) .
collaborative spectrum sensing among the cognitive radio nodes is expected to improve the ability of checking complete spectrum usage . due to hardware limitations ,
each cognitive radio node can only sense a relatively narrow band of radio spectrum .
consequently , the available channel sensing information is far from being sufficient for precisely recognizing the wide range of unoccupied channels . aiming at breaking this bottleneck , we propose to apply matrix completion and joint sparsity recovery to reduce sensing and transmitting requirements and improve sensing results .
specifically , equipped with a frequency selective filter , each cognitive radio node senses linear combinations of multiple channel information and reports them to the fusion center , where occupied channels are then decoded from the reports by using novel matrix completion and joint sparsity recovery algorithms . as a result , the number of reports sent from the crs to the fusion center is significantly reduced .
we propose two decoding approaches , one based on matrix completion and the other based on joint sparsity recovery , both of which allow exact recovery from incomplete reports .
the numerical results validate the effectiveness and robustness of our approaches . in particular ,
in small - scale networks , the matrix completion approach achieves exact channel detection with a number of samples no more than @xmath0 of the number of channels in the network , while joint sparsity recovery achieves similar performance in large - scale networks .
_ keywords _ : collaborative spectrum sensing , matrix completion , compressive sensing , joint sparsity recovery . |
age estimates are required for studying the evolutionary history of star clusters . based on such age estimates ,
a general framework has been developed starting from formation within the dense cores of giant molecular clouds ( gmcs ) ; a stage where the young stars are completely obscured by their dusty gas cocoons ; an emerging stage where the clusters become visible in the infrared ( ir ) and then visible in the optical as stellar winds and supernovae blow away the gas and dust ; a stage where an expanding bubble of ionized gas is blown ; and later stages with no evidence of ionized gas ( e.g. , see the lada & lada 2003 review article ) . in the past , two general methods for estimating the ages of unresolved extragalactic star clusters have been used .
the first requires obtaining high quality spectroscopic observations covering wavelength regions of lines which change as a function of time .
examples of this approach include bica & alloin ( 1986 ) , schweizer & seitzer ( 1993 ) , whitmore ( 1999 ) , bastian et al .
( 2009 ) and wofford et al .
while these spectroscopic observations typically provide high quality age determinations , they are limited to age - dating relatively small numbers of bright clusters due to constraints on the brightness of clusters that can be observed spectroscopically in a reasonable amount of time .
the standard method for age - dating large numbers of extra - galactic clusters compares photometry in several broadband filters ( e.g. , _ ubvi _ ) with predictions from population synthesis models ( e.g. , chandar et al .
2010 ) .
comparisons of age estimates using spectroscopic and photometric observations of the same clusters show good agreement in most cases .
an early example of this is the famous searle et al .
( 1980 ) paper where they compare integrated four - filter _ ugvr _ photometry of 61 star clusters in the magellanic clouds with the strength of different spectral features ( e.g. , balmer lines , @xmath14 band , etc . )
a recent example is wofford et al .
( 2010 ) who compared ages estimated from spectral energy distribution ( sed ) fitting with those derived from uv spectroscopy for 14 young ( @xmath15 myr ) clusters in the nuclear starburst region of m83 and found that the photometric ages are within a factor of 1.4 of the spectroscopic ones .
whitmore et al .
( 2010 ) find similar agreement between their photometric age estimates and the spectroscopic estimates from bastian et al .
( 2009 ) for clusters in the antennae galaxies . while generally satisfactory , both the spectroscopic and photometric techniques have limitations .
it is therefore important to develop independent methods for estimating the ages of clusters , especially for cases where spectroscopic or multi - band photometric observations are not available and it is necessary to estimate ages based on morphological appearance alone . here ,
we develop two new methods for estimating the ages of young star clusters in nearby galaxies based on high - resolution images at optical wavelengths .
the first method uses the morphology of the ionized gas and its position relative to the cluster stars , as measured from narrowband h@xmath0 emission , to estimate ages ( @xmath16 ) for clusters younger than @xmath17 myr .
this method relies on the general premise that the distribution of h@xmath0 will be largely coincident with the distribution of optical light in the youngest clusters ( i.e. , @xmath10 few myr ) , will be in a small ring - like structure around the optical stellar emission in slightly older clusters where massive star winds and supernovae have had time to blow a bubble ( i.e. , @xmath18 myr ) , and will be in a larger ring - like bubble for still older clusters ( i.e. , 510 myr ) .
many past observational and theoretical studies of hii regions and `` supershells '' in the milky way and nearby galaxies have laid the groundwork for this method .
for example , walborn ( 2002 ; see also walborn & parker 1992 and especially walborn 2010 for related discussions ) outlined an evolutionary cluster sequence based on observed properties of several well known clusters and ob associations in the milky way with ages ranging from @xmath19 myr to @xmath5 myr .
this sequence was then used to illustrate observed changes with age in the visually brightest stars , in the ionized gas and dust content , and in the existence of red supergiants .
much of this sequence was based on spectroscopy of individual stars .
while we are unable to make similarly detailed observations at the distance of m83 , many of the same basic correlations and underlying physical processes are relevant for the age sequence outlined in this paper .
theoretically , several works have made predictions for the size evolution of an expanding hii region over time .
oey & clark ( 1997 , 1998 ) model the size evolution due to mass loss and supernova - injected energy from cluster stars , and assume that expanding bubbles `` stall '' when their internal pressure equals the ambient pressure in the ism .
these simulations predict a strong dependence of bubble size on both the age and mass of the central cluster .
( e.g. , weaver et al .
1977 , oey & massey 1995 , oey & garcia - segura 2004 , dopita et al . 2006a , 2006b ) .
our observations of clusters in m83 reveal a strong dependence of bubble size on cluster age , and possibly a weak dependence on cluster mass , as discussed in section 6 .
the second method developed in this paper uses the surface brightness fluctuations of cluster stars to estimate their ages .
young clusters have strong pixel - to - pixel flux variations , due to the presence of massive , luminous stars . as a cluster ages , the bright ,
short - lived massive stars disappear , and these fluctuations fade in strength .
this technique is especially useful in the range @xmath20100 myr , ages that can be somewhat difficult to deal with using the sed method , because the predicted integrated colors loop back on themselves .
our primary target for this study is the spiral galaxy m83 . at a distance of 4.5 mpc
( corresponding to a distance modulus of @xmath21 ; thim et al .
2003 , and a pixel scale of 0.876 pc pixel@xmath22 ) , m83 , nicknamed the `` southern pinwheel , '' is the nearest massive grand - design spiral galaxy .
it is a slightly barred galaxy , with a hubble type sab(s)c ( 3rc ) . in this work ,
we make use of observations taken with the wide - field camera 3 ( wfc3 ) , which were described in detail in dopita et al .
( 2010 ) and chandar et al .
briefly , the observations are part of the early release science project 1 ( ers1 ) program 11360 ( pi : oconnell ) and were taken in august , 2009 .
observations were taken in several broadband ( `` uv '' - f225w , `` u '' - f336w , `` b '' - f438w , `` v '' - f555w , `` i '' - f814w , `` j '' - f110w , and `` h '' - f160w ) , and narrowband filters ( [ oiii ] - f373n , h@xmath23 - f487n , [ oii ] f502n , h@xmath0 - f657n , [ sii ] - f673n , paschen@xmath23 - f128n , and [ feii ] - f164n ) . in the appendix
we briefly investigate clusters in m51 and find that the age versus morphological category relationship derived for m83 is appropriate for this galaxy as well .
this paper is organized as follows : in section 2 we outline our working scheme for an evolutionary cluster classification system based on observables at optical wavelengths . in section 3
we investigate the correlation between h@xmath0 morphology and sed age estimates while section 4 examines the correlation between the strength of surface brightness fluctuations within clusters and their ages .
a catalog of smaller hii regions , apparently ionized by `` single stars , '' is presented in section 5 .
we examine the effect of cluster mass on h@xmath0 bubble size in section 6 , and summarize our primary results in section 7 .
the early formative stages of star cluster evolution are best studied in the ir or microwave part of the spectrum , since the stars remain embedded in their placental dust cocoons for the first million years or so ( lada & lada 2003 ) .
for example , in nearby groups and clusters in the milky way , individual young stellar objects ( ysos ) can be age - dated using a classification scheme ranging from category 0 to iii , as developed by wilking et al .
( 1989 ) and andre et al .
( 1993 ) , based on near - ir observations .
in this paper we outline a working classification scheme for cluster evolution in external galaxies .
we focus on the later stages based on the optical portion of the spectrum .
our primary goals are to test how well we can use h@xmath0 morphology and surface brightness fluctuations to age date star clusters .
the basic categories are defined below . in all cases
we assume that the clusters have radial profiles that are broader than the point - spread function ( psf ) , hence confusion with individual stars is not a major issue ( see chandar et al .
2010 ) . examples of clusters in categories 3 through 6 are shown in figure [ fig:18plot ] .
m83 clusters in categories 1 and 2 will be discussed in a later paper which presents our wfc3 observations in the @xmath24 and @xmath25 bands . _
category 1 : _ * opaque dust cloud * - core of a gmc - dark region on optical image with no associated ir / uvis source - often studied using millimeter observations of the hcn line ( e.g. , gao & solomon 2004 ) 0.1 in _ category 2a : _ - * embedded cluster * - weak ir source with no optical ( i band ) counterpart ( i.e. , av @xmath12 10 ) - no ionized h@xmath0 gas visible 0.1 in _ category 2b : _ - * obscured cluster * - strong ir and weak optical source ( 3 @xmath10 av @xmath10 10 ) 0.1 in _ category 3 : _ - * emerging cluster * - ionized gas spatially coincident with cluster stars - cluster has a reddish color due to dust - surface brightness fluctuations from individual stars are relatively small since the brightest , evolved stars have not yet appeared 0.1 in _ category 4a : _ - * very young cluster * - ionized gas in a small bubble surrounding the cluster - cluster has a bluish color since most of the dust has been expelled - surface brightness fluctuations from individual stars are strong 0.1 in _ category 4b : _ - * young cluster * - similar to category 4a but the ionized gas is now in a large bubble surrounding the cluster ( i.e. , radii larger than approximately 20 pc ) - surface brightness fluctuations from individual stars reach maximum 0.1 in _ category 5a : _ - * young / intermediate - age cluster * - no ionized gas is observed - surface brightness fluctuations are still present , but weaker 0.1 in _ category 5b : _ - * intermediate - age cluster * - ionized gas and surface brightness fluctuations among cluster stars are not observed - cluster has a slightly redder color due to aging of stars 0.1 in _ category 6 : _ - * old cluster * - no ionized gas or surface brightness fluctuations among cluster stars are observed , - cluster appears yellow / red with no evidence of dust in vicinity
in principle , it should be possible to approximately estimate the ages of young ( @xmath17 myr ) star clusters from the size of the ionized gas ( h@xmath0 ) bubble that surrounds them , as outlined in sections 1 and 2 . in this section
we test this idea by selecting a representative catalog of young clusters primarily based on their apprearance in the narrowband h@xmath0 image ( i.e. , the red color in figure [ fig : large_image ] ) , and then visually classify each source based on the scheme outlined in section 2 .
each region is then matched with the apparent central source of ionization from the chandar et al .
( 2010 ) catalog , and compared with the ages from that paper based on sed fitting of the _ _ ubvi__h@xmath0 bands .
we do not , at this point , separate the clusters by mass ( the effect of cluster mass on bubble size is discussed in section 6 ) .
basic properties of the selected clusters are presented in table 1 , including position ( ra and decl . ) and photometric measurements ( @xmath26 , @xmath27 , and @xmath28 ) for the central cluster .
the top row of figure [ fig:18plot ] shows examples of clusters in categories 3 - 6 , with different h@xmath0 morphologies as outlined in section 2 .
the bottom set of panels shows the location of each cluster in the category in a @xmath27 versus @xmath28 two - color diagram , where the photometry is drawn from the catalog described in chandar et al .
these panels show how well our morphological categories group the clusters in color - color space .
predictions from single stellar population g. bruzual & s. charlot ( 2009 ; private communications , hereafter bc09 ) models are superposed for comparison .
as expected , clusters in the two youngest categories ( 3 and 4a ) have somewhat higher extinction than those in later categories , which results in their colors spreading out along the reddening vector . the middle set of panels present histograms of ages determined from the photometric dating method used in chandar et al .
( 2010 ) , for all clusters within a given category .
these age estimates are included in table 1 .
there is a clear trend for clusters in later categories to have older ages . for clusters with associated h@xmath0 emission ,
i.e. , categories 3 , 4a , 4b , we find an rms scatter of approximately 0.1 in @xmath29 within each category .
while our sed dating method includes photometry in the narrowband filter directly in the fit , the photometry is measured in a small ( 3 pixel radius ) aperture that misses most or all of the h@xmath0 emission after category 3 .
hence the age estimates from the sed fitting and from the h@xmath0 morphology are largely independent of one another .
the rms scatter in photometric ages for clusters identified within morphological categories 5 and 6 increases to @xmath11 in @xmath29 , primarily because morphology becomes a cruder estimate in this age regime .
the good overall correlation between morphological classification and photometric age estimates from chandar et al .
( 2010 ) is also apparent in figure [ fig : large_image ] .
this figure includes the sed ages for the clusters where morphological classifications have been determined using the prescripts defined in section 2 .
the region covers a @xmath30 kpc@xmath31 area of m83 .
we note in particular the intermediate - age and old clusters in the upper central part of the image which have a fuzzy appearance due to the absence of bright young stars .
the appendix shows similar figures for selected regions in m51 .
figure [ fig : ha_vs_sed ] plots the morphological category assigned here versus @xmath32yr ) based on the sed age estimates from chandar et al .
the top panel shows the full range of cluster ages and categories . there is a clear , albeit nonlinear , correlation , with a steeper correlation between morphological category and age for clusters younger than @xmath33 yr .
the nonlinearity is to be expected since h@xmath0 morphology evolves rapidly between the ages of 110 myr .
note that in figure [ fig : ha_vs_sed ] we have subdivided the morphological types into a finer grid of categories ( based on subjective estimates of the h@xmath0 bubble size ) than outlined in section 2 .
the bottom panel of figure [ fig : ha_vs_sed ] shows one of the main results of this work , _ that between 2 and 10 myr , the clusters in m83 show a strong correlation between their h@xmath0 morphology and age_. this suggests that the morphology of ionized gas alone can be used as an age indicator .
the best linear fit , shown in the figure , is given by : @xmath34 ( i.e. , a 9 @xmath35 correlation ) for @xmath36 , where @xmath37 is the morphological category defined in section 2 .
this relationship is only relevant when h@xmath0 is present ( i.e. , categories 3 and 4 ) . a similar , flatter , and less well - defined relationship exists for category 5 and higher , as shown in the top figure : @xmath38 ( i.e. , a 5 @xmath35 correlation ) for @xmath39
. we note that a similar correlation exists between our morphological categories and ages estimated by comparing color magnitude diagrams of individual stars with stellar tracks for stars around some of our objects , as will be discussed in a study of 50 regions in m83 by h. kim et al .
( 2011 , in preparation ) .
while the dispersion in the estimated sed ages within each morphological category is remarkably small ( see figure [ fig:18plot ] and table 2 ) , a close inspection reveals four outliers in category 5b .
three of these clusters ( 44034 , 18032 and 74692 ) have no associated h@xmath0 emission , have relatively small pixel - to - pixel flux variations ( discussed in section 4 ) , and have optical colors consistent with those predicted for extinction - free intermediate age clusters , and therefore almost certainly have ages @xmath40 .
the sed fitting , however , erroneously assigned them lower ages ( @xmath36 ) , and higher extinctions ( i.e. , @xmath41 , 0.36 , 0.50 mag ) .
the fourth cluster ( 10114 ) has significantly bluer colors than the other clusters in category 5b .
in retrospect , this cluster probably belongs in morphological category 4a , since there does appear to be a small amount of h@xmath0 emission associated with it . in this case
it is the morphological classification which appears to be in error rather than the sed age estimate ( the latter is @xmath42 ) .
overall , we find good agreement between the previously determined sed age and currently determined morphological category for 95% of the clusters ( i.e. , 84 out of 88 cases ) , demonstrating that _ both _
techniques are quite reliable .
while the good correlation between the morphological category and sed age found in figure 3 is encouraging , this approach has several limitations .
in particular , placing the clusters into the different categories is subjective and hence not easily automated .
an additional complication is that an isolated , full 360 degree ring with a single dominant central cluster is very rare .
partial rings , multiple loops , and multiple clusters in the region are more typical situations . here , we attempt to better quantify the relationship between h@xmath0 morphology and cluster age by measuring the radii of the most prominent `` coherent '' ring or partial shell of ionized gas associated with the cluster stars . while the measurement of h@xmath0 bubble size is also difficult to automate , this more quantitative approach provides the opportunity for improving the correlation still further , and for exploring the impact that other physical parameters ( e.g. , cluster mass ) have on this relationship .
figure [ fig : ha_vs_r ] plots the measured h@xmath0 bubble size versus photometric age for clusters younger than @xmath5 myr ( top panel ) , and bubble size versus morphological category ( bottom panel ) .
there is clearly a strong correlation which suggests that h@xmath0 morphology alone can be used as an age indicator for clusters younger than @xmath5 myr .
this figure hints at other intriguing results as well .
the correlation between age and bubble size is strongest for clusters younger than @xmath43 , with typical bubbles growing from a few pc to @xmath44 pc in size . after this time , between ages @xmath436.9
, there is a large range observed in bubble size , from @xmath44 pc to nearly 80 pc , and no strong dependence on age .
these trends are illustrated by the solid lines , which represent simple linear fits to observations in the two different regimes .
the best linear fits for the top panel are given by : @xmath45 for r @xmath10 20 pc , and @xmath46 for r @xmath12 20 pc .
similar fits are shown in the bottom panel , but numerical values are not included here since the use of morphological category makes them more qualitative .
the large , open circles in the top panel of figure [ fig : ha_vs_r ] show the clusters with @xmath47 and @xmath48 .
these will be discussed further in section 6 .
as discussed in section 2 , bright individual stars are expected to be visible within m83 clusters at ages younger than @xmath49 myr , giving the clusters a mottled appearance with relatively large pixel - to - pixel flux variations .
after this , the flux variations have mostly disappeared , leading to a more uniform appearance among cluster stars .
compare , for example , the images in categories 4b and 5a in the top panel of figure [ fig:18plot ] , where a number of bright individual cluster stars are clearly visible , with the images for categories 5b and 6 , where individual cluster stars are no longer observed .
figure 2 also shows a variety of cases where older clusters appear as fuzzy objects while single individual stars are clearly observed within the young clusters . here , we use the strength of these surface brightness fluctuations to develop a new method for estimating the ages of clusters in m83 .
this is reminiscent of the surface brightness fluctuation method used to estimate distances to early - type galaxies ( e.g. , tonry & schneider 1988 ) , but in our case the clusters are all at the same distance and the surface brightness fluctuations are caused by differences in age .
one must exercise caution when using this technique for other galaxies , since the relationships derived below for m83 would need to be renormalized for galaxies at different distances .
we first need to remove the overall radial gradient in the cluster luminosity profile , since this gradient typically dominates over the peak - to - peak variations in pixel flux within the cluster .
we accomplish this by creating a median divided image ( with a @xmath50 smoothing box size ) , which effectively flattens the radial profile of the cluster , leaving behind primarily the smaller scale pixel - to - pixel variations that we are interested in here .
we then measure the rms scatter in the fluxes of pixels located within a 10 pixel radius .
a more sophisticated method is being developed by c. kaleida et al .
( 2011 , in preparation ) where the rms variations are measured within the half - light or effective radius r@xmath51 of the cluster , hence accounting for clusters of different sizes in a more systematic manner .
figure [ fig : sbf ] shows a comparison of the f555w image and the median divided f555w image in a region including clusters younger than @xmath52 ( blue circles and labels ) and older than this ( red circles and labels ) .
the rms measurements are indicated in the bottom panel .
one can easily see that old clusters ( e.g. , nos .
65 and 70 ) nearly disappear from the median divided image , resulting in low values of the rms , while bright , individual stars are observed in the younger clusters ( e.g. , nos .
32 and 49 ) , resulting in a higher rms .
small , barely visible dust lanes also contribute to larger values of rms for the young clusters . in a few very concentrated clusters
there are small artifacts at their centers , for example in cluster no .
64 . however , the rms is dominated by pixels outside of this central region , hence this is a relatively small effect .
we note that for four clusters , identified in table 1 , we have used a slightly smaller radius to avoid single bright stars that are almost certainly unrelated to the cluster ( e.g. , nos .
66 , 69 , 70 and 79 ) . in the upper panel of figure [ fig : m83_rms_2plot ]
we plot the rms in the pixel - to - pixel flux measured in the @xmath53 band versus the morphological category .
the resolution is better here for clusters younger than @xmath54 yr ( i.e. , category 3 and 4 objects ) . in the lower panel of figure [ fig : m83_rms_2plot ]
we plot the rms vs. the sed age , which gives better resolution for clusters older than @xmath33 yr .
these figures show the second main result of our work , _ that older clusters have smaller rms values of the pixel - to - pixel flux variations .
_ however , we also note , especially in the top panel , that the rms is degenerate , and increases in strength over the age range 27 myr ( i.e. , morphological categories 3 and 4 ) , peaks in category 4 , corresponding to an age of approximately 510 myr , and decreases in strength for clusters with older ages and in later categories .
there are several possible explanations for this behavior .
the first , and probably most important effect , is that the brightest stars ( i.e. , red and blue super giants ) do not appear until about 5 myr , and are largely gone by about 20 myr .
two other evolutionary effects may potentially contribute to the low rms measured for the very young clusters .
( 1 ) the central part of the cluster , where it is difficult to detect individual stars due to the high background , might emerge from its dust cocoon before the outskirts .
( 2 ) clusters may start out very compact and expand with age . hence crowding
may make it impossible to detect individual stars at very early ages .
there is both theoretical and observational support for both of these conjectures , as discussed below .
these trends are illustrated by the lines in figure [ fig : m83_rms_2plot ] , which represent linear fits to observations .
the best fits for the top panel are given by : @xmath55 ( i.e. , a 5 @xmath35 correlation ) for category @xmath104.7 , and @xmath56 ( i.e. , a 6 @xmath35 correlation ) for category @xmath574.7 .
for the bottom panel , only the trend for older clusters is shown , with @xmath58 for @xmath39 .
we will better quantify the correlation between cluster age and the rms in pixel - to - pixel fluxes in a future paper using a significantly larger number of clusters .
we note that these relationships will be different for galaxies at different distances , and in different filters .
for example , fluctuations will generally be larger for older clusters ( 1001000 myr ) in the near - ir due to the presence of asymptotic giant branch ( agb ) stars .
values of rms pixel - to - pixel flux variations for each cluster are included in table 1 , and the mean values for clusters in each morphological category are included in table 2 .
the rms method is particularly promising for estimating the ages of clusters between 10 and 100 myr , a range which can be somewhat difficult for the sed method because the integrated colors of clusters loop back on themselves during this period .
clusters in this age range can be identified initially by their lack of ionized gas .
it is instructive to look at the outliers in figure [ fig : m83_rms_2plot ] , just as we did for the correlation between sed age and morphological category in section 3.1 . in the bottom panel
the cluster with rms = 0.03 and @xmath59 ( i d = 44034 ) is an apparent outlier .
it is also one of the sources which we believe , based on the discussion in section 3.1 , has an incorrect sed age estimate .
based on its measured colors and assuming @xmath60 instead of 0.28 , this cluster has a likely age of @xmath61 , more appropriate for it s low rms value .
similarly , many of the data points just to the right of this one are the same category 5b outliers discussed in section 3.1 that have suspect ages .
other data points in the same region of the @xmath29 versus rms plot are located in the large bubble just below the nucleus of m83 , where it is difficult to measure the rms accurately due to the very high background .
we note that the four clusters with rms @xmath62 and @xmath63 are all in morphological category 3 , and are largely responsible for the decline in the measured rms for the youngest clusters .
many of the trends discussed above are seen in table 2 , which lists the mean values of several parameters as a function of morphological category . in particular , the trends in @xmath64 , @xmath29 , and h@xmath0 shell radius ( when present ) are clearly evident .
in addition , the double - valued nature of pixel - to - pixel rms values is evident , with a peak at category 4a .
one correlation that was mentioned briefly is the trend of increasing cluster size with age , quantified by the increase in the concentration index ( ci ) from values around 2.5 for categories 3 and 4a to @xmath65 for categories 5b and 6 .
we find the same trend in our m51 data , which is briefly discussed in the appendix .
this effect appears to be real , and is likely related to the rapid expansion of the clusters ( see , for example , mackey & gilmore 2003 ; bastian et al .
2009 ; pfalzner 2009 ) .
several different physical mechanisms may be responsible for this expansion , including the expulsion of leftover ism due to feedback from massive stars ( e.g. , goodwin & bastian 2006 ; baumgardt & kroupa 2007 ) , and heating by binary stars and stellar mass black holes ( e.g. , mackey et al .
2008 ; see portegies zwart et al .
2010 for a review of this subject ) . the early expansion of star clusters will be discussed in more detail in a future paper ( r. chandar et al .
2011 , in preparation ) .
while the dominant sources of ionizing flux responsible for the h@xmath0 emission in m83 are massive , young star clusters , which were studied in the previous sections , there is also a population of compact hii regions which are ionized by what appear to be single stars . in this section
we identify and study a sample of 22 hii regions with very small h@xmath0 radii and an unresolved central point source ( based on their ci , i.e. , the magnitude difference between 0.5 and 3 pixel radii ; see chandar et al .
color images of the selected sources are shown in figure [ fig : target_mosaic ] .
only the brighter candidates have been retained for this first exploratory study ; the sample could be increased by a factor of two or more if fainter , less distinct objects were included .
the measured colors of these sources , shown in figure [ fig : mini_cmd ] , either coincide with the colors predicted for the bluest ( youngest ) stars in the upper left portion of the diagram , or are found downstream along the reddening vector .
basic properties of these sources , including positions , @xmath26 , ci , colors and measured radii of the h@xmath0 bubble , are compiled in table 3 .
we note that it is unlikely that all of these objects are actually individual massive stars , hence the use of quotes around `` single star . ''
some , or even most of these may be a dominant star among a close grouping of stars which are either too close to the primary source , or too faint , to be detected .
what we can say is that a single ( or very close binary ) star dominates the light profile , resulting in a ci that is indistinguishable from a single star .
if we assume that all of these objects have a similar age , and that the distribution in the two - color diagram is primarily due to reddening , we can correct for the effects of reddening and extinction .
we show the corrected photometry in a color magnitude diagram in the left panel of figure [ fig : mini_cmd ] . here
, we have assumed an intrinsic color of @xmath66 , the color of the bluest object , and solve for the @xmath27 color excess of each hii region .
we find that this procedure moves most of the objects to a very young isochrone ( 4 myr is shown ) , although three sources scatter to the left of the models . these may be especially young stars , or the offset may result from observational uncertainties , since these are the three most reddened sources , and hence have the largest corrections . while many of these single - star hii ( i.e. , sshii ) regions are found near large regions of recent star formation , several of them are quite isolated ( see figure 3 from whitmore 2010 ) , raising the possibility that these massive stars formed in the field rather than in clusters or associations .
for example , five of the 22 sshii regions are @xmath671 kpc away from any region of active star formation .
if these are stars that have been dynamically ejected from their birthsites , i.e. , runaways , they must have velocities @xmath67200 km / s .
however , they do not appear to have the prominent bow - front morphologies typical of many runaway stars in the galaxy and magellanic clouds ( e.g. , gvaramadze et al .
velocity measurements are needed to establish if these massive stars formed in the field or are runaways from larger star - forming regions .
as summarized in the introduction , the spatial relationship between young @xmath68 yr star clusters and their hii regions has been previously studied both observationally and theoretically . to our knowledge however , no study has yet systematically measured the sizes of h@xmath0 bubbles in a nearby galaxy and correlated these sizes with the properties of their ionizing star clusters , as we have done here .
we have demonstrated that in m83 there is a good correlation between bubble size and age for clusters younger than @xmath43 , and that clusters with ages @xmath436.9 have a larger range in bubble size .
this suggests that some of the expanding bubbles at these older ages may have effectively `` stalled . ''
theoretical work suggests that bubble size should depend not only on the ambient pressure in the ism and the age of the cluster , but also on cluster mass . to look for evidence of a mass - dependence on bubble size
, we compare the sizes measured for h@xmath0 bubbles associated with @xmath696.9 clusters that are more and less massive than @xmath70 . at these ages
the bubbles are presumably approaching their stall radius , and the effect of cluster mass should be more apparent .
the large , open circles in the top panel of figure [ fig : ha_vs_r ] show the sizes of clusters with @xmath47 and @xmath48 .
while clusters at high and low masses have an overlapping range in cluster size , there is a tendency for more massive clusters to have larger bubbles , with median sizes of 18 pc and 41 pc for @xmath71 and @xmath72 , respectively .
a formal fit of h@xmath0 radius versus @xmath73 for these clusters gives a slope of @xmath74 , suggesting that there may be a weak correlation between bubble size and mass .
future studies that include a larger number of sources , selected in a more systematic way , are needed to confirm this result .
however , other effects may also be important .
for example , stochasticity in the number of massive stars formed in lower mass clusters can strongly affect their integrtated colors and hence their estimated ages . in this case , a lower mass cluster can have an intrinsically redder color than a higher mass cluster of the same age ( e.g. , fouesneau & lancon 2010 ) , and hence an older estimate for the age of the cluster .
in addition to the correlation between h@xmath0 bubble size and cluster age ( and possibly mass ) , we also found that the rms variations in pixel - to - pixel brightness correlates with cluster age .
this result has applications beyond star clusters , and can be used to constrain the ages of stellar populations in general .
for example , portions of m83 itself that contain young stars and star clusters , such as along the spiral arms , have larger pixel - to - pixel flux variations than portions of m83 dominated by older stars , such as between the arms ( see h. kim et al .
2011 , in preparation for a discussion ) .
this is also true in large portions of m82 and in portions of the tidal tails of galaxy mergers , where regions dominated by intermediate age star clusters ( @xmath67100 myr ) have small fluctuations in the surface brightness of the field stars .
we have used observations taken with the newly installed wfc3 camera on - board the _ hubble space telescope _ to develop two independent methods for age - dating young star clusters in the nearby spiral galaxy m83 .
our primary results are summarized below .
a working classification system , largely based on h@xmath0 morphology and pixel - to - pixel flux variations , was developed to map an observed age sequence onto a proposed sequence of cluster evolution .
the underlying evolutionary picture includes the formation of dense cores in giant molecular clouds ( gmcs ) ; a stage where the young stars are completely obscured by their dust cocoon ; an emerging stage where the clusters become visible in the ir , and then in the optical as stellar winds and supernovae blow away the dust ; a stage where an expanding h@xmath0 bubble is blown and the existence of very bright young stars leads to large pixel - to - pixel flux variations ; and later stages with no evidence of h@xmath0 and diminishing pixel - to - pixel flux variations . \2 .
we found that h@xmath0 morphology , i.e. , the size of the ionized gas bubble , provides a viable method for age - dating clusters in the range 110 myr .
this method is based on the simple premise that the gas in very young ( @xmath2 few myr ) clusters is largely coincident with the cluster stars , is in a small , ring - like structure surrounding the stars in slightly older clusters since the winds from massive stars have had time to push out the natal gas ( e.g. , @xmath75 myr ) , and is in a larger ring - like bubble for still older clusters ( i.e. , @xmath67510 myr ) .
if no h@xmath0 is associated with a cluster it is generally older than @xmath6710 myr .
we first made _ qualitative _ estimates based on the classification scheme outlined above , and find that the ages of the clusters , as determined from the sed method described in chandar et al .
( 2010 ) , correlate well with the morphological categories , with a scatter of @xmath670.1 in @xmath29 within each category for the clusters with h@xmath0 emission , and a scatter of @xmath670.5 in @xmath29 for the older clusters .
we then _ quantified _ this technique by correlating the measured radii of the most conspicuous h@xmath0-emitting ring or shell which appears to be physically related to the cluster with the sed ages determined in chandar et al .
we found tentative evidence for a weak correlation between bubble size and cluster mass , but a larger , more objectively selected sample will be required to confirm this .
we then used pixel - to - pixel flux variations to age - date clusters .
this technique is based on the fact that individual stars are bright enough to be visible within clusters when they are young ( e.g. , m@xmath6 , the approximate detection level , for the brightest stars with ages @xmath10100 myr ) , leading to relatively large pixel - to - pixel variations in flux . the strength of the fluctuations peaks in clusters with ages of @xmath67510 myr , presumably because this is when the brightest stars ( e.g. , red and blue super giants ) appear .
the number of luminous , evolved stars falls off for both younger and older clusters .
this degeneracy in age can be broken using other age indicators such as the h@xmath0 morphology .
the technique is especially useful for identifying clusters older than 100 myr .
a by - product of this study was the identification of 22 `` single - star '' hii regions in m83 . by assuming that all of these objects have a similar age , and that the distribution in the two - color diagram is primarily due to reddening , we corrected for the effects of reddening and extinction .
we found that this procedure moves most of the objects to a very young isochrone with an age approximately 4 myr .
some of these massive stars are located far from any star - forming region , indicating that they either formed in the field or were dynamically ejected from their birthsites at very high velocities .
these sshii regioins will be studied in more detail in h. kim et al .
( 2011 , in preparation ) . in the future , we will extend the classification system into the near - ir ( i.e. , categories 1 and 2 ) using our j and h observations .
we will also calculate the energy budget of cluster stars and compare with physical properties of the clusters and the ism .
finally , we will extend this analysis to other galaxies in the ers1 sample ( including `` low pressure '' systems such as the dwarf starburst galaxy ngc 4214 ) in order to determine whether the relationships are universal or are strongly dependent on environment .
we thank zolt levay for making the color images used in figures [ fig:18plot ] and [ fig : large_image ] .
this paper is based on observations taken with the nasa / esa _ hubble space telescope _ obtained at the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
the paper makes use of early release science observations made by the wfc3 science oversight committee .
we are grateful to the director of stsci for awarding director s discretionary time for this program .
r. c. is grateful for support from the nsf through career award 0847467 .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
the spiral galaxy m51 was originally used to perform a pilot study to test how well h@xmath0 morphology can be used to estimate cluster ages .
examples of the results are included in figures [ fig : m51_ha_morph_1 ] and [ fig : m51_ha_morph_2 ] .
these are similar to figure [ fig : large_image ] for m83 and show that even at twice the distance of m83 the quality of the images are comparable , and the morphological categories match sed age estimates at a similar level . a more complete analysis will be included in a future paper ( r. chandar et al . 2011 ; in preparation ) .
1&65815&204.244637&@xmath7629.8507043&@xmath766.10 & 2.57&@xmath761.35&@xmath760.12&0.24&6.52&1.1e3 & 3&2&0.082 + 2&63611&204.2426354&@xmath7629.8515924&@xmath766.42 & 2.35&@xmath761.27&0.07&0.34&6.54&2.1e3 & 3&1&0.091 + 3&41736&204.251687&@xmath7629.8640556&@xmath768.70 & 2.68&@xmath760.21&0.80&0.50&6.7&2.6e4 & 3&7&0.064 + 4&39744&204.2524132&@xmath7629.8651911&@xmath7611.03 & 2.58&@xmath761.33&0.23&0.30&6.56&1.2e5 & 3&7&0.119 + 5&49651&204.2564416&@xmath7629.8586912&@xmath767.17 & 2.50&@xmath760.38&0.68&0.50&6.48&7.1e3 & 3&3&0.063 + 6&32461&204.2880034&@xmath7629.8692326&@xmath767.67 & 2.62&@xmath761.36&0.30&0.44&6.52&8.6e3 & 3&5&0.073 + 7&53072&204.2889299&@xmath7629.8569125&@xmath766.64 & 2.61&@xmath760.93&0.56&0.50&6.48&4.2e3 & 3&4&0.076 + 8&93801&204.2795357&@xmath7629.8392001&@xmath767.10 & 2.50&@xmath761.37&0.08&0.32&6.54&3.6e3 & 3&1&0.106 + 9&50287&204.2865616&@xmath7629.8582915&@xmath767.35 & 2.66&@xmath760.54&0.69&0.50&6.70&7.4e3 & 3&6&0.072 + 10&59214&204.2818977&@xmath7629.8537395&@xmath767.56 & 2.53&@xmath761.08&0.36&0.50&6.54&9.3e3 & 3&6&0.105 + 11&24767&204.2493318&@xmath7629.8729735&@xmath766.89 & 2.32&@xmath760.46&0.70&0.50&6.56&5.1e3 & 3&7&0.096 + 12&58013&204.2475174&@xmath7629.8543366&@xmath766.49 & 2.43&@xmath761.29&0.15&0.42&6.46&5.4e3 & 3&7&0.104 + 13&52726&204.2458956&@xmath7629.8570619&@xmath768.13 & 2.78&@xmath761.50&@xmath760.06&0.26&6.42&1.2e4 & 3&7&0.130 + 14&12721&204.2908632&@xmath7629.8785688&@xmath765.98 & 2.43&@xmath761.30&0.32&0.50&6.46&4.3e3 & 3&5&0.070 + 15&19354&204.2525635&@xmath7629.8755537&@xmath766.06 & 2.83&@xmath761.39&0.27&0.44&6.44&2.5e3 & 3&7&0.065 + 16&51616&204.2799185&@xmath7629.8575701&@xmath769.31 & 2.43&@xmath761.33&@xmath760.02&0.20&6.58&1.8e4 & 4a&6&0.131 + 17&49790&204.2546965&@xmath7629.8585957&@xmath767.88 & 2.27&@xmath760.85&0.27&0.42&6.68&1.1e4 & 4a&10&0.091 + 18&17035&204.2906967&@xmath7629.8766101&@xmath767.23 & 2.24&@xmath761.48&@xmath760.16&0.00&6.70&1.8e3 & 4a&15&0.119 + 19&66897&204.2654298&@xmath7629.8502774&@xmath768.60 & 2.96&@xmath761.50&@xmath760.12&0.00&6.72&4.4e3 & 4a&20&0.134 + 20&39842&204.2520397&@xmath7629.8651314&@xmath7610.94 & 2.71&@xmath761.51&@xmath760.15&0.08&6.58&5.2e4 & 4a&13&0.086 + 21&31457&204.2855284&@xmath7629.8698283&@xmath768.83 & 2.19&@xmath761.31&@xmath760.03&0.04&6.78&8.9e3 & 4a&17&0.164 + 22&74860&204.262148&@xmath7629.8474511&@xmath768.59 & 2.67&@xmath761.43&@xmath760.14&0.00&6.78&5.5e3 & 4a&21&0.141 + 23&89929&204.2581863&@xmath7629.8412261&@xmath766.79 & 2.23&@xmath761.35&@xmath760.05&0.04&6.72&1.3e3 & 4a&17&0.118 + 24&25607&204.2492229&@xmath7629.8725982&@xmath767.85 & 2.61&@xmath760.66&0.42&0.50&6.70&1.2e4 & 4a&21&0.120 + 25&75041&204.2775285&@xmath7629.847385&@xmath767.31 & 2.24&@xmath761.61&@xmath760.15&0.12&6.52&2.7e3 & 4a&7&0.137 + 26&64462&204.2811447&@xmath7629.8512236&@xmath768.19 & 2.30&@xmath760.81&0.34&0.46&6.68&1.7e4 & 4a&7&0.121 + 27&61923&204.263904&@xmath7629.8523354&@xmath766.02 & 2.52&@xmath761.28&0.06&0.36&6.54&1.5e3 & 4a&9&0.105 + 28&21601&204.2861725&@xmath7629.8744932&@xmath769.59 & 2.27&@xmath761.48&@xmath760.12&0.00&6.74&1.5e4 & 4a&15&0.158 + 29&62771&204.2811659&@xmath7629.8519608&@xmath767.25 & 2.27&@xmath760.96&0.28&0.38&6.70&5.6e3 & 4a&16&0.130 + 30&13265&204.2547344&@xmath7629.8783208&@xmath767.64 & 2.44&@xmath761.63&@xmath760.27&0.00&6.64&2.3e3 & 4a&13&0.148 + 31&5990&204.2551409&@xmath7629.8813282&@xmath768.17 & 2.47&@xmath761.53&@xmath760.26&0.00&6.72&3.9e3 & 4a&46&0.124 + 32&67958&204.2887307&@xmath7629.8499286&@xmath769.84 & 3.20&@xmath761.35&@xmath760.04&0.02&6.78&1.1e4 & 4a&19&0.161 + 33&66054&204.2644967&@xmath7629.8506016&@xmath769.97 & 2.63&@xmath761.45&0.39&0.00&6.88&2.8e4 & 4a&49&0.127 + 34&69793&204.2694643&@xmath7629.8493238&@xmath7611.41 & 2.63&@xmath761.48&@xmath760.16&0.00&6.78&7.5e4 & 4a&25&0.136 + 35&7716&204.2816067&@xmath7629.8807039&@xmath769.17 & 2.47&@xmath761.43&0.41&0.00&6.88&1.4e4 & 4b&68&0.151 + 36&36651&204.2524088&@xmath7629.8668398&@xmath7612.17 & 3.09&@xmath761.58&@xmath760.22&0.00&6.78&1.1e5 & 4b&&0.070 + 37&35937&204.2537197&@xmath7629.8672182&@xmath7610.31 & 2.79&@xmath761.24&0.45&0.02&6.90&4.2e4 & 4b&&0.052 + 38&54268&204.2932226&@xmath7629.8563319&@xmath768.57 & 2.87&@xmath760.91&0.79&0.34&6.86&1.7e4 & 4b&32&0.108 + 39&23366&204.2529648&@xmath7629.8736403&@xmath769.39 & 2.43&@xmath761.29&@xmath760.08&0.02&6.78&1.3e4 & 4b&41&0.116 + 40&27423&204.2896066&@xmath7629.8718058&@xmath768.38 & 2.34&@xmath760.99&0.50&0.36&6.78&1.4e4 & 4b&71&0.170 + 41&78807&204.2629122&@xmath7629.8459862&@xmath769.71 & 2.46&@xmath761.55&@xmath760.23&0.00&6.78&1.6e4 & 4b&45&0.169 + 42&36729&204.2525635&@xmath7629.8667944&@xmath7612.15 & 3.08&@xmath761.45&0.00&0.00&6.74&1.0e5 & 4b&&0.073 + 43&70777&204.2607456&@xmath7629.8489709&@xmath768.18 & 2.68&@xmath761.23&0.49&0.08&6.88&6.7e3 & 4b&43&0.087 + 44&70769&204.2650559&@xmath7629.8489734&@xmath767.77 & 2.92&@xmath761.34&@xmath760.08&0.00&6.76&2.1e3 & 4b&18&0.108 + 45&37589&204.2519048&@xmath7629.8663287&@xmath7611.96 & 3.20&@xmath761.13&0.46&0.24&6.74&1.4e5 & 4b&&0.064 + 46&49893&204.2897929&@xmath7629.8585381&@xmath769.76 & 2.74&@xmath761.49&@xmath760.24&0.00&6.78&1.6e4 & 4b&48&0.189 + 47&37095&204.2523893&@xmath7629.8665896&@xmath7611.18 & 2.79&@xmath761.32&0.35&0.00&6.88&8.0e4 & 4b&&0.059 + 48&78154&204.257905&@xmath7629.8462475&@xmath768.28 & 2.76&@xmath760.86&0.18&0.00&7.86&3.4e4 & 5a&&0.080 + 49&66216&204.2897363&@xmath7629.8505371&@xmath768.87 & 2.56&@xmath760.98&0.70&0.34&6.84&2.5e4 & 5a&&0.145 + 50&76156&204.2556622&@xmath7629.8469956&@xmath7610.21 & 3.15&@xmath761.08&0.62&0.06&7.48&1.0e5 & 5a&&0.089 + 51&66123&204.2903301&@xmath7629.8505755&@xmath769.39 & 2.50&@xmath761.11&0.59&0.22&6.86&3.0e4 & 5a&&0.125 + 52&66069&204.2844254&@xmath7629.8505955&@xmath768.88 & 3.13&@xmath760.78&0.47&0.06&7.81&4.5e4 & 5a&&0.067 + 53&85836&204.2694279&@xmath7629.8431197&@xmath769.43 & 3.25&@xmath761.12&0.18&0.10&6.78&8.7e3 & 5a&&0.093 + 54&40779&204.2926898&@xmath7629.8645926&@xmath7610.17 & 3.06&@xmath760.65&0.49&0.08&7.96&2.1e5 & 5a&&0.044 + 55&30950&204.2578713&@xmath7629.8701108&@xmath7610.85 & 2.94&@xmath761.08&0.60&0.08&7.49&2.5e5 & 5a&&0.085 + 56&83925&204.2643496&@xmath7629.8439131&@xmath768.71 & 2.79&@xmath760.97&0.77&0.14&7.44&4.3e4 & 5a&&0.129 + 57&44034&204.2499199&@xmath7629.8626794&@xmath769.56 & 3.10&@xmath760.82&0.44&0.28&6.78&2.1e4 & 5b&&0.030 + 58&17159&204.2616672&@xmath7629.8765474&@xmath768.22 & 3.18&@xmath760.42&0.57&0.16&8.06&4.3e4 & 5b&&0.052 + 59&25716&204.2827697&@xmath7629.8725452&@xmath767.28 & 2.76&@xmath760.14&0.42&0.00&8.41&2.6e4 & 5b&&0.054 + 60&18032&204.2849161&@xmath7629.876142&@xmath767.72 & 2.98&@xmath760.72&0.61&0.36&6.78&5.7e3 & 5b&&0.064 + 61&74692&204.2760088&@xmath7629.8475135&@xmath766.99 & 3.17&@xmath760.04&0.70&0.50&6.74&3.3e3 & 5b&&0.075 + 62&85964&204.2565326&@xmath7629.843069&@xmath767.59 & 2.99&@xmath760.33&0.44&0.00&8.36&2.7e4 & 5b&&0.049 + 63&14748&204.2752747&@xmath7629.8776532&@xmath767.14 & 3.29&@xmath760.11&0.40&0.00&8.36&1.2e4 & 5b&&0.038 + 64&65733&204.2851579&@xmath7629.8507302&@xmath769.82 & 2.73&@xmath760.94&0.61&0.10&7.59&1.3e5 & 5b&&0.089 + 65&65479&204.2862739&@xmath7629.8508316&@xmath769.06 & 3.19&@xmath760.37&0.42&0.00&8.31&7.8e4 & 5b&&0.050 + 66&65304&204.2849235&@xmath7629.8508999&@xmath768.51 & 3.39&0.07&0.80&0.18&8.41&5.8e4 & 5b&&0.039 + 67&10114&204.2664853&@xmath7629.8796816&@xmath767.17 & 2.64&@xmath761.53&@xmath760.33&0.00&6.64&1.4e3 & 5b&&0.105 + 68&58911&204.2807662&@xmath7629.8539005&@xmath767.69 & 2.88&0.15&1.55&0.32&9.11&2.9e5 & 5b&&0.050 + 69&58363&204.2831825&@xmath7629.8541717&@xmath767.23 & 3.27&0.27&0.54&0.00&8.86&2.9e4 & 5b&&0.034 + 70&66553&204.2887134&@xmath7629.8504088&@xmath768.06 & 3.23&0.64&1.08&0.20&9.01&1.5e5 & 5b&&0.035 + 71&55591&204.2710795&@xmath7629.8556304&@xmath767.27 & 3.14&0.34&0.74&0.04&8.86&4.1e4 & 5b&&0.036 + 72&94866&204.2790178&@xmath7629.8386224&@xmath768.34 & 3.28&0.06&0.39&0.00&8.46&4.0e4 & 5b&&0.031 + 73&3113&204.2651882&@xmath7629.8828492&@xmath766.93 & 3.28&0.04&0.89&0.00&9.16&3.8e4 & 6&&0.031 + 74&12505&204.2762954&@xmath7629.8786591&@xmath766.67 & 3.26&0.44&0.97&0.00&9.06&2.2e4 & 6&&0.029 + 75&24528&204.2890323&@xmath7629.8730866&@xmath769.15 & 3.13&@xmath760.14&0.80&0.22&8.36&1.8e5 & 6&&0.040 + 76&55985&204.2695705&@xmath7629.8553997&@xmath767.97 & 2.91&0.02&1.01&0.00&9.26&2.1e5 & 6&&0.040 + 77&18044&204.2956504&@xmath7629.8761376&@xmath768.29 & 3.24&0.12&0.54&0.00&8.61&5.3e4 & 6&&0.034 + 78&54416&204.2898224&@xmath7629.8562536&@xmath768.01 & 3.15&@xmath760.37&0.91&0.38&7.96&6.1e4 & 6&&0.042 + 79&46572&204.2634717&@xmath7629.8608741&@xmath768.54 & 3.05&0.18&0.81&0.04&8.91&1.7e5 & 6&&0.058 + cccccccc 3 & @xmath77 & @xmath78 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 + 4a & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 + 4b & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 + 5a & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & & @xmath103 + 5b & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & & @xmath109 + 6 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & & @xmath115 + 1&3550&204.2548844&@xmath7629.8825632&@xmath765.51 & 2.13&@xmath761.63&@xmath760.22&9 + 2&3916&204.2554865&@xmath7629.8823689&@xmath765.19 & 2.30&@xmath761.60&@xmath760.28&8 + 3&10373&204.2549063&@xmath7629.8795771&@xmath765.24 & 2.28&@xmath761.49&@xmath760.16&7 + 4&11098&204.2672170&@xmath7629.8792752&@xmath766.89 & 2.02&@xmath761.64&@xmath760.33&4 + 5&29250&204.2745636&@xmath7629.8709998&@xmath764.41 & 2.09&@xmath761.21&0.02&8 + 6&35101&204.2745919&@xmath7629.8676963&@xmath765.23 & 2.11&@xmath761.49&@xmath760.39&6 + 7&37778&204.2667959&@xmath7629.8662227&@xmath765.93 & 2.21&@xmath761.13&0.08&9 + 8&47544&204.2823703&@xmath7629.8602099&@xmath765.86 & 2.10&@xmath761.51&@xmath760.14&7 + 9&47772&204.2689918&@xmath7629.8600630&@xmath765.09 & 2.12&@xmath761.31&0.27&7 + 10&48660&204.2894733&@xmath7629.8594131&@xmath764.26 & 2.25&@xmath761.62&@xmath760.48&2 + 11&49679&204.2535423&@xmath7629.8586761&@xmath765.78 & 2.18&@xmath760.49&0.32&9 + 12&50923&204.2552793&@xmath7629.8579399&@xmath766.13 & 2.23&@xmath761.15&0.25&5 + 13&51301&204.2887045&@xmath7629.8577311&@xmath765.19 & 2.18&@xmath761.27&0.15&2 + 14&52904&204.2462139&@xmath7629.8569822&@xmath764.69 & 2.23&@xmath761.59&0.14&1 + 15&56031&204.2828693&@xmath7629.8553764&@xmath766.15 & 2.18&@xmath761.36&0.02&6 + 16&56485&204.2849019&@xmath7629.8551313&@xmath766.19 & 2.04&@xmath761.42&0.06&8 + 17&56492&204.2829518&@xmath7629.8551272&@xmath766.24 & 2.24&@xmath760.12&1.07&9 + 18&56937&204.2841803&@xmath7629.8548844&@xmath768.06 & 2.13&@xmath760.21&0.90&6 + 19&60624&204.2516243&@xmath7629.8530186&@xmath764.99 & 2.20&@xmath761.28&@xmath760.02&7 + 20&61345&204.2496746&@xmath7629.8526196&@xmath764.79 & 2.23&@xmath761.25&@xmath760.32&4 + 21&72114&204.2810230&@xmath7629.8485123&@xmath766.34 & 2.08&@xmath760.77&0.33&2 + 22&88339&204.2804302&@xmath7629.8420518&@xmath764.49 & 2.30&@xmath761.08&@xmath760.13&7 + | we use new wfc3 observations of the nearby grand design spiral galaxy m83 to develop two independent methods for estimating the ages of young star clusters .
the first method uses the physical extent and morphology of h@xmath0 emission to estimate the ages of clusters younger than @xmath1 myr .
it is based on the simple premise that the gas in very young ( @xmath2 few myr ) clusters is largely coincident with the cluster stars , is in a small , ring - like structure surrounding the stars in slightly older clusters since massive star winds and supernovae have had time to push out the natal gas ( e.g. , @xmath3 myr ) , and is in a larger ring - like bubble for still older clusters ( i.e. , @xmath410 myr ) . if no h@xmath0 is associated with a cluster it is older than @xmath5 myr .
the second method is based on an observed relation between pixel - to - pixel flux variations within clusters and their ages .
this method relies on the fact that the brightest individual stars in a cluster are most prominent at ages around 10 myr , and fall below the detection limit ( i.e. , m@xmath6 ) for ages older than about 100 myr .
older clusters therefore have a smoother appearance and smaller pixel - to - pixel variations .
the youngest clusters also have lower flux variations , hence the relationship is double valued .
this degeneracy in age can be broken using other age indicators such as h@xmath0 morphology .
these two methods are the basis for a new morphological classification system which can be used to estimate the ages of star clusters based on their appearance .
we compare previous age estimates of clusters in m83 determined from fitting _ _ ubvi__h@xmath0 measurements using predictions from stellar evolutionary models with our new morphological categories and find good agreement at the @xmath7% level .
the scatter within categories is @xmath8 dex in @xmath9 for young clusters ( @xmath1010 myr ) and @xmath11 dex for older ( @xmath1210 myr ) clusters .
a by - product of this study is the identification of 22 `` single - star '' hii regions in m83 , with central stars having ages @xmath13 myr . |
let @xmath1 and @xmath2 be independent homogeneous poisson processes on @xmath0 , with finite intensities @xmath5 and @xmath6 , respectively .
furthermore , let @xmath3 and @xmath4 be probability laws on the strictly positive integers and @xmath7 and @xmath8 random variables with laws @xmath3 and @xmath4 , respectively .
we shall study translation invariant bipartite simple graphs whose vertices are the points in @xmath1 and @xmath2 and where , conditionally on @xmath1 and @xmath2 , the degrees of the vertices are i.i.d . with laws @xmath3 and @xmath4 , respectively . in @xcite , it is shown that such graphs exist if and only if @xmath9= \lambda_{\mathcal{b } } \mathbb{e}[y].\ ] ] here we focus on the question on whether there is an infinite component or not in a particular type of such graphs referred to as stable graphs .
our main result is that , if all degrees are almost surely equal to 2 , then all components are finite almost surely .
this implies that the percolation properties of the model are different compared to the case with a single poisson process , where there are strong indications that there is an infinite component when all vertices have degree 2 ; see @xcite .
we also present simulation results for other degree distributions .
first we formally define the objects that we will work with .
the * support * ( or point - set ) of the process @xmath1 is the random set @xmath10:= \ { x \in \mathbb{r}^d : \mathcal{r}(\{x\})>0\}$ ] and its points are called * red points*. analogously , @xmath11 $ ] denotes the point - set of the process @xmath2 , and its points are called * blue points*. in general , for a random point measure @xmath12 , we write @xmath13 $ ] for its support .
let @xmath14 be a random integer - valued measure on @xmath15 , with the same support as @xmath1 , and which , conditionally on @xmath1 , assigns i.i.d .
values with law @xmath3 to the elements of @xmath10 $ ] . similarly , for @xmath2 , let @xmath16 assign i.i.d .
values with law @xmath4 to the elements of @xmath11 $ ] .
the pairs @xmath17 and @xmath18 are independent marked poisson processes with positive integer - valued marks . for @xmath19
$ ] and @xmath20 $ ] , we write @xmath21 for @xmath22 and @xmath23 for @xmath24 , and we interpret this as the number of stubs ( half - edges ) at the red point @xmath25 and the number of stubs at the blue point @xmath26 , respectively . sometimes we refer to the stubs as red or blue depending on the color of the point to which they are attached . a matching scheme for a marked process @xmath27 is a point process @xmath28 on the space of unordered pairs of points in @xmath15 , with the property that a.s . for every pair
@xmath29 we have that @xmath30 , and such that in the graph with vertex set @xmath10 $ ] and edge set @xmath31 $ ] , each vertex @xmath25 has degree @xmath21 .
we refer to this as a * one - color multi - matching scheme*. a matching scheme for two marked processes @xmath17 and @xmath18 is a point process @xmath28 on the space of unordered pairs of points in @xmath15 with the property that almost surely , for every unordered pair @xmath32 $ ] , we have @xmath33 $ ] and @xmath34 $ ] , and such that in the bipartite graph with vertex set @xmath10\cup[\mathcal{b}]$ ] and edge set @xmath31 $ ] , each vertex @xmath19 $ ] and each vertex @xmath20 $ ] has degree @xmath21 and @xmath23 , respectively .
we refer to this as a * two - color multi - matching scheme*. if a vertex @xmath19 $ ] ( or @xmath20 $ ] ) has degree at most @xmath21 ( or @xmath23 ) , we talk about a * two - color partial multi - matching*. we only consider matching schemes @xmath28 that are * simple * ( that is , the resulting graph has no self - loops and no multiple edges ) and * translation invariant * ( that is , @xmath28 is invariant in law under the action of all translations of @xmath0 ) .
it is shown in @xcite that a two - color multi - matching scheme with these properties exist if and only if ( [ des ] ) holds , including also the case when @xmath35=\mathbb{e}[y]=\infty$ ] .
one - color multi - matchings have previously been studied in @xcite and generalize one - color matchings , which is the particular case in which each vertex has degree 1 ; see @xcite .
two - color matchings have been studied in @xcite and were generalized to multi - matchings in @xcite .
the focus in this paper is on a particular type of two - color multi - matching referred to as stable multi - matching , inspired by gale - shapley stable marriage @xcite . here and throughout
, @xmath36 denotes the euclidean norm on @xmath0 .
a one - color ( two - color ) multi - matching is said to be a * stable multi - matching * if almost surely , for any two distinct points @xmath37 ( of different color ) , either they are linked by an edge or at least one of @xmath25 and @xmath26 has no incident edges longer than @xmath38 .
for the two - color case it was proved in @xcite[proposition 1 ] that , if both @xmath3 and @xmath4 have finite means such that ( [ des ] ) holds , then there is an a.s .
unique stable two - color multi - matching that can be obtained from the following procedure .
given the two point configurations @xmath10 $ ] and @xmath11 $ ] , call two points @xmath25 and @xmath26 compatible if they are of different color and say that two compatible points @xmath25 and @xmath26 are mutually closest if @xmath25 is the closest compatible point of @xmath26 and vice versa .
first create an edge between each pair of mutually closest compatible points in @xmath10\cup[\mathcal{b}]$ ] and remove one stub from each of these points . in the next step , consider the set of points which still have at least one stub on them after the previous step .
call two such points compatible if they are of different color and if no edge was created between them in the previous step .
create an edge between each pair of compatible mutually closest points , and remove one stub from each of these points .
iterate the algorithm indefinitely .
the procedure is a straightforward generalization of the analogue construction for the one - color model .
in @xcite it was shown for the one - dimensional one - color model that an infinite component is unique .
our first result generalizes this to the two - color case .
[ unique ] for two poisson processes on @xmath0 with intensities @xmath5 and @xmath6 and any two degree distributions satisfying ( [ des ] ) , in the two - color stable multi - matching , there is at most one infinite component . as for percolation properties ,
it was proved in @xcite that the two - color stable multi - matching almost surely does not generate an infinite component when the only possible degrees for both processes are 1 and 2 , with a strictly positive probability of degree 1 for at least one of the processes .
furthermore , for @xmath39 , it was shown that there is an integer @xmath40 such that , if all vertices almost surely have degree at least @xmath41 , then there is almost surely an infinite component .
these results are generalizations of analogue results for the one - color case . for the one - color case
there are strong indications that the case with constant degree 2 does generate an infinite component : in @xcite it is shown that almost sure existence of an infinite component follows from the assumption that a certain finite event has large enough probability , and the assumption is strongly supported by computer simulations .
our main result is that the two - color model with constant degree 2 almost surely does not give rise to an infinite component .
the model is hence qualitatively different from the one - color version in this respect .
[ noinf ] almost surely , there is no infinite component in the two - color stable multi - matching with @xmath42 .
the rest of the paper is organized as follows .
proposition [ unique ] is proved in section 2 . in section 3 ,
we establish a consequence of percolation , more specifically , we show that almost sure existence of an infinite component in the two - color model with constant degree 2 implies that the number @xmath43 of vertices that would prefer to be matched to a point at the origin ( in a sense defined in section 2 ) is finite almost surely . then , in section 4 , we show that in fact @xmath44 almost surely for any two degree distributions in the two - color model , which proves theorem [ noinf ] .
finally , in section 5 , simulation results for other degree distributions are presented along with some suggestions for further work .
proposition [ unique ] is proved along the same lines as the analog one - color result .
we say that two edges @xmath45 and @xmath46 in @xmath31 $ ] * cross * each other if @xmath47 .
proposition [ unique ] follows by combining the fact that two edges that cross each other in the two - color stable multi - matching on @xmath0 must belong to the same component ( lemma [ unbo ] ) with the fact that an infinite component must be unbounded in both directions in any translation - invariant matching scheme ( lemma [ cross ] ) .
the latter follows from a version of the mass transport principle , which will also be used later .
a * mass transport * is a random measure @xmath48 on @xmath49 that is invariant in law under translations of @xmath15 . for borel sets @xmath50 ,
we interpret @xmath51 as the amount of mass transported from @xmath52 to @xmath53 .
write @xmath54 for the unit cube @xmath55 .
[ masst ] let @xmath48 be a mass transport . then @xmath56 for a proof , see e.g. @xcite[lemma 2.2 ] .
we say that an infinite component in a translation - invariant two - color multi - matching scheme is unbounded to the right ( left ) with respect to the red ( blue ) points if , for any @xmath57 , it contains infinitely many red ( blue ) points to the right ( left ) of @xmath58 ( @xmath59 ) .
[ unbo ] an infinite component in a translation - invariant two - color multi - matching scheme is almost surely unbounded in both directions with respect to both colors . just observe that a matching scheme that with positive probability is bounded in one direction with respect to ( at least ) one of the colors yields a contradiction with the mass transport principle : assume for instance that an infinite component has a rightmost red point @xmath25 .
for the mass transport where each point in the component sends out mass 1 to @xmath25 , we then have that the unit interval receives infinite mass with positive probability , while the expected mass that is sent out is bounded by @xmath60 .
next we show that crossing edges in the two - color stable multi - matching can not belong to different components . for the one - color stable
multi - matching this follows immediately from the definition of the matching , but here it requires a proof . for a point @xmath25 in
@xmath10 $ ] or @xmath11 $ ] , write @xmath61 for the length of the longest edge incident to @xmath25 and say that @xmath25 * desires * a point @xmath62 if @xmath63 . by the definition of the two - color stable multi - matching ,
if @xmath64 $ ] and @xmath65 $ ] are such that @xmath25 desires @xmath26 and @xmath26 desires @xmath25 , then @xmath25 and @xmath26 must have an edge between them .
[ cross ] for the two - color stable multi - matching with any two degree distributions , crossings edges belong to the same component .
suppose that the edges @xmath45 and @xmath46 in @xmath31 $ ] cross each other .
by symmetry we only need to consider two cases : assume that @xmath66 and @xmath67 are blue , and that @xmath68 and @xmath69 are hence red .
since @xmath70 and the edge @xmath45 is in @xmath31 $ ] , we have that @xmath68 desires @xmath67 .
similarly , since @xmath71 and the edge @xmath46 is in @xmath31 $ ] , we also have that @xmath67 desires @xmath68 .
it follows from stability of the matching that @xmath72 is in @xmath31 $ ] .
assume that @xmath66 is blue and @xmath67 is red , so that hence @xmath68 is red and @xmath69 is blue . since @xmath73 and @xmath45 is in @xmath31 $ ] , we have that @xmath66 desires @xmath67 .
analogously , we obtain that @xmath69 desires @xmath68 .
if @xmath74 , then @xmath68 desires @xmath69 , and the edge @xmath75 is in @xmath31 $ ] .
if @xmath76 , we have that @xmath77 , so @xmath67 desires @xmath66 , and the edge @xmath78 is in @xmath31 $ ] .
uniqueness of the infinite component in the two - color stable multi - matching now follows from lemma [ unbo ] and lemma [ cross ] by noting that two components that are both unbounded in both directions can not avoid having crossing edges .
for the two - color stable multi - matching , let @xmath43 denote the number of blue points that desire the origin . in this section we show that , if there is almost surely an infinite component in the case when all degrees are almost surely equal to 2 , then @xmath79 almost surely .
we then show in section 4 that in fact @xmath44 almost surely for any degree distributions , which gives theorem [ noinf ] . for the one - color model
, it was shown in @xcite that existence of an infinite component is equivalent to @xmath79 almost surely .
that model is strongly believed to percolate , but proving that @xmath79 almost surely has so far not been sucessful .
[ equi ] consider the two - color stable multi - matching with @xmath80 and @xmath42 .
if almost surely there is an infinite component , then @xmath79 almost surely . in order to prove proposition [ equi ] , we need to get a grip on the structure of the components in the two - color stable multi - matching with constant degree 2 .
this is more involved than in the one - color model , but it turns out that an infinite component must have a relatively simple structure ; see figure 1 .
the following lemmas are what we need .
[ samba ] consider the two - color stable multi - matching with @xmath42 .
for a red point @xmath66 and a blue point @xmath68 , suppose that @xmath81 and that @xmath82 $ ] .
let @xmath83 be a blue point such that @xmath84 $ ] .
if @xmath67 has a partner @xmath69 to the right of @xmath68 , then @xmath85 $ ] . furthermore
, the point @xmath67 can have at most one partner to the right of @xmath68 , and no partners to the left of @xmath66 .
as for the first claim , note that , since @xmath86 , it follows that @xmath69 desires @xmath68 , and since the edge @xmath84 $ ] , we must have that @xmath87 .
suppose for contradiction that @xmath68 does not desire @xmath69 .
then @xmath76 , which implies that @xmath88 , that is , @xmath68 desires @xmath69 . as for the second claim ,
suppose that @xmath67 has two partners , say @xmath69 and @xmath89 , to the right of @xmath68 .
the same argument as above shows that @xmath68 desires both @xmath69 and @xmath89 .
since @xmath69 and @xmath89 also desire @xmath68 , the edges @xmath75 and @xmath90 must be in @xmath91 and hence @xmath68 has degree at least 3 , a contradiction .
finally , suppose that @xmath67 has a partner , say @xmath92 , to the left of @xmath66 .
since @xmath93 and @xmath94 , it follows that @xmath66 and @xmath67 desire each other . hence @xmath95 $ ] .
[ mia ] consider the two - color stable multi - matching with @xmath42 .
suppose that @xmath45 and @xmath46 in @xmath31 $ ] are in the infinite component and that they cross each other , that is , @xmath47
. then @xmath96 .
in particular , @xmath67 and @xmath68 must be of different colors .
first note that , when @xmath67 and @xmath68 are of different colors , it is clear that @xmath67 desires @xmath68 and vice versa , implying that they must be matched in @xmath28 .
we are hence done if we can show that it is impossible for @xmath67 and @xmath68 to have the same color . to this end , assume that @xmath67 and @xmath68 are , say , blue , and note that both edges @xmath78 and @xmath75 can not be in @xmath31 $ ] , since then the vertices @xmath97 would form a finite component of their own .
so assume that @xmath98 $ ] .
then , by lemma [ samba ] , we must have @xmath99 $ ] . also by lemma [ samba ]
, the second partner of @xmath67 ( that is , not @xmath69 ) can not be to the right of @xmath68 or to the left of @xmath66 .
hence it must be inside @xmath45 .
write @xmath100 for this second partner of @xmath67 .
note that @xmath100 , in turn , can not have its second partner @xmath101 outside of @xmath45 .
indeed , @xmath101 can not be to the left of @xmath66 , since then , by lemma [ mia ] , the edge @xmath102 would be in @xmath31 $ ] , which would imply a finite component .
furthermore , @xmath101 can not be to the left of @xmath68 , since that would mean that @xmath68 and @xmath101 desire each other and would hence be connected , which would give @xmath68 degree 3 .
iterating this argument leads to infinitely many points inside @xmath45 , which is impossible .
the case when @xmath95 $ ] and @xmath103 $ ] is ruled out analogously .
call a point at @xmath25 a * bird * if it has one partner on each side of @xmath25 , a * left - beak * if it has both its partners to the left of @xmath25 and a * right - beak * if it has both its partners to the right of @xmath25 .
lemma [ mia ] implies that , immediately to the right of a right - beak in the infinite component , there must be a left - beak of the opposite color . in between such pairs of consecutive right - beaks / left - beaks
, there may be birds ; see figure 1 .
the following observations are consequences of this structure .
[ comp ] consider the two - color stable multi - matching with @xmath42 and let @xmath104 be the set of points in the infinite component .
any point in @xmath105 $ ] is almost surely connected no further away from itself in a given direction than to its second nearest neighbor in @xmath106 $ ] in that direction . between two consecutive points in
@xmath105 $ ] , there are at most three points in @xmath106 $ ] . as for ( i ) , assume that a red point @xmath66 in the infinite component is connected to a blue point @xmath107 and that the interval @xmath45 contains at least two other blue points in the infinite component .
first note that the path from @xmath66 in the direction away from @xmath68 can not re - enter the interval @xmath45 after having left it , since that would produce a configuration that violates lemma [ mia ] , and analogously for the path from @xmath68 in the direction away from @xmath66 .
hence each point in @xmath108 must be hit by precisely one of these paths before the path leaves @xmath45 ( which both paths eventually do , since there are only finitely many points in @xmath45 ) .
but if there are at least two blue points in the interval , at least one of the paths would generate a contradiction with lemma [ mia ] when leaving the interval : at most one of the blue points in @xmath45 can be a partner of @xmath66 and , immediately after hitting such a point , the path from @xmath66 must leave @xmath45 in order not to violate lemma [ mia ] when it does so . the rest of the blue points in @xmath45 must hence be on the path from @xmath68 , and it is not hard to see that this path then yields a contradiction with lemma [ mia ] when it leaves @xmath45 ( if not sooner ) . as for ( ii ) , note that , by lemma [ mia ] , a red left - beak in the infinite component must be followed immediately to the left by a blue right - beak and , similarly , a red right - beak must have a blue left - beak immediately to its right . in between a red left - beak and a red right - beak
there can be a red bird , but no further red point without blue points in between .
the maximum possible number of consecutive red points in the infinite component is hence three .
the next lemma is the last ingredient that we need to prove proposition [ equi ] .
it is a generalization of @xcite[lemma 2.4 ] .
[ plop ] let @xmath109 and @xmath12 be two simple point processes on @xmath0 with finite intensities , jointly ergodic under translations , and such that , in between two consecutive points from one of the processes , there are at most @xmath41 points from the other process
. for @xmath110 $ ] , write @xmath111 for the maximum of the distances from @xmath25 to the second nearest point of @xmath13 $ ] on the left and the second nearest point of @xmath13 $ ] on the right .
the number of points @xmath110 $ ] with @xmath112 is finite almost surely .
consider the mass transport in which , for each @xmath109-point @xmath25 , each of the intervals @xmath113 and @xmath114 between @xmath25 and the second nearest @xmath12-points @xmath26 and @xmath115 to the right and left , respectively , sends out mass equal to @xmath116 times its length .
note that any interval on @xmath0 is covered by at most @xmath117 such intervals
@xmath118 coming from each direction and hence the total mass sent out by any interval is dominated by its length .
the mass is distributed uniformly over the intervals @xmath119 and @xmath120 , respectively .
if there were infinitely many @xmath121 $ ] with @xmath122 , then the unit - interval would receive infinite mass .
this contradicts the mass transport principle , since the mass sent out by the unit - interval is dominated by 1 . combining lemma [ comp ] and lemma [ plop ] with @xmath12 and @xmath109 in lemma [ plop ] corresponding to red and blue points , respectively yields that the number of blue points in the infinite component that desire the origin is finite almost surely .
as for the finite components , it follows from lemma [ cross ] that edges of such components can not cross edges of the infinite component .
hence a finite component must be contained between two consecutive points in the infinite component .
consider the longest edge in the infinite component that covers a given finite component and note that , if the finite component contains vertices that desire the origin , then at least one end - point of this covering edge must also desire the origin . furthermore , clearly each covering edge can enclose only finitely many finite components . hence ,
if with positive probability there are infinitely many finite components that contain blue vertices that desire the origin , then with positive probability there are infinitely many vertices in the infinite component that desire the origin . by symmetry , this means that there is a positive probability of having infinitely many blue points in the infinite component desiring the origin , which is a contradiction .
in this section we show that @xmath44 almost surely for the two - color stable multi - matching with any degree distribution(s ) ( satisfying ( [ des ] ) ) . together with proposition [ equi ] , this proves theorem [ noinf ] .
[ viola ] in the two - color stable multi - matching , for any degree distributions , we have @xmath44 almost surely . by ergodicity ,
the event that there exists an infinite component has probability 0 or 1 . combining proposition [ equi ] and proposition [ viola ] rules out the latter option . to prove proposition [ viola ] , we need a multi - matching analogue of the monotonicity property proved in @xcite[lemma 17 ] for the two - color matching , that is , for the case with @xmath123 . more specifically , the following lemma states that adding an extra blue point makes the multi - matching no worse for red points .
[ beauty ] let @xmath124 be discrete disjoint sets , such that @xmath125 is non - equidistant and has no descending chains . for any given degree distribution(s ) , let @xmath126 be the partial two - color stable multi - matching between @xmath127 and @xmath128 , and @xmath129 the partial two - color stable multi - matching between @xmath130 and @xmath128 .
furthermore , for @xmath131 , write @xmath132 and @xmath133 for the @xmath41:th nearest partner of @xmath25 in @xmath126 and @xmath129 , respectively , with @xmath134 ( @xmath135 ) if @xmath25 has fewer than @xmath41 partners in @xmath126 ( @xmath129 )
. then @xmath136 assume for contradiction that there is a @xmath137 such that @xmath138 for some @xmath41 .
this means that @xmath139 has at least one partner in @xmath126 and that at least one of these @xmath126-partners is not connected to @xmath139 in @xmath129 .
let @xmath140 be the closest one of the @xmath126-partners that @xmath139 is no longer connected to in @xmath129 .
then @xmath139 desires @xmath140 in @xmath129 , since indeed all edges of @xmath139 in @xmath129 can not be shorter than the distance to @xmath140 .
since @xmath139 and @xmath140 are not connected in @xmath129 , it follows that @xmath140 does not desire @xmath139 in @xmath129 , that is , all edges of @xmath140 in @xmath129 are shorter than the distance to @xmath139 .
furthermore , @xmath140 has at least one partner in @xmath129 that @xmath140 is not connected to in @xmath126 ( since @xmath140 is connected to @xmath139 in @xmath126 but not in @xmath129 ) .
let @xmath141 be the closest @xmath129-partner of @xmath140 that @xmath140 is not connected to in @xmath126
. then @xmath142 . since @xmath139 is a partner of @xmath140 in @xmath126 , we have that @xmath140 desires @xmath141 in @xmath126 and , since @xmath140 and @xmath141
are not connected in @xmath126 , it follows that @xmath141 does not desire @xmath140 in @xmath126 .
hence all , edges of @xmath141 in @xmath126 are shorter than the distance to @xmath140 .
furthermore , @xmath141 has at least one partner in @xmath126 that @xmath141 is not connected to in @xmath129 ( since @xmath141 is connected to @xmath140 in @xmath129 but not in @xmath126 ) .
let @xmath143 be the closest @xmath126-partner of @xmath141 that @xmath141 is not connected to in @xmath129
. then @xmath144 . iterating the above
gives a sequence @xmath145 , such that @xmath146 .
note that we must have @xmath147 for @xmath148 , since @xmath149 has an edge to @xmath150 in @xmath129 , while all edges of @xmath151 in @xmath129 are shorter than the distance to @xmath152 , where @xmath153 .
similarly @xmath154 for @xmath148 .
hence @xmath145 constitute a descending chain and we have arrived at a contradiction .
we have to show that the number of blue points that desire the origin is infinite almost surely .
to this end , define @xmath155 : m_{x}>|x|-1 \}$ ] , that is , @xmath156 is the set of blue points that desire some point in the interval @xmath157 , and write @xmath158 for the cardinality of @xmath156 . the same argument as in @xcite[theorem 1.3(i ) ] applies to show that @xmath159 a.s .
implies that @xmath44 a.s .
hence we are done if we show that @xmath159 a.s .
the proof of this is analogue to the proof of @xcite[theorem 6(i ) ] for two - color stable matchings : fix any @xmath160 ; we will prove that @xmath161 . to this end , let @xmath162 be obtained from @xmath27 by adding @xmath41 independent uniformly random red points in @xmath157 . by a straightforward adaptation of @xcite[lemma 18(i ) ] , and independence between @xmath27 and @xmath163 , we have that the law of @xmath164 is absolutely continuous with respect to the law of @xmath165 . by @xcite[proposition 1 ] , almost surely all @xmath41 added red points are matched in the two - color stable multi - matching @xmath129 between @xmath162 and @xmath163 .
let @xmath126 be the two - color stable multi - matching between @xmath27 and @xmath166 .
by lemma [ beauty ] , the blue partners of the @xmath41 added red points were matched at least as far away in @xmath126 , so these blue partners lie in @xmath156 .
thus @xmath167 , as required .
as we have shown , there is no infinite component in the two - color stable multi - matching with constant degree 2 . is this the case also for other degree distributions ?
indeed , we know that @xmath44 for all degree distributions . however , the method employed to show that existence of an infinite component implies that @xmath79 relies heavily on the fact that the structure of an infinite component can be characterized in the degree 2 case .
this does not seem doable for other degree distributions , indicating that another approach is probably neccessary in the general case .
the result that @xmath44 if all components are finite almost surely , proved for the one - color model in @xcite , seems possible to extend to the two - color model , but does not seem to help in answering the above question .
table 1 contains simulation results for some other degree distributions .
( we have chosen to generate a fixed number of points of each color uniformly distributed on a finite cycle , which means that the points do not quite consitute two independent poisson processes . fixing the total number of points and
then randomizing their colours would give a finite representation of our process where the points indeed constitute independent poisson processes , but it would mean that all stubs are not matched and thus lead to difficulties in defining the largest connected component . ) as for deterministic degrees , the relative size of the largest connected component seems to go to 0 for degree 3 , suggesting that there is no infinite component . for constant degree 4 , there is also a clearly decreasing trend , while for constant degree 5 it is difficult to draw any conclusions
the relative size of the largest component is still large in the largest simulation , but the standard deviation is also large compared to the smaller simulations .
is it the case that there is never percolation for deterministic degrees in @xmath168 ?
if so , does this remain true also when there is randomness in the degrees ? indeed , for a mixture of degrees 2 and 3 and of degrees 3 and 4 , and for poisson(2)+1 degrees , the relative size of the largest component seems to decrease , while for poisson(2)+2 degrees we can not see this in simulations of the current size .
another question concerns the so called random direction stable multi - matching , which is obtained by assigning random directions to the stubs at the poisson points and then do a stable matching of the stubs subject to the restriction that a given stub is to be matched in its prescribed direction . in @xcite , it is shown that the one - color version of this model with constant degree 2 almost surely does not generate an infinite component .
is this true also in the two - color case ?
the two - color version of the model seems more difficult to analyze , since the component structure becomes more involved . | let red and blue points be distributed on @xmath0 according to two independent poisson processes @xmath1 and @xmath2 and let each red ( blue ) point independently be equipped with a random number of half - edges according to a probability distribution @xmath3 ( @xmath4 ) .
we consider translation - invariant bipartite random graphs with vertex classes defined by the point sets of @xmath1 and @xmath2 , respectively , generated by a scheme based on the gale - shapley stable marriage for perfectly matching the half - edges .
our main result is that , when all vertices have degree 2 almost surely , then the resulting graph does not contain an infinite component .
the two - color model is hence qualitatively different from the one - color model , where deijfen , holroyd and peres have given strong evidence that there is an infinite component .
we also present simulation results for other degree distributions . _ keywords : _ poisson process , random graph , degree distribution , percolation , matching .
ams 2000 subject classification : 60d05 , 05c70 , 05c80 . |
let @xmath0 , @xmath1 , @xmath2 , be observations from a doubly - indexed sequence @xmath3 of independent random real variables .
suppose that each @xmath4 has the pdf @xmath5 from a one - parameter family of pdfs @xmath34 . in this note
, we make * compound * maximum likelihood estimations @xmath35 of @xmath36 , subject only to a non - decreasing constraint : @xmath37 compound estimates @xmath38 that meet this constraint will be called * non - decreasing*. think of each value of the index @xmath13 specifiying consecutive levels of an explanatory variable , and @xmath39 is the response to that @xmath13-th level of the explanatory variable .
the goal of this note is to specify the maximum likelihood non - decreasing response estimate ( existence and uniqueness ) and provide an algorithm for its efficient computation .
of course , the same technology gives also the maximum likelihood non - increasing response function .
one simply reverses the ordering of the levels of the explanatory variables , and applies the same theory and algorithm to the reversely ordered data .
let @xmath0 , @xmath1 , @xmath2 , be observations from a doubly - indexed sequence @xmath3 of independent random real variables .
suppose that each @xmath4 has the pdf @xmath5 from a one - parameter family @xmath40 of pdfs @xmath34 .
set @xmath41 a real number @xmath42 is called * observable * if and only if @xmath43 .
* assumption 1 .
* one of two cases are assumed : 1 .
the pdfs in @xmath40 are for discrete real random variables .
2 . the pdfs in @xmath40 are for absolutely continuous real random variables .
* assumption 2 .
* we assume that @xmath44 and @xmath45 is a real interval of positive length .
consequently , @xmath45 includes the arithmetic means of finite sequences of observable real numbers .
* assumption 3 .
* let @xmath46 , @xmath47 be observations from the independent random variables @xmath48 , @xmath47 , with pdfs in the given family .
as in @xcite and on pages 337 341 of @xcite , the likelihood function is @xmath49 where @xmath50 .
we assume that , if each @xmath51 is observable and @xmath52 is the arithmetic mean of @xmath53 , @xmath54 , @xmath55 , then @xmath56 is strictly increasing for @xmath57 and strictly decreasing for @xmath58 .
the next lemma is an immediate consequence of these assumptions .
it will be applied frequently to subsets of the observations under discussion .
[ l : begin ] let @xmath59 , @xmath47 , be observations from the independent random variables @xmath48 , @xmath47 , with pdfs in the given family
. then @xmath60 and , for all @xmath61 such that @xmath62 @xmath63 also , @xmath64 by assumption 2 , @xmath60 . equation [ e : eqn0 ] follows directly from assumption 3 .
since @xmath45 is a real interval of positive length , there is some @xmath62 in @xmath45 .
therefore @xmath65 the formal context applies to many of the common one - parameter distributions , if one parameterizes them by their means and takes care to include some boundary distributions . here
are some examples , verified in an appendix : 1 .
bernoulli random variables with parameter @xmath66 $ ] .
note that we include the boundary cases @xmath67 and @xmath68 because they could be sample means .
this follows the usual practice for maximum likelihood estimators ( see page 318 of @xcite ) .
2 . poisson random variables parameterized by their means , with @xmath69 ( see page 644 of @xcite ) .
note the inclusion of the boundary value @xmath67 .
3 . let @xmath70 and @xmath71 the family of geometric random variables ( here including one constant random variable ) , parameterized by their means ( see page 644 of @xcite ) .
given @xmath72 , set @xmath73 . for @xmath74 , the pdf with parameter @xmath75
is defined as follows : @xmath76 for @xmath77 , @xmath78 for all @xmath42 except that @xmath79 .
4 . normal distributions @xmath80 with @xmath81 fixed and @xmath82 .
it is noted on page 317 of @xcite that @xmath56 has a unique maxium at @xmath83 . to verify the slightly stronger assumptions of this note ,
one slightly modifies the argument given in @xcite .
exponential random variables parameterized by their means ( page 638 of @xcite ) .
let @xmath84 . given @xmath72 , the pdf with parameter @xmath75
is defined as follows : @xmath85 the motivating example of the next section has binomial distributions .
the example is an application of a corresponding bernoulli random variable case .
the compound likelihood functions of the two situations differ by a non - zero factor @xmath86 that does not depend on any of the parameters @xmath87 .
@xmath86 has the form @xmath88 where @xmath89 is the number of successes ( @xmath68s ) in the bernoulli observations @xmath90 ,
@xmath54 , @xmath91 . within this formal context , and with each @xmath0
observable , it will be proved that there is a unique non - decreasing compound estimate @xmath7 of @xmath8 such that , among all non - decreasing @xmath92 , @xmath15 where 1 .
@xmath16 is the ( doubly - indexed ) vector of observations 2 .
@xmath17 is the compound likelihood function : @xmath18 there is an efficient algorithm to compute @xmath19 .
the notation @xmath20 $ ] denotes the integers in the @xmath21 $ ] . for @xmath22
$ ] let @xmath23 that is , @xmath24 is the sample mean of observations @xmath0 with @xmath25
. let @xmath26 be a complete list of the distinct components of @xmath12 .
let @xmath27}{\phi_i=\tau_r}$ ] . by a theorem
, there are integers @xmath28 such that @xmath29 $ ] . also , @xmath30 . for integers
@xmath31 $ ] , set @xmath32}{k \in [ a_r\ , \hdots \,m ] } \label{e : kappa}\ ] ] it will be proved that @xmath33}{\mu[\,a_r\ , \hdots\ , k]=\kappa_r } \label{e : br}\ ] ] here is the algorithm : 1
. set @xmath93 equal to @xmath68 .
compute @xmath94 according to equation [ e : kappa ] . then compute @xmath95 according to equation [ e : br ] . for @xmath96 $ ] ,
set @xmath97 equal to @xmath98 $ ] .
2 . proceed recursively .
suppose that @xmath99 and @xmath100 satisfy equation @xmath101 with @xmath102 , and @xmath103 for @xmath104 . if @xmath105 , the algorithm stops .
if @xmath106 , set @xmath107 . then compute @xmath108 according to equation @xmath109 .
then compute @xmath110 according to equation [ e : br ] . for @xmath111
$ ] , set @xmath97 equal to @xmath112 $ ] .
for a particular mathematics course , can one use various sat scores to predict performance on the final examination ?
a significant subset of the students failed to take the final examination , in some cases because they withdrew from the course ( with a record of w ) or they simply received a letter grade of f. in the data appendix , table [ tab : data ] tabulates the no - show counts by sat - r levels .
the data are noisy .
table [ tab : noshownd ] gives the maximum likelihood non - decreasing response estimate . in figure
[ fig : noshowvsrsat ] are plotted both the observed no - show rates for each sat - r score , and the maximum - likelihood non - decreasing response estimate . on the other hand ,
the maximum likelihood * non - increasing * response estimate is the constant @xmath113 for every sat - r level . to estimate the significance of the sharp difference between the maximum - likelihood non - decreasing and non - increasing response estimates , 10,000 data tables with the same structure as table [ tab : data ]
were generated under a null hypothesis of a constant no - show rate of @xmath113 regardless of sat - r score .
* for each simulated table , the maximum - likelihood non - decreasing response estimate @xmath114 and the maximum - likelihood non - increasing response estimate @xmath115 were computed .
this test statistic was tabulated : @xmath116 - [ g(330)-g(800)]\ ] ] * the simulated quantile ranking is 0.9899 for the same statistic for the actual table ( namely @xmath68 ) .
that is , 101 of the simulated tables had @xmath117 ( the maximum logically possible ) .
there are easier and more routine ways to reject the null hypothesis , but the non - decreasing response estimate itself has proved useful for directing resources to students who have a greater risk of failure .
the observed table is broadly consistent with an alternative hypothesis of the maximum - likelihood non - decreasing response estimate ; its simulated loglikelihood rank ( based on 10,000 simulated tables generated under this alternative hypothesis ) is 0.3476 .
cccc & & & + 330 - 430 & 15 & 0 & 0% + 440 - 530 & 69 & 9 & 13.0% + 540 - 580 & 35 & 5 & 14.3% + 590 & 5 & 1 & 20% + 600 & 3 & 1 & 33.3% + 610 - 660 & 19 & 7 & 36.8% + 680 - 750 & 5 & 2 & 40% + 800 & 1 & 1 & 100% + + & 152 & 26 & 17.1%
every @xmath118 defines a partition @xmath119 of @xmath120 $ ] as follows .
let @xmath121 , @xmath122 , @xmath123 , @xmath124 be a listing of the distinct components of @xmath75 without the repetition of any component .
let @xmath125 } { \theta_i = \tau_t } \ ] ] let @xmath126 , the index of @xmath75 , be equal to the number of @xmath127 $ ] such that @xmath128 [ l : step0 ] let @xmath129 have each @xmath130 .
suppose that @xmath118 is non - decreasing and has @xmath131 .
then there is some non - decreasing @xmath132 such that 1 .
@xmath133 has fewer members than @xmath119 or , @xmath134 and @xmath135 .
2 . for the compound likelihood function @xmath136 , @xmath137 suppose first that @xmath138 . let @xmath139 $ ] , the arithmetic mean of all the observations . by lemma [ l : begin ] , because every @xmath0 is observable , we have @xmath140 .
let @xmath141 be the constant vector with components equal to @xmath142 .
clearly @xmath143 is non - decreasing , @xmath144\}$ ] , and for all @xmath13 we have @xmath145\ ] ] thus @xmath146 .
by lemma [ l : begin ] , @xmath147 for the rest of the proof we assume that @xmath148 .
let @xmath149 be a complete list of the distinct components of @xmath75
. set @xmath150}{\theta_i=\tau_t}\ ] ] then @xmath151 . because @xmath131 , there is at least one @xmath152 $ ] such that @xmath153 . because @xmath75 is non - decreasing , there are integers @xmath154 in @xmath120 $ ] such that @xmath155 $ ] .
let @xmath156 be the vector with @xmath157 components @xmath158 .
then , @xmath159 where @xmath160 and @xmath161\backslash\{t\ } } \prod_{i \in a_t } \prod_{j=1}^{n_i } f(x_{i , j } \mid \theta_i)\ ] ] since @xmath148 and @xmath162 , we have @xmath163 . for @xmath127\backslash \{t\}$ ] and for all @xmath164 , we set @xmath165 for @xmath166 , we will select some @xmath167 and for all @xmath168 we will set @xmath169 . note that we will then have @xmath170 we break the selection of @xmath167 into 6 cases : 1 .
@xmath171 if @xmath172 .
2 . @xmath171 if @xmath173 , @xmath174 and @xmath175 .
3 . @xmath176 if @xmath173 , @xmath177 , and @xmath178 .
4 . @xmath179 if @xmath173 , @xmath180 and @xmath181 .
@xmath182 if @xmath173 , @xmath183 , and @xmath184
@xmath176 if @xmath173 , @xmath185 , and @xmath186 note that in every case , @xmath187 and thus @xmath188 : * in cases 1 , 2 , 3 and 6 , @xmath167 is the arithmetic mean of observable real numbers . by lemma
[ l : begin ] , @xmath167 is in @xmath45 . * in cases 4 and 5 , @xmath167 is equal to a component of @xmath75 . since @xmath118 , we have @xmath167 in @xmath45 . next , please note that in every case @xmath167 has been carefully selected to make @xmath143 be non - decreasing .
it will be argued shortly that the change from @xmath189 to @xmath190 for @xmath168 strictly increases @xmath191 . since @xmath163 , that will give us @xmath192 * in cases 1 , 2 , 3 and 6 ,
the change is from @xmath153 to @xmath193 . by lemma [ l : begin ] , @xmath194 * in case 4 , we have @xmath195 . by assumption 3
, @xmath196 is strictly increasing for @xmath197 .
therefore @xmath198 in case 4 , @xmath179 and thus @xmath199 * in case 5 , we have @xmath200 . by assumption 3
, @xmath196 is strictly decreasing for @xmath201 .
therefore , @xmath202 in case 5 , @xmath182 and thus @xmath203 lastly , consider item ( i ) of the lemma . * in case 1 , @xmath204 .
also @xmath133 has one member .
if @xmath119 has more than one member , item ( i ) is satisfied . if @xmath119 has one member , then @xmath205\}$ ] . since @xmath131 by hypothesis
, we have @xmath135 .
that , too , satisfies item ( i ) . * in cases 2 , 3 and 6 , @xmath134 .
also , for @xmath206 , there is no change in whether @xmath207 is equal to @xmath208 . however , for @xmath168 , by changing for @xmath209 to @xmath210 , we have made @xmath211 . * in case 4 , @xmath133 has one fewer partition member than @xmath119 , because @xmath212 where @xmath213 * in case 5 , again @xmath133 has one fewer partition member than @xmath119 , because @xmath214 where @xmath215 [ l : step1 ] let @xmath129 have each @xmath130 .
suppose that @xmath118 is non - decreasing with @xmath131 .
then there is some non - decreasing @xmath132 with @xmath146 such that @xmath137 apply lemma [ l : step0 ] to @xmath216 to get @xmath217 , and continue recursively to apply lemma [ l : step0 ] to @xmath218 to get @xmath219 , but stop if @xmath220 . in lemma
[ l : step0 ] , @xmath218 has the role of @xmath75 ( provided that @xmath221 ) and @xmath219 has the role of @xmath143 . with each recursion
@xmath222 and either @xmath223 has fewer members than @xmath224 , or @xmath225 because @xmath226 is finite , we can reduce the size of @xmath224 at most finitely many times .
let @xmath227 be the last time that @xmath228 is smaller than @xmath229 . then , after @xmath230 more steps of the recursion , @xmath231 .
[ t : exist ] there is a non - decreasing response estimate @xmath118 such that @xmath146 , @xmath148 and , for all non - decreasing @xmath232 , @xmath233 by lemma [ l : step1 ] , for any non - decreasing @xmath232 , there is some non - decreasing @xmath118 with @xmath146 such that @xmath233 note that , when @xmath146 , the partition @xmath119 induced by @xmath75 determines @xmath75 : @xmath234 because @xmath120 $ ] is finite , there are finitely many distinct partitions of it . with the condition @xmath146 imposed , that gives at most finitely many members of @xmath235 that can play the role of @xmath236 in lemma [ l : step1 ] . among these finitely many possibilities for @xmath143 in lemma [
l : step1 ] , choose some @xmath35 with @xmath237 being largest .
in particular , consider the @xmath232 that is constant with components equal to @xmath139 $ ] . because all @xmath0 are observable , by lemma [ l : begin ] we have @xmath238 it follows that @xmath239
[ l : splitting ] suppose that all @xmath130 .
let @xmath75 satisfy the conclusions of theorem [ t : exist ] .
let @xmath240 be a complete listing of the distinct components of @xmath75
. for @xmath31 $ ] , set @xmath241}{\theta_i=\tau_r}\ ] ] then 1 . there are integers @xmath28 in @xmath120 $ ] such that @xmath29 $ ] .
2 . for all @xmath242 $ ] , @xmath243 \le \mu(a_r)\ ] ] because @xmath75 is non - decreasing , item ( i ) is immediate . we will prove item(ii ) by contradiction .
suppose there is @xmath244 and @xmath245 $ ] such that @xmath243>\mu(a_r)\ ] ] because @xmath146 , we have @xmath246 . for some real @xmath247 to be chosen soon ,
we define @xmath143 as follows : for @xmath248 $ ] let @xmath249 $ } \end{cases}\ ] ] if @xmath250 , let @xmath251 $ ] . if @xmath252 , select @xmath253 , \tau_{r+1}\})\ ] ] please note that this selection makes @xmath143 be non - decreasing . also , @xmath61 : 1 . because every @xmath0 is an observable real number , @xmath251 $ ] puts @xmath61 by lemma 1 .
2 . because @xmath45 is a real interval and both @xmath207 and @xmath254 are in @xmath45
, @xmath255 puts @xmath61 .
therefore @xmath143 is a non - decreasing member of @xmath235
. let @xmath256 ( with empty products set equal to @xmath68 ) .
since @xmath162 and a factor of @xmath148 , we have @xmath163 . by assumption 3 , with @xmath257 $ ] , @xmath258 therefore , @xmath259 this contradicts the likelihood maximizing property of @xmath75 in theorem [ t : exist ] .
[ t : algo]suppose that all @xmath130 .
let @xmath75 satisfy the conclusions of theorem [ t : exist ] .
let @xmath240 be a complete listing of the distinct components of @xmath75
. for @xmath31 $ ] , set @xmath241}{\theta_i=\tau_r}\ ] ] then 1 . there are integers @xmath28 in @xmath120 $ ] such that @xmath29 $ ] .
2 . let @xmath260}{k \in [ a_r \hdots m]}\ ] ] and set @xmath261}{\mu[a_r \hdots k]=\kappa_r}\ ] ] then @xmath262 . item ( i ) is the same as item(i ) in lemma [
l : splitting ] , and is repeated here to establish the notation . because @xmath146 , we have @xmath30 for all @xmath244 .
suppose first there is some @xmath244 such that @xmath263 for some @xmath264 .
then @xmath265 $ ] is a convex combination with positive coefficients of @xmath266 for @xmath267 $ ] . for @xmath268
, we have @xmath269 . by the definition of @xmath270
, we have @xmath271 \ge \kappa_r\ ] ] therefore , @xmath272 .
this contradiction proves that @xmath273 for all @xmath274 $ ] .
next suppose that @xmath275 $ ] for some @xmath276 $ ] . by the definition of @xmath270 and of @xmath277 , * @xmath278>\kappa_r$ ] * when @xmath264 , @xmath279 \quad \rightarrow \quad \mu[a_r \hdots b_h ] \ge \kappa_r\ ] ] in particular , @xmath280 \ge \kappa_r$ ] .
we now show that @xmath281 \le \kappa_r$ ] . if @xmath282 this is immediate from the definitions of @xmath270 and @xmath277 .
suppose that @xmath264 .
there are positive integers @xmath283 and @xmath114 such that @xmath284=\dfrac{e \mu[a_r \hdots b_{q-1 } ] + f \mu[a_q \hdots t_r]}{e+f}\ ] ] therefore @xmath285&=\dfrac{(e+f ) \mu[a_r \hdots t_r]-e \mu[a_r \hdots b_{q-1}]}{f } \\ &
= \dfrac{(e+f ) \kappa_r - e \mu[a_r \hdots b_{q-1}]}{f}\\ & = \kappa_r + ( e / f ) \cdot \left\{\kappa_r-\mu[a_r \hdots b_{q-1 } ] \right\ } \\ & \le \kappa_r \end{aligned}\ ] ] because @xmath280 \ge \kappa_r$ ] .
next we show that @xmath286>\kappa_r$ ] .
there are positive integers @xmath283 and @xmath114 such that @xmath287=\dfrac{e \mu[a_r \hdots t_r ] + f \mu[(t_r+1 ) \hdots b_q]}{e+f}\ ] ] consequently @xmath288&=\dfrac{(e+f)\mu[a_r \hdots b_q]-e \mu[a_r \hdots t_r]}{f } \\ & = \dfrac{(e+f)\mu[a_r \hdots b_q]-e \kappa_r}{f}\\ & = \mu[a_r \hdots
b_q]+(e / f ) \cdot\left \{\mu[a_r \hdots b_q]-\kappa_r \right\}\\ & > \mu[a_r \hdots b_q ] > \kappa_r \end{aligned}\ ] ] because @xmath278>\kappa_r$ ] .
third , we argue that @xmath286>\tau_q$ ] .
we ve shown already that @xmath289 \le \kappa_r <
\mu[(t_r+1 ) \hdots b_q]\ ] ] there are positive integers @xmath283 and @xmath114 such that @xmath290=\dfrac{e \mu[a_q \hdots t_r]+f\mu[(t_r+1 ) \hdots b_q]}{e+f}\ ] ] it follows that @xmath291 + f \mu[(t_r+1 ) \hdots b_q]}{e+f } = \mu[(t_r+1 ) \hdots b_q]\ ] ] however , this contradicts lemma [ l : splitting ] ( since @xmath75 satisfies the conclusions of theorem [ t : exist ] ) .
the only possibility left for @xmath277 is to be equal to @xmath292 as desired .
suppose that all @xmath130 .
there is a unique @xmath118 that is non - decreasing and , for all non - decreasing @xmath232 , @xmath293 where @xmath17 is the compound likelihood function . by theorem [ t : exist
] , there is some @xmath75 such that satisfies the conclusions of that theorem . in particular
, @xmath75 satisfies equation [ e : eqn ] for all non - decreasing @xmath232 .
suppose that @xmath35 also satisfies equation [ e : eqn ] for all non - decreasing @xmath232 .
by lemma [ l : step1 ] , if @xmath294 , there would be some non - decreasing @xmath141 such that @xmath295
so we must have @xmath296 .
because each @xmath0 is observable , their arithmetic mean @xmath297 $ ] is in @xmath45 .
let @xmath232 which has the constant component @xmath298)$ ] . by assumption 3 , @xmath299 therefore , @xmath300 .
so @xmath35 satisfies the conclusions of theorem [ t : exist ] . note that theorem [ t : algo ] specifies @xmath301 uniquely .
first , the theorem determines @xmath302 $ ] as @xmath95 must equal @xmath303 .
once @xmath292 is determined , if @xmath106 the theorem then determines @xmath304 .
of course @xmath107 , and thus @xmath305 is specified .
however , since @xmath306 , we have for @xmath307 @xmath308 thus @xmath309 .
throughout this section , let @xmath310 be a positive integer , @xmath311 \rightarrow \mathcal d$ ] , and we think of each @xmath51 as an observation of a random variable @xmath312 , with @xmath313 independent with pdfs from the given family .
let @xmath314 $ ] and @xmath71 be the family of bernoulli random variables ( here including two constant random variables ) .
given @xmath72 , the pdf with parameter @xmath75 is defined as follows : @xmath315 here @xmath316 .
we show that assumption 3 holds .
let @xmath156 have @xmath244 ones and @xmath317 zeros .
thus @xmath318 and @xmath319 .
for any @xmath320 $ ] , @xmath321 * case 1 : * suppose @xmath322 .
so @xmath323 .
then @xmath324 for @xmath325 , this derivative is negative .
because @xmath326 is continuous in @xmath327 , the function @xmath328 is strictly decreasing on @xmath329 $ ] . since @xmath156 has all zeros , @xmath330 .
so assumption 3 holds : @xmath328 is strictly decreasing on @xmath329=[\overline y , 1]$ ] and strictly increasing ( trivially ) on @xmath331 $ ] . * case 2 : * suppose @xmath332 . then @xmath333 and @xmath334 for @xmath335 $ ] ,
this derivative is positive .
because @xmath326 is continuous in @xmath327 , the function @xmath328 is strictly increasing on @xmath329 $ ] . since @xmath156 has all ones , @xmath336 .
consequently , assumption 3 holds : @xmath328 is strictly increasing on @xmath337=[0,1]$ ] and @xmath328 is strictly decreasing ( trivially ) on @xmath338 $ ] .
* case 3 : * suppose @xmath339 and @xmath340
. then @xmath341 \\ & = \lambda^{r-1 } ( 1-\lambda)^{s-1 } ( r+s)\left[\dfrac{r}{r+s}-\lambda\right ] \\ \end{aligned}\ ] ] for @xmath342 , this derivative is positive for @xmath343 and negative for @xmath344 .
because @xmath326 is continuous in @xmath327 , it follows that @xmath328 is strictly increasing on @xmath345 $ ] and strictly decreasing on @xmath346 $ ] . since @xmath319 , assumption 3 holds .
let @xmath69 and @xmath71 be the family of poisson random variables ( here including one constant random variable ) . given @xmath72 , the pdf with parameter @xmath75 is defined as follows : @xmath347 here @xmath348 is the set of non - negative integers .
we show that assumption 3 holds .
for any @xmath349 , let @xmath350 where @xmath351 is a factor that does not depend on @xmath327 .
* suppose that @xmath330 .
then @xmath352 and hence @xmath353 note that @xmath354 is negative for all @xmath355 . thus @xmath328 is strictly decreasing on @xmath356 and strictly increasing ( trivially ) on @xmath331 $ ] .
* suppose that @xmath357 . then @xmath358 \end{aligned}\ ] ] note that @xmath354 is positive on @xmath359 and @xmath354 is negative on @xmath360 .
since @xmath361 , we have @xmath328 continuous on @xmath362 .
therefore , @xmath328 is strictly increasing on @xmath363 $ ] and @xmath328 is strictly decreasing on @xmath364 .
let @xmath70 and @xmath71 be the family of geometric random variables ( here including one constant random variable ) , parameterized by their means .
given @xmath72 , set @xmath73 . for @xmath74 , the pdf with parameter @xmath75
is defined as follows : @xmath76 for @xmath77 , and thus @xmath365 , let @xmath366 for all @xmath42 except that @xmath367 .
here @xmath348 is the set of non - negative integers . for any @xmath349 , set @xmath368 .
then for @xmath369 @xmath370 for @xmath371 , @xmath372 if @xmath156 is a vector of zeros and @xmath67 otherwise
. please note that @xmath328 is continuous on @xmath373 : * suppose @xmath330 .
then @xmath156 is a vector of zeros and @xmath372 .
for @xmath369 , we have @xmath374 clearly @xmath328 is continuous on @xmath375 ; it is also continuous at @xmath67 because @xmath376 .
* suppose @xmath357
. then @xmath156 has at least one non - zero component and thus @xmath377 . for @xmath378 , @xmath379
clearly @xmath328 is continuous on @xmath375 .
since @xmath380 , we have @xmath381 .
thus @xmath328 is continuous at @xmath67 as well .
* suppose that @xmath330 .
then @xmath156 is a vector of zeros and @xmath372 . for @xmath369 , @xmath382
since @xmath328 is continuous on @xmath383 and has a negative derivative on @xmath375 , we know that @xmath328 is strictly decreasing on @xmath383 . thus @xmath328 is strictly decreasing on @xmath384 and strictly increasing ( trivially ) on @xmath331 $ ] .
* suppose that @xmath357 . for @xmath369 , @xmath385 \\
& = ( 1+\lambda)^{-2}t p^{t-1 } ( 1-p)^{t \overline{y}-1 } \left[\overline{y}p-(1-p ) \right]\\ & = ( 1+\lambda)^{-2}t p^{t-1 } ( 1-p)^{t \overline{y}-1 } \left [ \dfrac{1+\overline{y}}{1+\lambda}-1 \right ] \\ & = ( 1+\lambda)^{-2}t p^{t-1 } ( 1-p)^{t \overline{y}-1 } \left [ \dfrac{\overline{y}-\lambda}{(1+\lambda)(1+\overline{y } ) } \right ] \end{aligned}\ ] ] for @xmath378 , all the factors in the previous line are positive except for the factor @xmath386 .
thus for @xmath387 , we have @xmath388 and for @xmath389 we have @xmath390 . since @xmath328 is continuous on @xmath373 , we have @xmath328 strictly increasing on @xmath391 $ ] and strictly decreasing on @xmath364 .
let @xmath392 and , for a fixed @xmath81 , let @xmath71 be the family of normal random variables with standard deviation @xmath393 .
given @xmath72 , the pdf with parameter @xmath75 is defined as follows : for all real @xmath42 @xmath394 here @xmath395 .
we ll argue that assumption 3 holds . for all real @xmath327 , let @xmath396 where @xmath227 is a positive factor that does not depend on @xmath327 .
then @xmath397 all factors immediately above for @xmath398 , except the last one , are positive for all @xmath327 .
so @xmath354 is positive for @xmath327 in @xmath399 and @xmath354 is negative for @xmath327 in @xmath400 . since @xmath328 is continuous on @xmath401 , we have @xmath328 is strictly increasing on @xmath402 $ ] and strictly decreasing on @xmath364 .
let @xmath84 and @xmath71 be the family of exponential random variables parameterized by their means .
given @xmath72 , set @xmath403 . in terms of @xmath404 , the pdf with parameter @xmath75 is defined as follows : @xmath405 here @xmath406 .
we ll argue that assumption 3 holds .
note that , since @xmath348 has only positive numbers , @xmath407 . for any real @xmath408 and with @xmath403 , let @xmath409 then @xmath410 it follows that @xmath411 if if @xmath412 and @xmath413 if @xmath414 . because @xmath415 is continuous on @xmath375
, we have @xmath328 strictly increasing on @xmath416 $ ] and strictly decreasing on @xmath417 .
table [ tab : data ] has the no - show counts , tabulated by sat - r scores .
99 klugman , stuart a. , panjer , harry h. , and willmot , gordon e. _ loss models , second edition _
, john wiley and sons , inc .
, hoboken , new jersey , 2004 , pages 337 341 and pages 627 645 .
casella , george and berger , roger l. _ statistical inference , second edition _ duxbury press , pacific grover , california , 2002 , pages 315 318 . | let @xmath0 , @xmath1 , @xmath2 , be observations from a doubly - indexed sequence @xmath3 of independent random variables ( all of them discrete , or all of them absolutely continuous ) .
suppose that each @xmath4 has the pdf @xmath5 from a one - parameter family of pdfs @xmath6 .
mild assumptions are described under which there is a unique compound estimate @xmath7 of @xmath8 such that 1 . for integers @xmath9 in @xmath10
$ ] , @xmath11 ( @xmath12 is non - decreasing with respect to the index @xmath13 ) .
2 . among all non - decreasing vectors of parameters @xmath14 , @xmath15 where 1 .
@xmath16 is the ( doubly - indexed ) vector of observations 2 .
@xmath17 is the compound likelihood function : @xmath18 an efficient algorithm is described to compute @xmath19 .
the notation @xmath20 $ ] denotes the integers in the real interval @xmath21 $ ] . for @xmath22
$ ] let @xmath23 that is , @xmath24 is the sample mean of observations @xmath0 with @xmath25 . here is the theorem that justifies the algorithm : 1 .
let @xmath26 be a complete list of the distinct components of @xmath12 .
let @xmath27}{\phi_i=\tau_r}$ ] .
there are integers @xmath28 such that @xmath29 $ ] .
also , @xmath30 .
2 . for integers
@xmath31 $ ] , set @xmath32}{k \in [ a_r\ , \hdots \,m]}\ ] ] then , @xmath33}{\mu[\,a_r\ , \hdots\ , k]=\kappa_r}\ ] ] |
there is strong evidence for relativistic motions in jets that emanate from active galactic nuclei ( agns ) .
the lorentz factors of blazar jets lie in the range @xmath4 @xcite . in the case of agns
there have indeed been indications from a growing body of data that the associated relativistic jets undergo the bulk of their acceleration on scales that are of the order of those probed by very - long - baseline radio interferometry . in particular , the absence of bulk - comptonization spectral signatures in blazars has been used to infer that jet lorentz factors @xmath5 are only attained on scales @xmath6 @xcite . the first theoretical clues to the necessity of relativistic motion in grb s arose from the compactness problem @xcite .
the requirements that the source be optically thin can be used to obtain direct limits on the minimal lorentz factor , @xmath7 @xcite .
recently , superluminal expansion was observed in the afterglow emission of grb 030329 @xcite .
the main source of power of agn and grb jets is the rotational energy of the central black hole @xcite and/or its accretion disk .
the naturally occurring low mass density and hence high magnetization of black - hole magnetospheres suggests that the relativistic jets originate directly from the black - hole ergosphere as poynting - dominated outflows , whereas the disk surface launches a slower , possibly non - relativistic wind that surrounds and confines the highly relativistic flow .
although , dissipative processes may directly transfer the electromagnetic energy to emitting particles , the commonly held view is that it is first converted into the bulk kinetic energy and only subsequently channeled into radiation through shocks and other dissipative waves @xcite . in the limit of ideal mhd
such conversion can be achieved only via magnetic forces and the efficiency of this mechanism is the main subject of our investigation .
since most of the acceleration takes place far away from the source the space - time is basically flat .
we also use an isentropic equation of state @xmath8 , where @xmath9const and @xmath10 .
since we are interested in the magnetic acceleration of cold flows , we make @xmath11 very small , so the gas pressure is never a dynamical factor .
this relation enables us to exclude the energy equation from the integrated system and overcome the stiffness of relativistic mhd in magnetically dominated regime .
given the condition of axisymmetry the poloidal magnetic field is fully described by the so - called magnetic flux function @xmath12 , the total magnetic flux enclosed by the axisymmetric loop circle @xmath13const .
moreover , for stationary flows there are 5 quantities that propagate unchanged along the magnetic field lines and thus are functions of @xmath14 alone . these are @xmath15 , the rest - mass energy flux per unit magnetic flux ; @xmath16 , the angular velocity of magnetic field lines ; @xmath17 , the total angular momentum flux per unit rest - mass energy flux ; @xmath18 , the total energy flux per unit rest - mass energy flux ; and @xmath11 , the entropy per particle . for cold flows ( @xmath19 , @xmath20 )
we have @xmath21 , @xmath22 , @xmath23 , and @xmath24 where @xmath25 is the lorentz factor , @xmath26 is the magnitude of the poloidal velocity , @xmath27 and @xmath28 are the magnitudes of poloidal and azimuthal magnetic field , @xmath29 is the total electric current flowing through a loop of cylindrical radius @xmath30 , @xmath31 is the ratio of the poynting flux to the matter ( kinetic plus rest - mass ) energy flux , and @xmath32 is the poynting flux per unit rest - mass energy flux .
it is easy to see that the lorentz factor @xmath25 can not exceed @xmath18 .
in the simulation we deal with winds , or unconfined flows , and jets , or flows confined with a funnel . in order to avoid complications due to numerical diffusion through the jet boundary we use solid funnels of paraboloidal shape , @xmath33 where @xmath34 and @xmath30 are the cylindrical coordinates .
the initial configuration corresponds to a non - rotating , purely poloidal magnetic field with approximately constant magnetic pressure across the funnel .
the plasma density within the funnel is set to a small value so that the outflow generated at the inlet boundary can easily sweep it away . in order to speed this process up the longitudinal component of velocity inside the funnel and at the inlet is set to @xmath35 .
in fact , we use the same type of initial and boundary conditions as described in our study of lower magnetization jets @xcite .
we constructed a grid of models with different funnel power geometry ( @xmath36 ) including unconfined wind , different initial magnetization ( @xmath37 ) , and both solid and differential rotation at the base .
figure [ sigma - ev - a1 ] shows the evolution of @xmath31 along the jet boundary for the models with @xmath38 , solid rotation , and the maximum magnetization at the base @xmath39 and 620 ( for solid rotation @xmath40 increases with r ) .
one can see that all solutions exhibit transition to the particle dominated regime ( @xmath41 ) .
prior to reaching the equipartition they are described by the same law @xmath42 and after this the evolution of @xmath31 slows down significantly .
the dependence of the equipartition radius @xmath43 on @xmath40 can be approximated by @xmath44 figure [ gamma ] confirms similar evolution inside the jets .
the normal pressure at the jet boundary is close to @xmath45 , where @xmath46 is the spherical radius . in astrophysical conditions this pressure has to be balanced by the pressure of confining medium .
figure [ gpw ] shows our results for the unconfined wind .
one can see that magnetic acceleration in this case is much less effective in agreement with previous studies @xcite .
the initial energy - to - mass flux ratio of jets in our simulations yields an upper limit on the terminal lorentz factor @xmath47 . in order to make further comparisons of our numerical models with observations we need to select suitable dimensional scales .
the key scale in the problem of magnetic acceleration is the light cylinder ( or the alfvn surface ) radius , @xmath48 .
if the jets are launched by a rapidly rotating black hole in the center of an grb or agn then @xmath49 where @xmath50 . in this estimate
we assume that the angular velocity of the magnetic field lines is half of that of a maximally rotating ( rotation parameter @xmath51 ) black hole .
according to the results shown in equation ( [ req ] ) , the jets enter the matter - dominated regime at a cylindrical radius @xmath52 the corresponding distance from the black hole is @xmath53 where @xmath54 is the jet opening half - angle .
strong shock waves can appear only if @xmath55 .
for grb jets , with @xmath56 and @xmath57 this distance is remarkably close to the radius of wolf - raye stars .
thus , in the collapsar scenario the conditions for development of strong fast shocks and associated gamma - ray emission are already satisfied when the jets break away from the progenitor star .
the actual location of shocks , however , also depends on the scale of the central engine variability which puts it to larger distances .
lower lorentz factor of agn jets imply lower magnetization parameter , @xmath58 .
combined with the black hole mass @xmath59 this gives @xmath60 cm which is similar to the size of the `` blazar zone '' inferred from the observations .
thus , the slow nature of magnetic acceleration of relativistic jets is in very good agreement with the observational constraints .
the velocity profile of jets with a solid body rotation at their base shows strong tangential discontinuity at the jet boundary which favors the photon - breading mechanism of gamma - ray emission from agn jets @xcite . on the contrary , the gradual decline of lorentz factor towards the jet boundary which we see in the models with differential rotation at the base
significantly reduces its efficiency .
this implies that gamma - gay observations can be used to determine whether the agn jets originate directly from the black hole magnetospheres or from the magnetospheres of their accretion discs . like
other researches @xcite we find that magnetic acceleration of highly magnetized unconfined winds is less ineffective .
this is a rather uncomfortable conclusion since the successful mhd model of pulsar wind nebulae @xcite requires the pulsar wind to be particle dominated at the location of its termination shock .
the two possible solutions to this problem suggested so far utilize the alternating structure of magnetic field in the wind from oblique rotators ( ` striped wind ' , @xcite ) .
firstly , the alternating magnetic field can dissipate before reaching the termination shock , e.g. @xcite , with eventual convertion of the released heat into the bulk motion energy . secondly , it can dissipate inside the shock layer of the termination shock itself@xcite . in this case
one can not directly use the standard shock equations in order to determine the downstream state .
in validating the basic features of the simplified semi - analytic solutions , our numerical results go a long way toward establishing an `` mhd acceleration and collimation paradigm '' for relativistic astrophysical jets . in particular , they demonstrate that even jets with extremely high initial magnetization can be effectively accelerated via the ideal magnetic mechanism with more than a half of the poynting flux converted into the bulk motion energy of the flows .
the highest lorentz factor reached in the simulations is @xmath61 which is well within the range deduced for grb jets .
the slow character of magnetic acceleration allows dissipationless energy transport over large distances , the property which is deduced from observations and which is rather difficult to explain in other models of jet generation . | we present numerical simulations of cold , axisymmetric , magnetically driven relativistic outflows .
the outflows are initially sub - alfvnic and poynting flux - dominated , with total to rest - mass energy flux ratio up to @xmath0 . to study the magnetic acceleration of jets we simulate flows confined within a funnel with rigid wall of prescribed shape , which we take to be @xmath1 ( in cylindrical coordinates , with @xmath2 ranging from 1 to 2 ) .
this allows us to eliminate the numerical dissipative effects induced by a free boundary with an ambient medium .
we find that in all cases they converge to a steady state characterized by a spatially extended acceleration region .
for the jet solutions the acceleration process is very efficient - on the outermost scale of the simulation more than half of the poynting flux has been converted into kinetic energy flux , and the terminal lorentz factor approached its maximum possible value ( @xmath3 ) .
the acceleration is accompanied by the collimation of magnetic field lines in excess of that dictated by the funnel shape .
the numerical solutions are generally consistent with the semi - analytic self - similar jets solutions and the spatially extended acceleration observed in some astrophysical relativistic jets . in agreement with previous studies
we also find that the acceleration is significantly less effective for wind solutions suggesting that pulsar winds may remain poynting dominated when they reach the termination shock . 2 |
it is well known that a real scalar field under the influence of certain nonlinear potentials supports long - lived , spatially - localized , time - dependent configurations known as oscillons @xcite .
even though there are no topological or nontopological conserved charges associated with such objects , they exhibit remarkably long lifetimes , far exceeding naive estimates from linear approximations to the theory @xcite . this property has prompted much interest in studying oscillons in more detail in two @xcite and three @xcite spatial dimensions .
structures with similar qualitative behavior have been observed in many different physical systems , from vibrating grains @xcite to acoustic oscillations in the sun @xcite . in the past two years , it has been shown that oscillons are not limited to real scalar field models ; they have been found in 2d abelian - higgs models @xcite , and in the standard model @xcite .
they may also play a role in cosmology @xcite , possibly having an impact on the dynamics of inflationary reheating , in symmetry - breaking transitions , and in false vacuum decay @xcite . in @xmath0 de sitter models , oscillons not only form from near - thermal initial conditions , but survive to retain approximately 50% of the total energy of the universe @xcite .
so far , most oscillon properties , such as their energy , radius , and lifetime , have been obtained numerically .
it has been shown that they have a near - constant energy and a near - periodic motion with frequency just below the mass of vacuum excitations @xcite . in spite of this progress ,
it is fair to say that a proper analytical understanding of oscillons is still lacking . in the present work
, we will derive all physically relevant properties of scalar oscillons in quartic polynomial potentials .
starting from a simple _ ansatz _ for their profile , we develop a _ nonperturbative _ analytical procedure to obtain their energy , lifetime , and minimum and maximum values of their frequency , radius , and core amplitude .
we compare our analytical results to numerical simulations obtaining excellent agreement .
we begin with the lagrangian for a @xmath1-dimensional , spherically - symmetric real scalar field in flat spacetime , @xmath2 where a dot denotes a time derivative , and @xmath3 .
inspired by the numerical solutions , which show that an oscillon is well approximated by a gaussian undergoing near - periodic motion in its amplitude and with an effective radius oscillating with small amplitude about a mean value @xcite , we approximate the oscillon solution as @xmath4 where @xmath5 is the displacement from the vacuum , @xmath6 .
this approximation works particularly well in @xmath7 and @xmath8 , although it clearly fails to reproduce the oscillon s large @xmath9 behavior , where @xmath10 $ ] .
since oscillons have been observed in a diversity of quartic polynomial potentials , we write @xmath11 as @xcite @xmath12 where the @xmath13 s are constants and @xmath14 is the vacuum energy . substituting eqs .
[ ansatz ] and [ potential ] in eq .
[ lagrangian1 ] , @xmath15,\end{aligned}\ ] ] with @xmath16 where @xmath17 and @xmath18 is the linear frequency .
( primes will also denote partial derivatives with respect to @xmath19 . ) for potentials with @xmath20 , @xmath21 is the potential energy of a nonlinear oscillator with at least one equilibrium point ( at @xmath22 ) for any value of @xmath23 . for conservative motion we can write the energy as , @xmath24 = \left(\frac{\pi}{2}\right)^{\frac{d}{2}}r^d v(a_{\rm max}),\end{aligned}\ ] ] where @xmath25 is the positive turning point ( @xmath26 0 ) of @xmath27 .
we also write the expression for the frequency @xmath28 associated with the conservative motion between the two turning points , @xmath29 where @xmath30 is the period of oscillation , @xmath31^{\frac{1}{2}}$ ] , @xmath32 , and @xmath33 is related to @xmath25 by @xmath34 .
it was noted in @xcite that the location of @xmath25 relative to the inflection points of @xmath21 is an accurate indicator of oscillon existence .
this can be understood by noting that since the inflection points separate regions of opposite curvature in @xmath21 , the particular region probed by @xmath25 indicates the degree to which the field configuration experiences the stabilizing nonlinearities of the potential @xmath11 .
following @xcite , we therefore adopt @xmath35 as an indicator of oscillon stability : increasingly negative(positive ) values lead to increased stability(instability ) . defining @xmath36 , @xmath37 in figure [ fig : energyi ] we show the ( parabolic ) level curve corresponding to @xmath38 .
since values inside ( outside ) the parabola have @xmath39 @xmath40 , it is easy to show that , for a given @xmath23 , there exists a stable region provided that @xmath41 where @xcite @xmath42^{-1}.\ ] ] as the figure indicates , when @xmath43 , there is only one value of @xmath25 located in the stable region , given by @xmath44 .
in figure [ fig : energyi ] , the continuous and dashed lines show curves of constant energy ( eq . [ energy ] ) as a function of @xmath23 and @xmath25 for a symmetric double - well potential , @xmath45 that is , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 in eq .
[ quantities were made dimensionless with @xmath51 and the primes have been suppressed . ]
we also plot the parabola @xmath52 marking the boundary between the stable region ( @xmath39 , above the curve ) and the unstable region ( @xmath53 , below the curve ) .
note that the presence of an oscillon in the unstable region does not necessarily mean that it will immediately decay ; the degree of stability is given roughly by the vertical distance of a point to the curve @xmath54 .
( continuous line ) , and @xmath55 ( dashed line ) , together with the parabolic level curve @xmath56 .
vertical dashed line locates the asymptote of @xmath54.,width=288,height=288 ] using the curve @xmath54 in fig .
[ fig : energyi ] , we can describe the trajectory modeling the onset of an oscillon from an initial configuration starting at a somewhat arbitrary point @xmath57 . on its way to an oscillon
, the configuration will radiate excess energy until reaching point @xmath58 . to determine the location of point @xmath58 , note that the onset of the oscillon phase is marked by a minimum radius .
( in what follows it will be clear that this _ must _ be true if radiation radically slows down . ) taking the minimum radius given by eq .
[ minimum - radius ] , and using that the oscillon stage should begin in the stable region , fixes the location of point b to @xmath59 and @xmath43 , as shown in figure [ fig : energyi ] .
as we shall see , this is confirmed by numerical results . for the potential of eq .
[ examplepot ] , @xmath60 and @xmath61 , in excellent agreement with the inset in figure [ fig : radiusevol ] .
since our model approximates an oscillon as having constant energy , we input these values into eq .
[ energy ] , to obtain the oscillon energy , @xmath62 : @xmath63 where the numerical value given is for @xmath8 and the potential of eq .
[ examplepot ] .
compare with numerical results @xcite , where , roughly , @xmath64 during the oscillon s lifetime .
in addition to the above estimate for @xmath62 , we can derive an absolute lower bound for the oscillon energy .
this is found by noting that the dashed line in fig .
[ fig : energyi ] shows a critical energy , denoted @xmath65 , below which the energy curves do not probe the stable region . using the @xmath54 condition
, one can eliminate @xmath23 from the expression of the energy ( eq .
[ energy ] ) , which becomes a function only of @xmath25 .
remarkably , this function has a minimum , the asymptote energy of figure [ fig : energyi ] . for the potential of eq .
[ examplepot ] , @xmath66 ( @xmath8 ) and @xmath67 ( @xmath7 ) .
and @xmath68 in the symmetric double well of eq .
[ examplepot ] ( continuous lines ) , and for an asymmetric double well [ @xmath69 with @xmath70 and @xmath71 ( dashed lines ) . the analytical predictions at the points @xmath72 and @xmath73 are indicated by arrows and dots . the insets show the minimum radius and frequency for the symmetric double well.,width=288,height=288 ] having arrived at point @xmath58 ( after radiating excess initial energy ) the oscillon proceeds leftward along its line of constant energy , @xmath74 ( traveling rightward would cause @xmath23 to tend to zero , which is unphysical ) .
when the oscillon passes point @xmath75 , its distance from @xmath54 begins to increase and its instability grows . to estimate the point of oscillon decay we note
that , at point @xmath76 , the vertical distance to the @xmath54 curve becomes infinite .
mathematically , point @xmath76 is found by taking the limit @xmath77 in the expression @xmath54 and solving for @xmath78 , the decay amplitude . for a cubic potential
this gives , @xmath79 .
for a quartic potential , @xmath80 \simeq 0.84,\ ] ] where @xmath81 was obtained above and the numerical result is for the potential of eq . [ examplepot ] in @xmath8 . to obtain the final radius @xmath82 , we evaluate eq .
[ eplateau ] at point @xmath76 : @xmath83 . solving for @xmath82 , @xmath84 where @xmath85 . in @xmath8 ,
one gets a cubic equation for @xmath82 . for the potential of eq .
[ examplepot ] , with @xmath86 and @xmath87 , we obtain , @xmath88 . in figure
[ fig : radiusevol ] , we compare numerical and analytical values for @xmath89 ( left arrows and dots ) and @xmath73 ( right dots ) for this potential ( continuous lines ) and for an asymmetric potential ( dashed lines ) .
[ the numerical simulation measures @xmath23 by fitting a gaussian to the oscillon and reading off the radius . ] since we know @xmath90 throughout the oscillon s life , we can use the potential @xmath21 to compute @xmath33 and then use eq .
[ frequency ] to obtain @xmath91 .
for example , using @xmath92 gives the decay ( or critical ) frequency , @xmath93 . in the inset of figure [ fig : radiusevol ]
we compare analytical ( dot ) and numerical values .
it should be noted that this model correctly predicts that the oscillon frequency tends to a maximum ( as observed in , for example , @xcite ) as it approaches decay .
due to the remarkable near - periodicity of an oscillon , its spectral function in frequency space displays a series of narrow peaks , with the dominant one just below the mass frequency @xcite .
only a small radiating tail penetrates the region with @xmath94 . as in ref .
@xcite , we model the power spectrum of the oscillon core amplitude @xmath95 around its dominant frequency peak with a breit - wigner profile . since the power is proportional to the square of the amplitude of a given frequency component , we write @xmath96^{-1},\ ] ] where @xmath97 is the square of the amplitude of the frequency component of @xmath95 with frequency @xmath28 , and @xmath98 is a constant to be determined .
given that the amplitude of this primary frequency peak is approximately that of the oscillon itself ( because the first peak is narrow and holds most of the power ) @xmath99 , the oscillon s core amplitude .
since the constant @xmath100 setting the width of the resonance peak is related to the ( inverse ) timescale associated with the instantaneous radiation rate of the oscillon , it is natural to let @xmath101 , where @xmath102 is the oscillon energy .
most of the ( small - amplitude ) radiation leaking from the oscillon is contained in a mode with wavelength commensurate with the size of the oscillon that is , with wavelength @xmath103 .
the frequency of this wave is therefore @xmath104 .
now consider an infinitesimally thin shell of radius @xmath23 around the oscillon with volume @xmath105 , which is filled with the outgoing radiation wave ( @xmath106 ) traveling with speed @xmath107 . within this thin shell
, the wave can be approximated as a @xmath0 plane wave of constant amplitude ; hence its energy density is @xmath108 , where @xmath109 is the amplitude of the radiation wave in the shell . since
@xmath110 is given by @xmath111 , the total radiation energy contained in the thin shell is @xmath112 .
the radiation travels outward at a speed @xmath113 , and @xmath114 .
the amount of energy lost by the oscillon per unit time , @xmath115 , is therefore @xmath116^{-1}.\ ] ] if the oscillon is long lived , @xmath117 . in this case , and writing @xmath118 , we have @xmath119 where @xmath120^{-1}.$ ] to integrate eq .
[ radequation ] , we take @xmath121 to be constant , a good approximation for an oscillon .
then , @xmath122 where @xmath123 is the asymptotic energy as @xmath124 and @xmath125 the initial energy .
the oscillon decays at energy @xmath126 in a time @xmath127 given by @xmath128 if @xmath129 , the lifetime will be approximately zero , as expected . when @xmath130 , the lifetime becomes independent of @xmath131 and tends to a maximum , @xmath132 . to evaluate this expression ,
note that @xmath133 is simply the plateau energy obtained in eq .
[ eplateau ] . for @xmath123 , being the minimum energy the oscillon may possess , it is most natural to choose @xmath134 . for the other parameters in @xmath121 , we analytically calculate the average of each quantity over the stable phase of the oscillon s life ( @xmath135 in figure [ fig : energyi ] ) . for the potential @xmath136
, we obtain @xmath137 , which is quite accurate . finally , note that eq .
[ radequation ] admits the solution , @xmath138 .
thus , if @xmath139 . in @xmath7 , one can show numerically that the lowest energy oscillon has @xmath140 .
we thus see why @xmath141 in @xmath7 : the oscillon reaches its attractor value and stops radiating .
we have analytically computed all relevant physical properties of scalar field oscillons in quartic polynomial potentials , confirming that oscillons are indeed attractors in field configuration space and that their lifetimes can be estimated by studying their main radiating mode .
our results suggest why @xmath7 oscillons seem to be stable : they reach their attractor state before decaying .
we intend to extend our approach to oscillons in models with gauge fields and with more complicated potentials .
this work was partially supported by a national science foundation grant phy-0653341 .
e. honda and m. choptuik , phys .
d * 65 * , 084037 ( 2002 ) ; g. fodor , p. forgcs , p. grandclment , and i. rcz , phys .
d * 74 * , 124003 ( 2006 ) ; v. a. koutvistsky and e. m. maslov , j. math . phys . *
47 * , 022302 ( 2006 ) . | we develop an analytical procedure to compute all relevant physical properties of scalar field oscillons in models with quartic polynomial potentials : energy , radius , frequency , core - amplitude , and lifetime .
we compare our predictions to numerical simulations of models with symmetric and asymmetric double well potentials in three spatial dimensions , obtaining excellent agreement .
we also explain why oscillons have not been seen to decay in two spatial dimensions . |
pre - planetary nebulae ( ppne ) represent a fleeting stage of stellar evolution .
these transitional objects arise between the rapid mass loss phase at the end of the asymptotic giant branch ( agb ) and the ionized planetary nebula stage .
agb stars are important because they are the galaxy s main mechanism for the replenishment of dust and gas into the interstellar medium ( ism ) , ejecting most of their main sequence mass within a few times @xmath9 years because of their high mass loss rates . by looking at the circumstellar material around ppne
, we can see an imprint of the activity of the earlier agb .
ppne are also in a stage where the geometry of the mass loss from the central star usually changes from spherically symmetric to an axially symmetric bipolar outflow , as illustrated in the hst images of afgl 2688 @xcite . in these images ,
a spectacular bipolar structure is superimposed on a set of concentric shells of gas and dust ejected near the end of the agb . in the study reported here ,
we have attempted to characterize the mass loss behavior of the progenitor agb stages of three systems that are now bipolar . using the direct detection of infrared emission from dust lost over as much as @xmath9 years during the agb ,
it is possible to study the mass loss histories of these objects .
there has been a dearth of observations of large extended shells of evolved stars out to the distances sampled in this project primarily because of the difficulty in observing in the far infrared where the cool dust has its peak emission . with the spitzer space telescope @xcite ,
the far infrared is now accessible at a resolution and sensitivity sufficient to potentially resolve the structures of these dust shells , and thereby to test theories of mass loss in agb stars .
several tantalizing cases were observed by iras , showing that large dust shells exist around some evolved stars .
for example , @xcite found that r crb shows a very large shell with a radius of 4 parsecs , and reported a dust shell with a diameter of 30 - 40 arcminutes ( @xmath10 pc ) around w hya .
however , these observations could not resolve the structure that may be present in the shells , or address the possibility of multiple shells .
if the mass loss during the agb is constant , we should expect to see a smooth envelope with column density declining as @xmath11 , where @xmath12 is the displacement from the star in the sky .
however , if the mass loss rate fluctuates as the star goes through thermal pulses during the agb , as the models suggest ( e.g. , * ? ? ?
* ) , then we should see enhanced emission in the form of multiple dust shells around our objects . for example , @xcite reported on the basis of iso data that there are large concentric shells in afgl 2688 and afgl 618 .
this result is among a small set of observations showing evidence for periodic mass loss caused by thermal pulsation as well as the time scale between thermal pulses predicted by agb evolutionary models ( e.g. , * ? ? ?
confirmation of these shells at higher resolution and greater sensitivities motivated our selection of afgl 2688 as a target for this study .
there have been no previous claims of large - scale dust emission from oh 231.8 + 4.2 ( hereafter oh 231.8 ) or iras 16342@xmath03814 . however , they are presently very young ppne with high mass loss rates of approximately @xmath13 , which suggests that the emission from dust produced during the agb may be readily visible . here
we report on the lack of very extended emission , spherically symmetric or otherwise , from afgl 2688 and oh 231.8 .
however , we do see possible evidence for a very large diffuse dust shell around iras 16342@xmath03814 with a radius of 400 arcseconds , though this extended emission might also be from galactic cirrus .
afgl 2688 ( the egg nebula ) , oh 231.8 , and iras 16342@xmath03814 were observed with the mips instrument @xcite at 70 and 160 @xmath1 m on the spitzer space telescope .
these ppne were observed along two mutually perpendicular scan paths in order to sample dust emission out to about 800 arcseconds from the central source , and to determine background levels . at 70 @xmath1 m ,
the scan paths are @xmath14 in one direction and @xmath15 in the other , while at 160 @xmath1 m , the scan paths are @xmath16 and @xmath17 .
the pixel scale is 9.2 arcseconds at 70 @xmath1 m and 16 arcseconds at 160 @xmath1 m .
however , the central sources were not directly observed in order to avoid persistence artifacts from these bright sources .
this observing strategy limits the azimuthal coverage of the sources , but should give an adequate estimate of the presense of well - defined shells or asymmetries such as large - scale bipolarity . the basic science calibrated data ( bcd )
was reduced using the software package mopex from the spitzer science center ( ssc ) .
mopex was used for rejecting outliers as well as for co - adding and mosaicking the individual bcd frames to create the final image .
further reduction was done manually to remove striping due to bright latents as described in the mips data handbook .
the correction involved finding the median value measured for every pixel from a series of bcd frames far from the central source , where we assume there to be no extended emission , then subtracting this median from each data frame before co - adding and mosaicking .
the median subtraction removes both the bright latents and the uniform component of background emission .
figure [ fig : initial_mosaic ] shows the final mosaicked images .
the most prominent features of the 70 @xmath1 m images of the bipolar targets are the diffraction spikes from the point spread function ( psf ) of the central source .
although we did not image the central star , the wings of the psf are still present and bright out to about 200 arcseconds from the source . at 160 @xmath1 m , the psf is less pronounced but we see much more background than at 70 @xmath1 m from galactic cirrus .
the galactic cirrus emission presents a problem in determining the true sky background since it covers much of our fields . for both the 70 and 160 @xmath1 m images
, we use the median surface brightness of regions of low and uniform brightness far from the central source as an estimate of the sky background .
this procedure removes a uniform background from the images , retaining possible enhanced emission from the source and galactic cirrus . the uncertainty in the background ( see below ) at 160 @xmath1 m is higher than at 70 @xmath1 m because it is more difficult to find a large patch of uniform surface brightness to estimate the sky background .
the sensitivity of our data is estimated by examining the standard deviation of a patch of uniform background far from the source both before and after the median background subtraction method described above . for afgl 2688 , before background subtraction , the mean of a patch in the eastern scan path at 70 @xmath1 m is 14.1 mjy sr@xmath4 with a standard deviation of 1.11 mjy sr@xmath4 .
some of the variance in the background is likely from the effects of bright latents from the detector pixels , which are removed by the median background subtraction described above .
after median subtraction , the mean background is at 0.6 mjy sr@xmath4 with a standard deviation of 0.92 mjy sr@xmath4 . the mean surface brightness at 160 @xmath1 m before background subtraction is 31.45 mjy sr@xmath4 with a standard deviation of 1.67 mjy sr@xmath4 .
after background subtraction , the mean is at 0.33 mjy sr@xmath4 with a standard deviation of 1.51 mjy sr@xmath4 .
the sensitivity for oh 231.8 is better than for afgl 2688 , probably because of the lower sky background levels .
the sensitivity for iras 16342@xmath03814 is comparable to that of afgl 2688 at 70 @xmath1 m , but worse at 160 @xmath1 m due to a higher sky background and greater presence of galactic cirrus .
the mean background and sensitivities for all three sources at 70 and 160 @xmath1 m are given in tables [ tab:70_sensitivity ] and [ tab:160_sensitivity ] , respectively .
since the psf pattern from the central source is so prominent at 70 @xmath1 m , we attempted to fit a model psf to our images in order to subtract the bright central source in search of extended dust emission near the source .
the _ stinytim _
software provided by the ssc was used to generate model psfs .
we chose to use the default mips throughput curve along with a blackbody spectrum at several temperatures ranging from 20 to 100 k. see figure [ fig : model_psf ] for examples of the model psfs .
multidimensional fitting was carried out for our images in order to minimize the residuals of the psf subtraction using various x - y shifts as well as the scaling of the psf flux values .
the best fit was found by minimizing the square root of the sum of the squares of the differences in flux .
we found several limitations in the model psfs when attempting to subtract the psf from the images .
even for the best - fit psfs , the subtraction still leaves some obvious psf residuals at a level of @xmath18 relative to the peak as extrapolated from the model psf .
these residuals show that the wings of the model psf are not well characterized at that flux level .
the psf subtraction also tends to leave portions of the region within the airy ring with negative flux values .
the most likely reason is a non - linear pixel response as the pixels approach saturation so that although the model has a higher brightness value closer to the core of the psf , the actual pixel values are leveling off for our bright sources .
unfortunately , there are presently no empirical data on the mips 70 or 160 @xmath1 m psf to compare to the model psf at distances greater than 100 arseconds from the peak .
beyond about 240 arseconds , psf features are no longer present in the mips images so our ability to detect extended emission there is only limited by background and integration time .
the results of psf subtraction for afgl 2688 are shown in figure [ fig : afgl2688_psfsub ] .
note the remaining features near the north and at a position angle of @xmath19 east of north , about 150 arcseconds from location of the source in the psf subtraction in figure [ fig : afgl2688_psfsub]a .
this residual can also been seen in the roughly the same region of the 70 @xmath1 m psf subtraction of oh 231.8 in figure [ fig : oh231_psfsub ] .
it seems unlikely that these features are physical , but the northern feature in afgl 2688 does align with its bipolar outflows seen in the optical and infrared , but which are observed within the inner 10 arcseconds of the ppn .
the blob in afgl 2688 at a position angle of @xmath19 also roughly aligns with the outflow direction seen in previous co observations , also at around 10 arcseconds from the central star . the fact that the features we see are at nearly 200 arcseconds from the source and lie almost on top of the diffraction rays of the original psf casts doubt upon their reality .
this dilemma can not be resolved until a well sampled empirical psf out to several arcminutes is available .
we may be seeing possible features in the wings of the actual psf that the model has failed to reproduce . as a substitute for an empirical psf , we used oh 231.8 to subtract the psf from afgl 2688 .
figure [ fig : afgl2688_psfsub]b shows the result .
the residuals from using oh 231.8 as an empirical psf appear to be much smaller than those obtained using the model psf , reducing the residual of 30 mjy sr@xmath4 to less than 10 mjy sr@xmath4 at 100 arcseconds .
note that there are almost no psf features , such as diffraction spikes , remaining in figure [ fig : afgl2688_psfsub]b compared to figure [ fig : afgl2688_psfsub]a .
however , it may be problematic to use oh 231.8 + 4.2 as an empirical psf because it may have extended emission as well , although it would be remarkably fortuitous if both sources had extended emission having precisely the same morphology .
it is also problematic to scale oh 231.8 to the same surface brightness as afgl 2688 because in order to subtract the psf , the whole image needs to be scaled , thus scaling the background as well .
because afgl 2688 is over twice as bright as oh 231.8 , but the background in afgl 2688 is not , scaling to the level of the afgl 2688 psf would overemphasize the background in oh 231.8 so that the residuals after the psf subtraction might be dominated by the background in oh 231.8 .
we can nevertheless see that the psf subtraction using oh 231.8 results in residuals at a level @xmath6 10 times below that from using the model psf at 150 arcseconds , which suggests that oh 231.8 is either a point source at 70 @xmath1 m or that its extended structure has the same orientation , scale , and shape as in afgl 2688 . the two - dimensional psf subtraction at 160 @xmath1 m was done in a similar way using the model _ stinytim _ psf .
there are psf features remaining after this subtraction as well ( see figure [ fig : afgl2688_psfsub ] ) ; the ends of the diffraction spikes at 200 arcseconds from the source in the original image are not completely removed .
we did not attempt a psf subtraction of the 160 @xmath1 m image of iras 16342@xmath03814 , because that the psf of the central source is weak enough at this wavelength that its features are below the background and the extended structures . in order to investigate the hypothesis of spherical shells around a central source ,
we have determined an azimuthal average of the intensity around the central source .
such an average would show enhancements for spherically concentric shells projected as circularly symmetric features from episodic mass loss during the agb phase .
an azimuthal average also has the advantage that it greatly reduces the parameter space necessary to fit the model psf .
the source position and intensity are found by fitting to the brightest airy ring visible in our images , which is the second brightest airy ring of the psf ( see figure [ fig : model_psf ] ) , since the source itself does not appear in our images .
the psf airy ring used for scaling the psf is located 76 arcseconds from the source at 70 @xmath1 m and 170 arcseconds from the source at 160 @xmath1 m .
we also find that the azimuthal averages provide a subtraction with less residuals than using the two dimensional subtraction because localized irregularities of the psf are averaged away .
the one dimensional psf subtractions for afgl 2688 and iras 16342@xmath03814 at 70 @xmath1 m were done using both the model psf and oh 231.8 , and are shown in figure [ fig : afgl2688_70_1d ] . the profiles of afgl 2688 and oh 231.8 follow each other almost exactly out to beyond 200 arcseconds from the source .
this further indicates that oh 231.8 and afgl 2688 are both strongly centrally concentrated at 70 @xmath1 m , unless they both have exactly the same azimuthal excesses .
they both also match the model psf very well out to 100 arcseconds beyond which both appear to have slight excess ( @xmath20 1% of the extrapolated peak of the psf ) , above the model psf . once again
, this excess in the wings is more likely due to the model psf not reproducing the wings of the actual psfs rather than resulting from physical emission associated with the sources . at 160 @xmath1 m
, we only use the model psf for subtraction because of possible contamination due to galactic cirrus in oh 231.8 .
at 70 @xmath1 m , the psf - subtracted images of both afgl 2688 and oh 231.8 are similar .
the psf residuals extend to about 150 arcseconds from the source , which may be obscuring some physical extended features . however , there are no obvious circularly symmetric residuals in the psf subtraction even with the confusion near the central source , which would be expected if there had been spherically symmetric mass loss above a dust mass loss rate of @xmath21yr@xmath4 ( see discussion section ) .
previous studies with iso by @xcite led to the claim of large shells around afgl 2688 having radii of 150 and 300 arcseconds on the basis of a one - dimensional scan of the egg nebula with isophot at 120 and 180 @xmath1 m .
we find , with better spatial resolution and azimuthal coverage using spizer , that at 70 @xmath1 m , there are no signs of shell - like extended emission in our field . at 150 arcseconds from the source ,
the psf subtraction residuals have a surface brightness of about 16 mjy sr@xmath4 using the model psf and about 1 mjy sr@xmath4 using oh 231.8 as the empirical point source(figure [ fig : afgl2688_70_1d ] ) .
note also that there is no discernible excess emission above the fluctuations in the background ( @xmath22 mjy sr@xmath4 ) at 300 arcseconds from the source .
the 160 @xmath1 m image ( see figure [ fig : afgl2688_psfsub ] ) shows roughly uniform enhanced emission to the edge of our field of view at 1000 arcseconds from the source in the eastern scan path ; the northern scan show enhanced emission out to about 300 arcseconds , then dropping off to background levels .
the enhanced emission in the two paths is asymmetric as it does not fall off with distance in the eastern scan path .
there does not appear to be any structure to the emission in either scan , other than that which may be attributable to the psf diffraction spike in the north .
because the residual extended emission shows no symmetry about the central star , and because it shows no intensity falloff in the eastern scan path , this emission most likely arises from irregularly distributed galactic cirrus in this direction ; this is supported by the relatively lower quality iras 100 @xmath1 m image , which shows substantial extended cirrus beyound our survey region .
the isophot observation of a possible shell at 300 arcseconds was based on a @xmath23 linear scan with @xmath24 pixels , at a position angle of 8 degrees east of north centered on the source , which in our case would be sampling the emission in the northern scan .
we see with better azimuthal coverage with mips that the emission in both of our scans is likely strongly contaminated with galactic cirrus .
the presence of the inner shell reported by speck et al . at 150 arcseconds can not be directly confirmed by our observations because the 160 @xmath1 m psf airy ring at 170 arcseconds from the source overlaps that region .
the surface brightness at the airy ring is about 17 mjy sr@xmath4 after background subtraction ( but before the psf subtraction ) .
this is comparable to the background subtracted surface brightness of 20 ( 30 ) mjy sr@xmath4mjy sr@xmath4 at 120 ( 180 ) @xmath1 m reported by @xcite . in the azimuthally averaged surface brightness of our 160 @xmath1 m image ( figure [ fig : afgl2688_160_1d ] ) , we see a similar result as the 70 @xmath1 m data . speck et al .
reported a brightness value of the putative shell at 300 arcseconds from the source to be 20 mjy sr@xmath4 at 120 @xmath1 m and 60 mjy sr@xmath4 at 180 @xmath1 m , without background subtraction .
the 120 @xmath1 m emission is about 10 mjy sr@xmath4 above their background as extrapolated from the surface brightness far from the star in their plot . at 180 @xmath1 m
, the excess emission is about 15 mjy sr@xmath4 above the background . in the mips
160 @xmath1 m image , we find that the surface brightness at 300 arcseconds from the source is about 2 mjy sr@xmath4 after background subtraction .
although the comparison is not at the same wavelength , the excess emission we detect is 7 times below that measured by iso .
when comparing the two scans in our observations separately , we see that the eastern scan has roughly uniform brightness of 4 mjy sr@xmath4 out to 1000 arcseconds , while the emission in the northern scan drops off at 400 arcseconds from the source from about 2 mjy sr@xmath4 to the level we use as the background in the data reduction .
both the asymmetry in the enhanced emission in both scans and the low surface brightness in the spitzer data at 300 arcseconds from the source lead us to conclude that the shell at this distance reported on the basis of iso data is probably not associated with the source .
neither the 70 nor 160 @xmath1 m images of oh 231.8 show any signs of spherically symmetric extended emission .
the psf subtraction at 70 @xmath1 m leaves similar residuals as for afgl 2688 , though the brightnesses of these residuals are much lower because the central source is intrinsically less bright .
the strongest psf residuals are along the diffraction spikes , similar in shape to those from afgl 2688 .
this further suggests that these residuals are not physical features , but are rather psf artifacts not accounted for by the _
stinytim _ model .
in the 160 @xmath1 m azimuthally averaged plots , both scans appear to have rather uniform emission , with the northern scan having slightly higher surface brightness at approximately 2 mjy sr@xmath4 , while the eastern scan has an average surface brightness of 1 mjy sr@xmath4 ( figure [ fig : oh231_70_1d ] ) .
the 160 @xmath1 m image shows some slightly clumpy emission along the northern scan path , slightly beyond the psf diffraction spike ( see figure [ fig : initial_mosaic ] ) at 320 arcseconds .
the clumps , with a width of about 80 arcseconds , have a surface brightness of about 3.6 mjy sr@xmath4 on a 2 mjy sr@xmath4 diffuse background .
the iras 100 @xmath1 m image shows diffuse galactic cirrus in this region , which may be the origin of these clumps .
they are aligned with the bipolar axis , but unfortunately also with the diffraction ray so it may also be a psf feature . since this study is concerned with investigating spherically symmetric emission , we will defer the question of their nature .
the azimuthally averaged surface brightness shows no enhanced emission attributable to the source ( figure [ fig : oh231_160_1d ] ) .
in contrast to the previous two sources , iras 16342@xmath03814 does show some evidence for what may be a large patchy shell at both 70 and 160 @xmath1 m .
the enhanced emission out to a radial distance of @xmath6 400 arcseconds is consistent with rough circular symmetry about the star in the coverage area available ( figure [ fig : initial_mosaic ] ) .
this radius corresponds to a physical size of about 4 parsecs , assuming the distance to iras 16342@xmath03814 to be 2 kpc @xcite .
this possible shell structure is rather patchy , with a large arc in the western scan at 70 @xmath1 m . at 160 @xmath1 m
, the same arc is present , but is more diffuse .
nevertheless , the azimuthal average shows an excess above the background at @xmath6 400 arcseconds from the central source .
the brightest portion of this feature occurs at a position angle of 250 degrees east of north , which is the same direction as the axis of the bipolar nebula seen in the optical .
the ratio of the 70 to 160 @xmath1 m average surface brightness , is roughly constant ( @xmath25 ) as a function of radial distance from the source .
this ratio yields a color temperature of @xmath26 k , and , if a @xmath27 emissivity law is assumed , a dust temperature of @xmath28 k. the constant color temperature suggests that the dust is heated primarily by the interstellar radiation field . because there are a number of interstellar cirrus features in the general direction of iras16342 - 3814 , we can not definitively argue that the patchy arc of emission in the scan paths around this source is caused by mass loss from the star .
the iras 100 @xmath1 m image of this region with an overlay of the coverage of our data ( figure [ fig : iras16342_iras100 ] ) shows that there is indeed a substantial likelihood that this emission is cirrus . with this caveat in mind
, we proceed in the analysis below on the assumption that the extended emission has resulted from mass loss from the star to obtain some indicative numbers for the mass loss rate necessary to produce such a shell . in the following analysis
, we will use the term `` mass outflow rate '' to indicate the total mass flowing outward per unit time across a spherical shell at a particular radius ; this includes both the stellar component and the interstellar medium being swept up ; we will use the term mass loss rate to refer to mass loss from only the star . making a few assumptions about the mass outflow and the dust
, we can set some limits on the mass outflow rate of iras 16342@xmath03814 . assuming that the mass loss is spherically symmetric
, we can fit a model of the expected brightness profile to the observed radial profile of the source .
figures [ fig : iras16342_70_1d ] and [ fig : iras16342_160_1d ] show the azimuthally averaged radial profiles at 70 and 160 @xmath1 m , respectively .
we fit only the 70 @xmath1 m radial profile because there is less contamination by galactic cirrus compared to 160 @xmath1 m . following the derivation by @xcite ,
if we assume a @xmath29 density profile , appropriate for constant mass outflow , where @xmath30 is the distance from the star , and a constant expansion velocity , @xmath31 , then the surface brightness @xmath32 has the form : @xmath33 \quad\mbox{for } ~ b < r_{min } \label{eq : shell_model}\end{aligned}\ ] ] where @xmath12 is the projected physical distance from the source in the sky , @xmath34 is the opacity coefficient in units of @xmath35 g@xmath4 , @xmath36 is the blackbody function , and @xmath37 and @xmath38 are the outer and inner radii of the shell .
we further assume that the outflow velocity is constant at 15 km s@xmath4 , and that the dust is composed of small astronomical silicate grains with @xmath34 = 104 @xmath35 g@xmath4 at 70 @xmath1 m ( using a = 0.1 @xmath1 m , @xmath39
g @xmath40 , and @xmath41 from @xcite ) . we also adopt a constant temperature of 26 k throughout the shell and envelope .
figure [ fig : iras16342_model ] gives a comparison between different models having various mass outflow rates .
the best - fit model is a shell with a 4.2 pc radius and a thickness of 1 pc , with a dust outflow rate of @xmath42 yr@xmath4 superimposed on a smooth envelope with a constant dust outflow rate of @xmath43 yr@xmath4 .
this implies a rather high gas outflow rate on the order of @xmath44 yr@xmath4 in the shell and @xmath45 yr@xmath4 in the smooth envelope , assuming a gas to dust mass ratio of 200 . using oh 231.8 as the empirical psf
, the fit only requires a shell component with a mass outflow rate of @xmath44 yr@xmath4 without an underlying smooth envelope .
these fits depend on the choice of psf , but once the psf is chosen , the mass outflow rates are insensitive to residuals from psf subtraction , since we only fit for the region beyond 150 arcseconds from the central source ( see figure [ fig : iras16342_70_1d ] ) .
however , our value for the mass outflow rate is highly dependent on the assumed temperature of the dust . for typical ism dust temperatures between 22 and 35 k ,
the inferred dust outflow rate is related empirically to the adopted temperature by : @xmath46 at 70 @xmath1 m .
if t @xmath47 26 k , we would infer a substantially lower mass outflow rate . given the model thickness of the shell and assuming a typical expansion velocity of 15 km s@xmath4
, we estimate the maximum duration of the enhanced mass loss event which produced the shell to be about 65,000 years .
this is an upper limit because the internal velocity dispersion in the shell broadens the shell as it expands . a 1 km s@xmath4 dispersion , for example ,
would reduce the duration of the mass loss that produced the shell by about a third .
if we use an age of 40,000 years for the duration of the enhanced mass loss and approximate the shell as spherical , the total dust mass would be about 0.04 @xmath48 , giving a total shell mass of @xmath6 8 @xmath48 , assuming a gas - to - dust mass ratio equal to 200 .
the velocity of the shell is probably slower than the molecular outflow velocity of a typical agb star and the total amount of mass lost by the agb star is likely smaller than this amount because of the interaction with the ism out to the distances we are observing ( see below ) .
the mass of the shell can be estimated directly from the infrared emission using the equation @xmath49 , where @xmath50 is the total flux density in the shell and @xmath51 is the distance to the source .
the integrated flux density from the limited azimuthal coverage that we have between radii of 300 and 400 arcseconds is 6.4 jy at 70 @xmath1 m , corresponding to a dust mass of 0.015 @xmath52 in our observed portion of the shell , assuming a temperature of 26 k for the dust .
we have about one fifth of the full azimuthal coverage with data at this distance from the star , and can estimate the total dust mass of the shell if we approximate the shell as isotropic with an averaged surface brightness of 2.5 mjy sr@xmath4 at 70 @xmath1 m from the radial profile , with a width of 100 arcseconds , and a temperature of 26 k. the total dust mass of the extrapolated shell is then about 0.03 @xmath48 , comparable to the above estimate .
because the shell is so large , the amount of interstellar matter that may have been swept up and accumulated in the shell during the agb could be significant . for an ism hydrogen density of 1 @xmath40
, a 4 parsec radius shell would have swept up about 7 @xmath48 , which would account for almost all the mass that we may be measuring .
the accumulation of interstellar material could potentially lower the stellar mass loss rate estimated from our model by a large factor .
we therefore emphasize that the mass outflow rate estimate is only an upper limit to the stellar mass loss rate .
one of the difficulties in addressing the mass loss history of iras 16342@xmath03814 with the current data is its location in the galaxy ; with a scale height of only 150 pc , there is possible confusion with diffuse galactic cirrus .
an unfortunate alignment with background emission is an alternative to the existence of a shell produced by mass loss .
further observations with better azimuthal coverage are needed to test the circumstellar shell hypothesis .
the extended dust emission in afgl 2688 and oh 231.8 seems to show that these two objects do not have as long a mass loss history as one might have anticipated .
our results for these two sources probe the region beyond @xmath6 100 arcseconds from the central star , which corresponds to @xmath53 yrs ago for afgl 2688 and @xmath54 yrs ago for oh 231.8 , given a distance of 420 pc @xcite and 1.2 kpc @xcite , respectively , and an expansion velocity of 20 km s@xmath4 .
we can can set a limit to the dust mass loss rate by inverting equation [ eq : shell_model ] : @xmath55 the dust mass loss rate can be simplified as a function of the observed surface brightness @xmath56 and the projected distance from the source in the sky @xmath57 with the assumption that @xmath58 : @xmath59 if @xmath60 , then the full form of equation [ eq : mdot_limit ] must be used because the @xmath61 term in the denominator of equation [ eq : mdot_limit ] becomes important and will cause the dust mass loss rate inferred from a given surface brightness value to increase drastically . for the cases such as afgl 2688 and oh 231.8 where we do not see a clear envelope associated with the source , we can not be sure that the condition @xmath62 holds since @xmath37 is indeterminate .
the equations above also show the temperature dependence in the form of a power law to approximate the blackbody function between 22 and 35 k. from these equations we see that the temperature dependence at 160 @xmath1 m is less steep than at 70 @xmath1 m , but the contamination by galactic cirrus near the sources makes estimating upper limits problematic at 160 @xmath1 m .
we can establish an upper limit to the mass loss rates for afgl 2688 and oh 231.8 at 70 @xmath1 m based up on the surface brightness of the residual left from the psf subtraction at 200 arcseconds from the source , which is a compromise between a location far enough from the source that psf subtraction errors are small and being close enough to the source that there could plausibly be a circumstellar envelope . for afgl 2688 , 200 arcseconds corresponds to a radial distance from the source of 0.4 parsecs , using a distance to the source of 420 parsecs @xcite .
the surface brightness after psf subtraction with the model psf is about 14 mjy sr@xmath4 at 70 @xmath1 m .
assuming an expansion velocity of 20 km s@xmath4 as observed in co by , a dust temperature of 30 k , @xmath34 = 104 @xmath35 g@xmath4 appropriate for carbon dust @xcite , and that the possible envelope has a radius significantly greater than 0.4 pc , then the dust mass loss rate upper limit is @xmath63 m@xmath3 yr@xmath4 .
similarly for oh 231.8 , the residual emission from the psf subtraction is 3 mjy sr@xmath4 at 200 arcseconds from the central star , which corresponds to 1.3 parsecs at a distance of 1.3 kpc .
this surface brightness implies a dust mass loss rate of @xmath5 m@xmath3 yr@xmath4 assuming an outflow velocity of 15 km s@xmath4 , @xmath34 = 98 @xmath35 g@xmath4 ( appropriate for silicate dust ) , and a temperature of 30 k. there are several potential explanations for the lack of extended emission seen in this study .
one is that , even with spitzer s increased sensitivity , the dust emission is below the threshold for detection .
since the distances we are studying are far from the central star , we would expect the temperature of the dust to be determined by the ambient interstellar radiation field .
if the dust temperature is about 20 k , typical for the ism , and an emissivity @xmath64 , the surface brightness would be 12 times stronger at 160 @xmath1 m , than at 70 @xmath1 m where the resolution and sensitivity of mips is better .
we would be more likely to detect cool dust emission at 160 @xmath1 m than at 70 @xmath1 m .
however , confusion with the galactic cirrus is also likely at 160 @xmath1 m because it is about the same temperature as any hypothetical extended dust emission associated with the source that is heated primarily by the interstellar radiation field . for comparison
, our detection of the possible shell from iras 16342 - 3184 has a ratio of @xmath65 , corresponding to a color temperature of 32 k , which requires a higher than average interstellar radiation field .
the upper limits to the mass loss rates derived in this study can be compared with theoretical agb evolutionary models , particularly those that predict enhanced mass loss rates due to thermal pulsation near the end of the agb .
the spatial coverage in this study between @xmath6 200 to 1000 arcseconds from the central star translates into a probe of the history of mass loss between @xmath66 yrs to @xmath67 years ago for afgl 2688 and @xmath68 yrs to @xmath69 yrs ago for oh 231.8 , using an expansion velocity of 20 km s@xmath4 for both sources .
the models by @xcite show that for a 2.0 m@xmath3 progenitor , during the last few @xmath70 yrs of agb evolution , there are several thermal pulses which result in enhanced mass loss rates peaking at about @xmath71 m@xmath3 yr@xmath4 during the end of each pulse . using a gas to dust mass ratio of 200
, we find that for afgl 2688 and oh 231.8 , the upper limit to the total mass loss rate during the above time intervals is about @xmath72 m@xmath3 yr@xmath4 and @xmath73 m@xmath3 yr@xmath4 , respectively .
these limits are close to the sensitivity necessary to see shells that may be the result of thermal pulsation on the agb .
although we do not detect the shells reported by @xcite , which they attribute to thermal pulses , the signatures of thermal pulses may still be present , but below our current sensitivity .
we also consider the possibility that oh 231.8 does not show extended emission in the far infrared because of interactions with a binary companion , which caused the central star to lose mass more rapidly and more recently than during the evolution of a lone agb star . a companion to qx pup ( the central star of oh 231.8 ) ,
is evidenced by optical spectra consistent with a companion of stellar type a0 v @xcite .
qx pup also shows a peculiar paradox of being an m9iii @xcite agb star , while its bipolar activity and morphology display all the signs of post - agb activity of typical ppne . having a close companion would enhance the mass loss rate and provide a mechanism for generating the bipolar outflows @xcite , thus shortening its mass loss history enough that we should not be surprised to see no emission far from the source .
in contrast to the collimated outflow from oh 231.8 , afgl 2688 has a spherically symmetric envelope seen in @xmath74co @xcite and evidenced by the partial , concentric , circular arcs present in hst images @xcite .
since a binary interaction would not cause the past spherically symmetric mass loss , a possible companion is probably not a good explanation for initiating mass loss , although the present bipolar morphology of afgl 2688 is consistent the possibility that a binary interaction has altered the mass loss in more recent times .
tracing the mass loss history during the agb phase should be relatively straightforward via mapping the emission from the circumtellar envelope .
but in practice , it has been difficult because molecular - line observations are ultimately limited by the photodissociation of molecules in the outer envelope regions , and far - infrared observations of dust emission have been limited by the generally low angular resolution of the space - based telescopes which have been available for this purpose in the past ( iras and iso ) .
hence the reported detections of very extended emission in a few dying stars with iras and iso has generally been recognized as an important milestone in the study of mass loss . however , such detections have also raised the very important question of why only a few select objects like y cvn or ry dra @xcite not particularly known for having high mass loss rates show extended envelopes , whereas large numbers of stars having high co - determined mass loss rates do not reveal the presence of such envelopes .
is this because the radial density law for most of these high mass - loss stars is significantly steeper than in objects like y cvn or ry dra , and in particular steeper than @xmath75 ( an issue of profound importance for the evolutionary times of stars through the agb phase and theories of mass - loss ) , or are the claimed detections of extended envelopes really a result of poorly characterised instrumental artifacts ?
the spitzer data presented in this paper clearly show that the presence of these shells can not be confirmed at a level well below the intensities expected from speck et al.s results .
our non - detections call into question not only the iso results on afgl 2688 , but all results on the detection of extended envelopes in other objects using the same linear scan technique described by speck et al .. more detailed mapping of many high mass - loss rate objects like afgl 2688 , to search for such shells , is crucially needed .
spitzer observations of extended envelopes of expanding , dusty outflows from bipolar agb and post - agb stars reveal that there may be a very large dust shell having a radius of 400 arcseconds around iras 16342@xmath03814 .
the combination of the presence of nearby cirrus emission and the limited azimuthal coverage in the images makes the conclusion for a shell uncertain , but if the shell is indeed the result of an episodic mass loss event , then it would represent one of the largest dust shells found so far . our observations of afgl 2688 at 70 @xmath1 m do not show the dust shells at 150 and 300 arcseconds from the source reported by @xcite . since the dust shell may be very cool , the non - detection at 70 @xmath1 m does not alone rule out dust shells . however , we find that there is only a slight excess at 160 @xmath1 m of 2 mjy sr@xmath4 above the background at the reported location of the outer shell ( @xmath76 arcseconds ) . with greater azimuthal coverage than was previously obtained with isophot
we find that there is substantial contamination by galactic cirrus in the region at 160 @xmath1 m , with galactic cirrus emission above the sky background present throughout the entire eastern scan path of our observations .
unfortunately , at 160 @xmath1 m we can only set an upper limit to the emission from a dust shell at 150 arcseconds from the source because one of the airy rings of the psf overlaps that region .
we also see no extended emission from oh 231.8 at either 70 or 160 @xmath1 m other than that attributable to galactic cirrus .
in fact , using oh 231.8 as an empirical point source for psf subtraction from afgl 2688 at 70 @xmath1 m appears to substantially reduce psf subtraction residuals compared to using the model psf .
the limitation of our method of observation is that we have only two radial directions to probe possible extended emission . for cases like iras 16342@xmath03814 where the shell is patchy
, better azimuthal information would help to resolve whether the origin of the emission is from the star or from galactic cirrus .
the mips data show that observing very extended emission in the far infrared is possible , but difficult because of the prominence of galactic cirrus emission at these wavelengths .
this work is based on observations made with the spitzer space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was provided by nasa through an award issued by jpl / caltech .
mm and rs was partially funded for this work from a go spitzer award and an ltsa award ( no . 399 - 20 - 40 - 06 ) from nasa .
.the 70 @xmath1 m mean surface brightness and sensitivity for a region of uniform background far from the source , before and after the median background removal described in the text .
[ cols="^,^,^,^,^ " , ] .
the bars in the images are 200@xmath77 in length for scale and the @xmath78 marks the location of the point source .
top : afgl 2688 .
center : oh 231.8 .
bottom : iras 16342@xmath03814 .
note the prominence of the psf features at 70 @xmath1 m in all the images . at 160 @xmath1 m ,
the psf is not apparent in iras 16342@xmath03814 .
north is up and east is to the left . ]
3814 after the subtraction of the model psf .
we find that the best fit to the radial brightness profile obtained from subtracting the model psf is a two - component model with a shell of radius 4.2 pc and thickness 1 pc requiring a dust mass loss rate of @xmath79 @xmath48 yr@xmath4 superimposed upon an envelope from a constant dust mass loss of @xmath80 @xmath48 yr@xmath4 ( dashed blue line ) .
the asterisks show the radial profile found from using oh 231.8 as a substitute for the model psf in the psf subtraction will fit using only the shell component with a dust mass loss rate of @xmath79 @xmath48 yr@xmath4 ( dashed orange line ) .
points closer than a radial distance of @xmath6 150 arcseconds are unreliable because they are most affected by psf subtraction residuals and are not used in the fit . ] | we present the results of far - infrared imaging of extended regions around three bipolar pre - planetary nebulae , afgl 2688 , oh 231.8 + 4.2 , and iras 16342@xmath03814 , at 70 and 160 @xmath1 m with the mips instrument on the spitzer space telescope . after a careful subtraction of the point spread function of the central star from these images ,
we place constraints on the existence of extended shells and thus on the mass outflow rates as a function of radial distance from these stars .
we find no apparent extended emission in afgl 2688 and oh 231.8 + 4.2 beyond 100 arcseconds from the central source . in the case of afgl 2688 , this result is inconsistent with a previous report of two extended dust shells made on the basis of iso observations .
we derive an upper limit of @xmath2 m@xmath3 yr@xmath4 and @xmath5 m@xmath3 yr@xmath4 for the dust mass loss rate of afgl 2688 and oh 231.8 , respectively , at 200 arcseconds from each source .
in contrast to these two sources , iras 16342@xmath03814 does show extended emission at both wavelengths , which can be interpreted as a very large dust shell with a radius of @xmath6 400 arcseconds and a thickness of @xmath6 100 arcseconds , corresponding to 4 pc and 1 pc , respectively , at a distance of 2 kpc .
however , this enhanced emission may also be galactic cirrus ; better azimuthal coverage is necessary for confirmation of a shell .
if the extended emission is a shell , it can be modeled , with some assumptions about its dust properties , as enhanced mass outflow at a dust mass outflow rate of @xmath7 m@xmath3 yr@xmath4 superimposed on a steady outflow with a dust mass outflow rate of @xmath8 m@xmath3 yr@xmath4 . because of the size of the possible shell , it is likely that this shell has swept up a substantial mass of interstellar gas during its expansion , so these estimates are upper limits to the stellar mass loss rate .
we find a constant color temperature of 32 k throughout the circumstellar envelope of iras 16342@xmath03814 , which is consistent with heating by the interstellar radiation field . |
most galaxies harbor weakly- or non - active central supermassive black holes ( bhs ) . roughly once every
@xmath6 years in each galaxy , a star enters the bh s tidal disruption radius within which the tidal force of the bh exceeds the star s self gravity , and hence the star gets tidally disrupted @xcite . in these so - called tidal disruption events ( tdes ) , the stellar debris feeds a burst of strong accretion that generates a bright flare of electromagnetic radiation .
several dozen such flares have been reported at various wavelengths , e.g. hard x - ray/@xmath7-ray @xcite , soft x - ray ( reviewed by * ? ? ?
* ) , uv ( e.g. * ? ? ?
* ; * ? ? ?
* ) and optical ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if a large fraction of the stellar debris is accreted by the bh , the total energy budget is @xmath8 , where @xmath9 is the efficiency of energy release from accretion and depends on the bh spin .
however , the observed radiation energy is usually @xmath10 , much smaller than expected . especially , the energies of the optically discovered tdes lie at the lower end ( @xmath11 ) . a possible solution to the `` efficiency problem
'' could be that @xmath12 because most of the stellar debris does nt reach the bh but is blown away by a wind at a radius of @xmath13 to 100 schwarzschild radii @xcite
. if the wind is dense enough , the effective photosphere could be far from the surface of the disk , and the radiation energy suffers from adiabatic loss before escaping , so the spectrum appears redder and the total radiation energy lower .
another possibility is that the spectral energy distritubion ( sed ) peaks between the observable far uv and soft x - ray windows .
we can see that tdes release a huge amount of radiation energy , which will have a significant impact on the interstellar medium in the immediate vicinity of the bh .
for example , as pointed out by @xcite , if most of the flare s energy is radiated in the extreme uv ( @xmath14 ) , the bright flash can ionize gas of mass @xmath15 out to a radius @xmath16 where @xmath17 , @xmath18 is the number density of h nuclei and we have assumed each h nucleus consumes @xmath19 ( ionization potential of h and he + bond energy of h@xmath20 + kinetic energy of free protons and electrons ) . in eq.([eq:20 ] ) and ( [ eq:7 ] ) , recombination is ignored because the timescale @xmath21 is expected to be long
. then the energy of the ionizing flash will be slowly reprocessed into potentially observable h and he recombination lines if the gas density is high , or forbidden lines ( of metals ) if the gas density is low .
however , eq.([eq:20 ] ) and ( [ eq:7 ] ) are true only if there is no dust extinction .
if dust is well mixed with the gas , the v - band extinction is proportional to the gas column density and we have @xmath22 for sightlines with @xmath23 @xcite .
the v - band extinction of the gas with a column depth of @xmath24 is @xmath25 therefore , for our fiducial galactic center density of @xmath26 , the ionizing flash will be significantly attenuated at a radius of @xmath27 , unless dust is significantly depleted in the gas before the tde occurs . the radiation energy will be deposited into dust grains , which then re - radiate in the infrared ( ir ) .
a similar process has been studied in the context of gamma - ray bursts ( grbs ) by @xcite and @xcite . as we show in this work ,
if a significant amount of dust exists in the pc - scale vicinity of the bh , the dust ir emission from tdes is much brighter than that from grbs , lasts longer , and is hence easier to observe . before moving forward to calculations of the dust ir emission , we summarize our current ( very limited ) knowledge about the pc - scale dust content around galactic nuclei . in the context of active galactic nuclei ( agn ) ,
the accretion disk and broad line region are surrounded by a thick dusty torus .
the dusty torus absorbs a significant fraction of nuclear luminosity and re - radiates mid - ir continuum as observed in most agns .
typically , the sky covering factor of the torus is @xmath28 and inner boundary is roughly given by the dust sublimation radius @xmath29 , where @xmath30 is the disk luminosity @xcite .
the torus is dynamically active @xcite , possibly coming from a clumpy wind launched from the accretion disk @xcite . for weakly - active galactic nuclei with luminosities @xmath31 ,
the accretion disk can no longer sustain a wind outflow rate required by the mid - ir emission extrapolated from high - luminosity agns , so the torus is expected to disappear @xcite .
this has not been observationally confirmed , because the dust emission no longer dominates in the mid - ir when @xmath32 . as pointed out by @xcite , even at these luminosities
, the disk outflow can still provide a small but significant toroidal obscuration as long as the column density exceeds @xmath33 . in the context of non - active galactic nuclei ,
the nuclear dust content is hard to probe due to the overwhelming starlight and telescopes limited angular resolution ( 1 arcsec @xmath34 50 pc at a distance of 10 mpc ) .
currently , the only observable case is our galactic center .
radio observations ( of tracer molecules ) revealed @xmath35 of molecular clouds and diffuse molecular gas at @xmath36 from the bh , mostly in a torus - like structure ( known as circum - nuclear disk , cnd ) with a sky covering factor @xmath37 viewed from the bh .
inward from the cnd , due to the radiation from the central star cluster ( @xmath38 ) , the gas becomes atomic and then ionized .
the photo - dissociation region and hii region are both dusty and are observable through atomic fine - structure lines @xcite , free - free continuum @xcite , recombination lines @xcite , and mid / far - ir dust thermal emission @xcite .
the densities of different regions estimated by these works are @xmath39 ( hii regions ) , @xmath40 ( warm neutral medium and molecular streamers ) , and @xmath41 ( molecular cores )
. the dust within a few parsecs of the galactic center may originate from infalling molecular clouds / clumps @xcite or from in- _ situ _ dust formation @xcite .
dust grains may be destroyed by two processes : sublimation and sputtering .
the former may be unimportant because the dust sublimation radius is @xmath42 in weakly- or non - active galactic nuclei .
the latter process is significant only in the shock heated hot ionized medium where the gas temperature @xmath43 .
we note the fact that the galactic center is unusually dusty may be because the milky way is a gas - rich star - forming galaxy .
the nuclear dust content of elliptical galaxies is unknown but may be much less than in our galaxy .
as we show in this work , the pc - scale dust content around galactic nuclei can be probed by searching for the dust ir emission after a tde is discovered . in section [ sec :
estimation ] , we give an order - of - magnitude estimation of the ir emission from the dust heated by tdes . in section [ sec : rt ] , we calculate the bolometric lightcurve from a 1-d time - dependent radiative transfer model . in section [ sec : discussion ] , we discuss the detectability and implications , as well as some possible issues in our simple calculations .
a short summary is given in section [ sec : summary ] .
lacking the knowledge of the dust distribution , we assume spherical symmetry for simplicity .
consider a point source with uv - optical luminosity @xmath44 as a function of time @xmath45 in the rest frame of the source .
we assume a uniform spatial distribution of dust grains with number density @xmath46 , spanning from inner radius @xmath47 to outer radius @xmath48 . as the uv
- optical radiation propagates through the cloud , dust grains will be heated up . for a grain of radius @xmath49 at radius @xmath50 from the source , its temperature @xmath51 is determined by the equilibrium between heating and radiative plus grain - sublimation cooling @xcite @xmath52 where @xmath53 is the effective attenuation of the uv - optical flash by inner dust shells , @xmath54 is the `` retarded '' time ( corresponding to the time segment of the uv - optical flash ) , @xmath55 @xcite is the absorption efficiency factor of uv - optical radiation , @xmath56 is the planck - averaged absorption efficiency factor , @xmath57 is the stefan - boltzmann constant , @xmath58 is the density of the grain material , @xmath59 is the mean atomic mass , and @xmath60 is the chemical binding energy .
for the temperature range of interest in this work @xmath61 , we use an intermediate absorption coefficient between astronomical silicate and graphite @xcite @xmath62 where the grain radius is normalized to a typical value of @xmath63 .
the thermal sublimation rate is a function of temperature and can be estimated by @xmath64 we take @xmath65 , @xmath66 , and @xmath67 as representative values for refractory grains @xcite . from eq.([eq:3 ] ) , we get a characteristic grain survival time at temperature t @xmath68 \ \mt{s } \end{split}\ ] ] if the duration of the tde uv - optical flash is @xmath69 , the `` sublimation temperature '' ( above which dust grains fully sublime ) can be defined by @xmath70 , i.e. @xmath71 \ \mt{k}\ ] ] it can be shown that the critical temperature at which radiative cooling equals to sublimation cooling is @xmath72 , so the sublimation cooling term in eq.([eq:1 ] ) is usually negligible and we can solve for the `` sublimation radius '' where the grain temperature equals to @xmath73 @xmath74^{1/2 } \
\mt{cm}\ ] ] if the whole dusty layer is optically thick to the uv - optical radiation from the central source , the total radiation energy @xmath75 will be absorbed and re - radiated at a typical wavelength of @xmath76 . because the dust sublimation rate increases steeply with grain temperature ( eq . [ eq:3 ] ) , grains at radii @xmath77 sublime fully before a significant fraction of uv - optical energy
is absorbed . from eq.([eq:5 ] ) , we can see that dust ir emission usually cuts off at wavelength shorter than @xmath78 , unless there is more dust at @xmath79 than the uv - optical radiation can cause to sublime .
the duration of the dust emission can be estimated by the light - crossing time @xmath80 , where @xmath81 is the speed of light and @xmath82 .
the ir luminosity of the dust emission can be estimated by @xmath83 where @xmath84 is the sky covering factor of the dusty clouds viewed from the bh and @xmath85 is the further attenuation of the dust ir emission by outer dust shells that are not heated up .
note that , when the dusty cloud is far enough from the central source that sublimation is not significant , most ir emission comes from the inner edge of the cloud and @xmath86 should be replaced by @xmath47 .
also , eq.([eq:9 ] ) is a conservative estimate because when @xmath87 , say the plane of the dust torus is oriented at an angle @xmath88 wrt .
the line of sight , the light - crossing time is a factor of @xmath89 shorter than @xmath90 .
in this section , we model the heating and cooling of dust grains with time - dependent radiative transfer in a 1-d spherical symmetric grid .
the bolometric lightcurve of dust ir emission is calculated . dust is distributed uniformly in the spherical layer between the inner radius @xmath47 and outer radius @xmath48 .
we assume all dust grains have the same initial radius @xmath91 and the number density @xmath46 is independent of radius .
we divide the whole dust layer into @xmath92 thin shells of thickness @xmath93 and the middle point radius of the @xmath94-th shell ( @xmath95 ) is @xmath96 . due to sublimation
, the grain radius in the @xmath94-th shell @xmath97 is a function of the retarded time @xmath98 ( local time since the arrival of the radiation front ) .
the total optical depth from @xmath47 to @xmath96 is @xmath99 where the extinction coefficient factor @xmath100 is from nearly equal contributions from absorption and scattering at uv - optical wavelengths @xcite .
note that the photons with scattering angle @xmath101 , after traveling a distance @xmath102 , are delayed by @xmath103 so the scattered photons suffer from significant delays . to keep the model simple and conservative , we ignore the absorption of the scattered photons .
we also ignore the small contribution to dust heating by the ir radiation from other shells , because of the time delay and the extinction at ir wavelength being small .
for instance , @xmath104 for @xmath105 @xcite .
actually , the heating flux on dust grains from uv and ir photons bouncing between different layers is a factor of @xmath106 smaller than the flux directly from the central source ( @xmath107 instead of @xmath108 on the left hand side of eq .
[ eq:1 ] ) .
therefore , on the light - crossing timescale ( a few years ) , the grain temperature is a factor of @xmath109 smaller and hence the reprocessed emission is around @xmath110 . combining eq.([eq:1 ] ) , ( [ eq:2 ] ) and ( [ eq:3 ] ) , we use a simple explicit scheme to solve the time evolution of grain temperatures @xmath111 and radii @xmath112 in each shell .
we take the time step to be one tenth of the local grain survival time @xmath113 we have tested that the results are not affected by a factor of 2 variations in @xmath114 in all cases presented . as shown below , if we only consider the dusty region to be at radius @xmath115 , the grain temperatures are always smaller than @xmath116 ( the temperature at which the radiative cooling equals to sublimation cooling ) .
since the sublimation cooling term in eq.([eq:1 ] ) drops fast with decreasing temperature [ @xmath117 , we ignore the sublimation cooling term in our calculations . from @xmath118
, we get the volume emissivity ( @xmath119 ) of the @xmath94-th shell @xmath120 as a function of retarded time through @xmath111 and @xmath112 . at any observer s time @xmath121
, we can add up the contributions from all the shells at different radii @xmath50 and emission latitudes @xmath88 to get the total ( isotropic equivalent ) ir luminosity @xmath122 where @xmath123 is the attenuation of the ir emission from position @xmath124 at @xmath125 by other dust shells along the line of sight .
the observer s clock starts ( @xmath126 ) when the radiation front from the central source arrives , so the ir photons emitted from position @xmath124 at retarded time @xmath125 will reach the observer at @xmath127 which gives @xmath125 as a function of @xmath121 , @xmath50 and @xmath88 . to calculate the ir optical depth @xmath128
, we integrate the absorption ) , scattering has a smaller cross section than absorption , so we ignore the contribution to the ir optical depth from scattering . ] by dust shells along the line of sight . as shown in fig.([fig : tauir ] ) , there are two cases : ( 1 ) if the emission latitude @xmath129 , the line of sight intersects once with all the outer shells ( @xmath130 ) ; ( 2 ) if the @xmath131 , the line of sight intersects once with all the outer shells and twice with the inner shells at radii @xmath132 .
we denote the angle between the light ray and radial direction as @xmath133 and it satisfies @xmath134 to calculate the optical depth contributed by each shell , we need to know the dust content at the shell s retarded time when the ir light ray arrives @xmath135/c \mbox { , if $ r_\mt{ab } > r$}\\ & t_\mt{r } + \left [ -r\cos \theta + r_\mt{ab}\cos \theta_\mt{ab } \right ] /c
\mbox { , if $ r_\mt{ab } < r$ } \end{cases}\ ] ] note that when the ir light ray arrives at an inner shell for the first time , we have @xmath136 and hence @xmath137 . in the model , the grain number density @xmath138 is fixed and the decreasing of dust content in a shell ( due to sublimation ) is captured by the decreasing of grain radii with its retarded time @xmath139 . when @xmath139 is smaller than @xmath140 of the initial grain radius @xmath141
, we consider the grain sublimation to be complete and the shell becomes dust free .
therefore , we can calculate the optical depth of the ir emission from position @xmath142 at @xmath125 : ( 1 ) the contribution from outer shells @xmath143 ( 2 ) the contribution from inner shells ( only when @xmath144 ) @xmath145 where i = 1 , 2 means the first / second time the ir light ray intersects with the absorbing shell .
note that the planck - averaged absorption efficiency factor @xmath146 ( eq . [ eq:6 ] ) depends on the grain size @xmath49 in the absorbing shell ( at @xmath147 , @xmath148 ) and the radiation temperature is assumed to be planckian .
we note that , as the ir light ray propagates through the dust shells , the spectrum will change due to reddening . however , in all computed cases , the absorption at ir wavelength ( @xmath149 ) is not very large , so the changing of spectrum is ignored . ] @xmath51 of the emitting shell ( at @xmath50 , @xmath125 ) .
we run the 1-d radiative transfer model for four cases , which are summarized in table [ tab : cases ] .
the following parameters are fixed in all four cases : the radial thickness of the dusty cloud @xmath150 , initial grain radius @xmath151 ( all grains having the same initial radius ) and the total dust optical depth @xmath152 ( @xmath153 ) .
the grain number density is constant with radius and is given by , the grain number density @xmath46 corresponds to mass density of @xmath154 . using a dust - to - gas mass ratio of 0.01 , we get a hydrogen number density of @xmath155 . ]
@xmath156 where the extinction coefficient factor @xmath157 ( equal contributions from absorption and scattering ) .
.summary of the four cases , where we vary the inner edge of the dusty cloud @xmath47 , the uv - optical luminosity @xmath158 and duration @xmath159 of the tde source .
the lightcurve of the source is assumed to be flat .
the following parameters are fixed in all four cases : the radial thickness of the dusty cloud @xmath160 , initial grain radius @xmath151 ( all grains having the same initial radius ) and the total dust optical depth @xmath152 ( @xmath153 ) .
[ cols="^,^,^,^",options="header " , ] the bolometric lightcurves for all cases are shown in fig.([fig : lc ] ) , from which we draw the following conclusions : ( 1 ) the luminosities are determined by the total energy input @xmath161 and the light - travel delay , and are typically @xmath162 , consistent with the estimation in eq.([eq:9 ] ) ; ( 2 ) the lightcurves rise linearly until @xmath159 ( due to the increasing emitting area ) and when @xmath163 , the lightcurves are nearly flat ( where the projected surface area is slightly smaller ) . ] . to illustrate the dust sublimation process , in fig.([fig : profile ] ) we show the grain temperature and radius profiles for case i ( red curve in fig . [
fig : lc ] ) at different retarded time @xmath164 .
we can see that the dust sublimation front is very sharp and moves outwards with time . at @xmath165 ,
the dust grains in the inner quarter ( radius - wise ) of the cloud have fully sublimed . in case i , the ir emission peaks at @xmath166 .
the temperature profile far beyond the sublimation front is exponential , as expected from the @xmath53 term in eq.([eq:1 ] ) .
we note that , at larger radii where the temperature drops to much below 1000 k , the absorption coefficient adopted in eq.([eq:6 ] ) may not be a good approximation .
however , since most ir emission comes from the inner - most shells where the grain temperatures are high , our results are not affected .
as shown in section [ sec : estimation ] and [ sec : rt ] , for @xmath167 ( the fraction of uv - optical radiation from the tde intercepted by dust in the central few parsecs ) , the dust ir emission has a typical bolometric luminosity of @xmath168 and peaks at @xmath169 ( @xmath170 ) . for a luminosity distance of @xmath171 ( redshift @xmath172 ) ,
the peak flux density is @xmath173 .
this is photometrically detectable by current generation of telescopes , including the stratospheric observatory for infrared astronomy ( sofia , with the flitecam and forcast instruments ) and the _ spitzer _ space telescope ( hereafter _ spitzer _ , with the irac instrument ) . in the near future ,
_ james webb _ space telescope ( jwst ) will be able to perform photometric and spectroscopic measurements .
we note that ground - based telescopes are much more sensitive at @xmath174 ( k band ) than at @xmath175 .
observations at @xmath176 by space telescopes are more useful at detecting dust ir emission due to the following two reasons : ( 1 ) at wavelengths close to the optical , the flux could be dominated by other bright sources such as the accretion disk ( as shown in fig .
[ fig : sed ] below ) or star light ; ( 2 ) since the extinction ratio @xmath177 @xcite , dust emission at @xmath178 may be significantly absorbed by outer dusty layers ( and re - radiated at longer wavelengths ) .
however , on one hand , the confusion by the disk emission could be ruled out by long - term monitoring , since the disk accretion rate is expected to drop fast with time while the luminosity from dust emission stays more - or - less constant for many years . on the other hand ,
a combination of high uv - optical luminosities ( @xmath179 ) and small inner edge radii ( @xmath47 ) can lead to relatively high grain temperatures , such as in case i ( red curve in fig .
[ fig : lc ] ) .
for example , the flux density ratio @xmath180 of a blackbody at @xmath181 is only 0.72 , so ground - based k - band observations may play an important role in studying dust ir emission . to the authors knowledge ,
ir observations at rest - frame wavelength @xmath182 have only been reported from two tde candidates , sdss j0952 + 2143 @xcite and _ swift _ j1644 + 57 @xcite .
below , we first show that the mid - ir flux from sdss j0952 + 2143 , although consistent with the dust emission from a tde , is likely permanent emission from dust - obscured starburst regions or an agn .
then , we also show that the mid - ir flux from _ swift _ j1644 + 57 may be produced by other radiation sources , such as the accretion disk or the external shock driven by the interaction between disk outflow and circum - nuclear medium .
therefore , we are not sure whether the dust ir emission calculated in this work has been detected . at last ,
we point out two tde candidates where the dust ir emission may be currently detectable .
sdss j0952 + 2143 ( redshift z = 0.079 ) was reported as a tde candidate due to its strong but transient emission lines ( of e.g. h , he , fe ) , which are interpreted as the light echo of a powerful euv - x - ray outburst ( not directly observed ) .
photometry from sloan digital sky survey showed a fading optical continuum since dec 2004 and the outburst is believed to have happened before this time .
near - ir ( j , h and k@xmath183-band ) images showed quiescent ( from 1998 to 2008 ) emission of @xmath184 dominated by the extended host galaxy .
a _ spitzer _
mid - ir ( @xmath185 ) spectrum obtained in jun 2008 showed two pronounced silicate peaks at @xmath13 and @xmath186 , which resemble the seds of many palomar - green quasars @xcite .
the integrated luminosity was @xmath187 .
the total amount of energy radiated in the observation window since the outburst is @xmath188 and the grain temperature can be estimated to be @xmath189 .
therefore , the mid - ir data is consistent with the scenario of emission from the circum - nuclear dust heated by a tde .
however , as pointed out by @xcite , the whole sed from optical to mid - ir could also be produced by dusty starburst regions .
in addition , an x - ray luminosity of @xmath190 may be due to a weak agn , which could also contribute some mid - ir flux .
another mid - ir observation is needed to tell whether the mid - ir emission is transient or permanent .
j1644 + 57 ( redshift @xmath191 ) is widely believed to be a special tde where a relativistic outflow is launched from the accretion disk and produces bright emission from @xmath7-ray to radio wavelengths ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this event received excellent observational coverages in the time domain ( a few to @xmath192 days ) and wavelength domain ( @xmath193 to @xmath194 ) . due to heavy host extinction ( @xmath195 ) ,
the uv and optical fluxes are strongly suppressed . for the purpose of this work , we pay attention to the _ spitzer _ @xmath196 ( rest - frame @xmath197 ) and @xmath198 ( rest - frame @xmath199 ) data and see if it originates from the dust ir emission . in fig.([fig
: mir_lc ] ) , we plot the @xmath196 and @xmath200 band lightcurves , with the flux from the host galaxy included and the host extinction is not corrected .
the latest data points ( at rest - frame @xmath201 ) are considered as host flux , although the lightcurves are not completely flat . in a similar way , the host flux in j , h , and k bands are obtained and reported by @xcite . in fig.([fig : sed ] ) , we plot the seds with the radio data from @xcite and the host - subtracted ir data from @xcite at rest - frame time @xmath202 and @xmath203 . on the high - frequency side ,
the @xmath196 and @xmath200 fluxes are consistent with being the rayleigh - jeans tail of a blackbody component ( especially in the lower panel at rest - frame @xmath204 ) .
as pointed out by @xcite , the blackbody component may originate from the accretion disk . on the low - frequency side ,
the radio emission has been extensively studied and is well explained by the external shock produced by the interaction between the disk outflow and circum - nuclear medium .
the details of the outflow are different in different models , but at @xmath205 days , the blast wave should more or less settle to the sedov - taylor evolution . in the standard external shock model ( see e.g. * ? ? ?
* ) , electrons are accelerated to a relativistic powerlaw distribution and then produce a broken powerlaw spectrum via synchrotron radiation . from fig.([fig : sed ] ) , we can see that , if there is no spectral break between the radio and ir , the @xmath196 and @xmath198 fluxes are consistent with the extrapolation of the powerlaw around @xmath206 . for an electron energy distribution of @xmath207 ,
a spectral break may be produced at the synchrotron cooling frequency @xmath208 , because @xmath209 when @xmath210 and @xmath211 when @xmath212 .
if the blast wave is sub - relativistic , the synchrotron cooling frequency is given by @xmath213 where @xmath214 is electron mass , @xmath215 is electron charge , @xmath216 is thomson scattering cross - section , @xmath60 is the strength of magnetic field , @xmath45 is the dynamical time . from eq.([eq:24 ] ) , we can see that , as long as @xmath217 , the synchrotron cooling break is expected to be above @xmath218 , and hence the @xmath196 and @xmath198 fluxes are consistent with being produced by the external shock .
the magnetic field energy density in the shocked region is usually assumed to be a fraction @xmath219 of the thermal energy density and hence depends on the circum - nuclear medium density and @xmath219 .
these two parameters are very uncertain , so a firm conclusion can not be drawn .
on the other hand , since the dust ir emission typically lasts for a few years , we encourage observations at @xmath220 on some recently - discovered tde candidates , especially the most nearby ones assasn-14ae @xcite and -14li @xcite .
the reported luminosities and total radiation energies above are respectively @xmath221 and @xmath222 ( former ) , @xmath223 and @xmath224 ( latter ) .
we note that , if the seds of these two events are reddened by host - galaxy extinction , the true luminosities and energies are higher . taking the reported @xmath225 and @xmath161 as conservative values and assuming the pc - scale dust covering factor @xmath226 , we expect the peak flux densities to be @xmath227 for assasn-14ae and @xmath228 for assasn-14li and peak wavelengths to be between 3 and 10 @xmath229 for both events . to confirm the nature of dust ir emission , other possibilities can be ruled out in the following way : ( 1 ) unlike _ swift _ j1644 + 57 , the two assasn tde candidates are not expected to have strong relativistic outflows and are hence radio quiet .
actually , @xcite recently reported weak radio emission from assasn-14li , which was interpreted as a weak outflow driving an external shock . however , extrapolation from the radio to ir gives a flux much smaller than the predicted dust emission , so there is no contamination from the external shock .
( 2 ) with a good multi - wavelength coverage from mid - ir to optical , other components ( such as the disk emission ) could be separated out .
( 3 ) the contribution from the host galaxies could be subtracted by multi - epoch or high spacial resolution observations . in future jwst mid - ir spectra , we can look for the smoking - gun evidences of dust , such as the silicate feature at @xmath230 or polycyclic aromatic hydrocarbon ( pah ) features .
\(1 ) if a significant amount of dust exists within @xmath231 , the refractory elements al , ca , si and fe , which normally condense into solids , will gradually return to the gas phase .
observing their abundances rising with time would be a direct evidence of dust sublimation .
the abundances could be measured through absorption lines .
\(2 ) if the dusty clouds lie along the line of sight , as dust grains sublime , the time evolution of dust extinction may be observable with sufficient uv and optical wavelength coverages .
\(3 ) for simplicity , we have been assuming the source lightcurve to be flat , but tdes actually have uv and optical variability down to at least timescales of days .
therefore , it may be possible to measure the time dependent dust ir emission in response to the variable uv and optical continuum .
this so - called `` dust reverberation mapping '' method has been successfully used to determine the sublimation radius of the dusty torus in agns since the work by @xcite . in this subsection , we discuss some potential issues that may affect the details of the dust ir emission ( but the general result in this work is unaffected ) .
\(1 ) to solve the grain temperature , we have assumed instantaneous thermal equilibrium in eq.([eq:1 ] ) .
this is valid because the heat capacity of a single grain is small , and hence the cooling time is short .
regardless of the composition , the upper limit of the total internal energy of a grain is @xmath232 . at a temperature of @xmath233 ,
the radiative cooling rate is @xmath234 .
therefore , the cooling time is @xmath235 , which means that dust grains reach thermal equilibrium nearly instantaneously .
\(2 ) we have ignored dust destruction processes other than thermal sublimation . as mentioned by @xcite in the grb context , photo - ionization of dust grains could lead to significant grain charging ( positive ) .
the electric field at the surface can be so large that the grain fragments ( `` coulomb explosion '' ) or emits individual ions ( `` ion field emission '' ) , depending on how strong the grain structure is .
if the grain charge is @xmath236 ( in electron units ) , the electrostatic binding energy of an electron at the grain surface is @xmath237 ( @xmath215 being the electron charge ) .
photo - electric charging stops when the electrostatic binding energy exceeds the incoming photon s energy . when exposed to the intense x - ray radiation from grbs , the grain charge reaches @xmath238 @xcite ; but in the context of tdes where the radiation energy is in uv and optical , the grain charge @xmath239 is smaller than @xmath240 .
the surface electric field is @xmath241 , corresponding to a stress @xmath242 .
ion field emission happens when the electric field approaches @xmath243 ( to break a typical @xmath244 molecular bond across a distance of @xmath245 ) .
coulomb explosion happens when the electrostatic stress is greater than the tensile strength of grains , which is uncertain . in the grb literature ( e.g. *
* ) , a typical grain tensile strength @xmath246 is used .
therefore , once the grain charge reaches @xmath238 , both coulomb explosion and ion field emission can happen .
however , neither of these two grain destruction mechanisms are important in the tde context because the grain charge is too small . even for relativistic tdes like _ swift
_ j1644 + 57 with intense x - ray emission , since the x - rays are relativistically beamed into a narrow angle , most dusty clouds only receive the more - or - less isotropic uv - optical radiation .
\(3 ) a few other complexities , which are ignored in our simple model , can be taken into account in future works .
first , the heating by the scattered uv - optical photons and dust ir radiation could be included by considering the time delay and related absorption coefficient . in this way ,
a spectrum of dust ir emission can be calculated and compared with future observations .
second , the clumpiness of dusty clouds will affect the radiative transfer .
since every clump will have a bright ( hot ) side and a dark ( cold ) side , dust temperature distributions are different in clumpy and smooth environments .
this is considered in modeling the ir emission from the agn torus @xcite .
last , in the radiative transfer model in section [ sec : rt ] , a single dust radius of @xmath247 is used , but the true interstellar grain population has a broad size distribution , extending from a few @xmath248 ( pah molecules ) to about @xmath249 ( or even larger ) .
most mass is in the larger grains , with half - mass grain radius typically @xmath250 ( half of mass is in grains with radius above this value ) , and the smaller grains contribute most surface area and hence most uv extinction ( e.g. * ? ? ? * ; * ? ? ?
however , at the same distance from the central source , smaller grains sublime faster than larger ones ( mostly due to higher grain temperatures ) , so it is unclear which grain size contributes most ir emission .
a systematic study of grain size distribution and chemical composition could be done in the future with dust continuum radiative transfer codes such as dustymoshe / dusty/ ] ] and hyperion , although the dynamical propagation of radiation coupled with dust sublimation requires some modification of the codes .
we consider the impact of the uv - optical radiation ( @xmath251 ) from tdes on the circum - nuclear medium . if the medium is dusty , the uv - optical radiation energy will be absorbed by dust within @xmath252 from the bh and re - radiated in the ir .
we calculate the bolometric lightcurve from a 1-d radiative transfer model , taking into account the time - dependent heating , cooling and sublimation of dust grains .
we show that the dust emission peaks at @xmath2 and has typical luminosities between @xmath3 and @xmath4 ( with sky covering factor @xmath253 ranging from 0.1 to 1 ) .
this is detectable by current generation of telescopes , including the stratospheric observatory for infrared astronomy ( sofia ) and the _ spitzer _ space telescope . in the near future ,
_ james webb _ space telescope ( jwst ) will be able to perform photometric and spectroscopic measurements . in future jwst mid - ir spectra
, we can look for the smoking - gun evidence of dust , such as the silicate or polycyclic aromatic hydrocarbon ( pah ) features .
observations at rest - frame wavelength @xmath254 have only been reported from two tde candidates , sdss j0952 + 2143 @xcite and _ swift _ j1644 + 57 @xcite .
the _ spitzer _ @xmath255 flux of sdss j0952 + 2143 , although consistent with the dust emission from a tde , is likely permanent emission from dusty starburst regions or an agn .
we also show that the _ spitzer _ @xmath196 and @xmath256 fluxes of _ swift _ j1644 + 57 may originate from other sources than the dust ir emission , such as the accretion disk or the external shock ( driven by the interaction between disk outflow and circum - nuclear medium ) .
long - term monitoring is need to tell whether the mid - ir emission from the two event is transient or permanent .
currently , a firm conclusion can not be drawn .
at last , we show that the two nearby tde candidates assasn-14ae and -14li are good candidates to search for the dust ir emission .
if the pc - scale dusty clouds have a sky covering factor of @xmath226 , the conservatively estimated fluxes for the two events are , respectively , @xmath257 and @xmath258 at @xmath259 .
agns are known to be surrounded by a dusty torus that provides obscuration and ir emission .
however , little is known about the dust content in the pc - scale vicinity of weakly- or non - active galactic nuclei .
currently , the only observable example is our galactic center , where the dusty clouds has sky covering factor @xmath260 . the dust ir emission can give a snapshot of the pc - scale dust content around weak- or non - active galactic nuclei , which is hard to probe otherwise .
we acknowledge helpful discussions on the galactic center environment with j. lacy .
this research was funded by a graduate fellowship ( named `` continuing fellowship '' ) at the university of texas at austin . n.j.e .
acknowledges support from a grant from the national science foundation , ast-1109116 . | recent uv - optical surveys have been successful in finding tidal disruption events ( tdes ) , in which a star is tidally disrupted by a supermassive black hole ( bh ) .
these tdes release a huge amount of radiation energy @xmath0 into the circum - nuclear medium .
if the medium is dusty , most of the radiation energy will be absorbed by dust grains within @xmath1 from the bh and re - radiated in the infrared .
we calculate the dust emission lightcurve from a 1-d radiative transfer model , taking into account the time - dependent heating , cooling and sublimation of dust grains .
we show that the dust emission peaks at @xmath2 and has typical luminosities between @xmath3 and @xmath4 ( with sky covering factor of dusty clouds ranging from 0.1 to 1 ) .
this is detectable by current generation of telescopes . in the near future ,
_ james webb _ space telescope will be able to perform photometric and spectroscopic measurements , in which silicate or polycyclic aromatic hydrocarbon ( pah ) features may be found .
observations at rest - frame wavelength @xmath5 have only been reported from two tde candidates , sdss j0952 + 2143 and _ swift _ j1644 + 57 .
although consistent with the dust emission from tdes , the mid - infrared fluxes of the two events may be from other sources .
long - term monitoring is needed to draw a firm conclusion .
we also point out two nearby tde candidates ( assasn-14ae and -14li ) where the dust emission may be currently detectable .
the dust infrared emission can give a snapshot of the pc - scale dust content around weakly- or non - active galactic nuclei , which is hard to probe otherwise .
[ firstpage ] galaxies : nuclei ism : dust infrared : ism methods : analytical |
the interplay of quantum mechanical interference and electron - electron interaction in semiconductor quantum dots ( qds ) has attracted a lot of experimental and theoretical interest in recent years.@xcite experimentally the ground state properties of a qd can be probed by coulomb blockade ( cb ) measurements at near - zero temperature , in which the conductance through the dot is blocked except at particular gate voltages , causing the conductance as a function of gate voltage to be a series of sharp peaks . in particular , the spacing between neighboring peaks is related to the second difference of the ground state energy with respect to electron number @xmath2 , @xmath3 , which is often called the addition energy .
the ground state spin of the dot can be inferred from the motion of the peak position in a magnetic field.@xcite except in small dots with high geometrical symmetry,@xcite whose ground state properties show shell structure following hund s rules , most experimental qds are asymmetric due to either the irregular shape of the confining gate or impurity disorder , and the number of electrons is usually several hundred.@xcite the properties of qds show apparently random but reproducible fluctuations as some external parameters such as the chemical potential , the shape of the confining potential , external magnetic field and so on , vary . in this case , only statistical properties are of physical significance .
while the statistical behavior of peak height fluctuations can be well described by a non - interacting random matrix theory ( rmt ) model,@xcite a similar model , the constant interaction ( ci ) plus rmt , fails dramatically in describing the peak spacing distribution .
the importance of e - e interaction was well recognized , but different theoretical approaches differ quantitatively or even qualitatively in many aspects of e - e interactions effects . the numerical studies for small dots@xcite yield a gaussian - like peak spacing distribution without any even / odd pairing effect , which agrees with experimental results on much larger dots .
on the other hand , the statistical approaches for large dots@xcite predict a much more pronounced even / odd effect at zero temperature that is absent in experimental results.@xcite besides conductance peak spacings , another quantity that has attracted a lot of interest is the ground state spin in the cb regime . the occurrence of non - minimal ground state spin has been suggested as an explanation for the absence of a bi - modal distribution in conductance peak spacings,@xcite and for the kinks in the parametric motion of the peaks.@xcite a recent experimental study on the change in cb peak position as a function of magnetic filed by folk , et al.@xcite found that three out of five even-@xmath4 states had @xmath5 . as in the case of peak spacing distributions ,
previous theoretical studies on ground state spin distributions in qds gave very different results : simulations on small tight - binding lattice models with random site energies predict a dominant fraction of high spin ground states,@xcite but results from ks - sdft studies of small disordered parabolic quantum dots@xcite showed that the probability of high - spin ground states never exceeds @xmath6 even in the strong interaction regime . despite intensive efforts to understand the role of e - e interaction in qd systems , there are still many questions that require further investigation .
in this paper , we focus on the following issues : ( 1 ) the first one is concerned with the evolution of statistical properties of qds as the interaction strength increases .
the e - e interaction strength is usually characterized by the wigner - seitz radius @xmath7 , which is defined in 2d as @xmath8 , where @xmath9 is the electron density and @xmath10 is bohr radius.@xcite previous theoretical studies gave different answers to this question , indicating the necessity of further theoretical exploration . in the meantime , this issue has attracted the experimental interest , and some relevant experiments are underway .
( 2 ) the second issue to be explored is the effect of confining potential .
previous theoretical studies were based on abstract statistical assumptions , chaotic model external potentials , or gaussian impurities .
none of them have considered the realistic confining potential imposed in actual experiments . though it is usually assumed that the statistical properties of qds are of universal significance
, the validity of that is still to be verified . (
3 ) the third issue that is worthy of further elaboration is how statistical properties of qds evolve as the size of a qd system , mainly the electron number , increases , which is important not only in solving the puzzling discrepancy between numerical small - dot calculations and semi - analytic large - dot statistical analysis , but also in understanding the fundamental question in mesoscopic physics of how macroscopic properties emerge from microscopic physics via the mesoscopic regime . in our previous study,@xcite
we reported a kohn - sham spin density functional theory ( ks - sdft ) study of spin and conductance peak spacing distributions in 2d isolated quantum dots in a classically chaotic quartic oscillator potential with electron number @xmath2 up to 200 at @xmath11 .
we found that for large electron number and asymmetric external potential , the peak spacing distribution is mainly gaussian - like and there is no observable even / odd effects .
we also found enhanced probability of high spin ground state implying stronger interaction effects than expected from rpa approaches . in this paper , we extend the previous study in two directions : ( 1 ) similar calculations are done for weaker ( @xmath12 ) interactions to investigate how statistical properties of quantum dots evolve as the interaction strength varies .
( 2 ) we extended previous studies in isolated qds , in which electrons are confined by an analytic chaotic potential without explicit consideration of the coupling with external gate , to qd systems in which the confining potential is calculated by solving the electrostatic poisson equation with realistic boundary conditions . in the latter case , with more controllable parameters available , we are able to study statistical properties of qds as a function of both e - e interaction strength ( @xmath7 ) and electron number ( @xmath2 ) in a more systematic way .
the outline of the paper is as follows : in the next section , the ks - sdft method and the numerical techniques used in our study are concisely described . in section iii ,
we report the statistical properties of isolated quantum dots . in section iv ,
the results for realistic quantum dots are presented .
section v concludes the paper with discussion about the significance of these results .
in ks - sdft , the ground state energy of an interacting system with electron number @xmath2 and total spin @xmath13 in the local external potential @xmath14 is written as a functional of spin densities @xmath15 with @xmath16 denoting spin - up and spin - down , respectively , @xmath17 = t_{s}\left [ n^{\alpha } , n^{\beta } \right ] + \int n({\mathbf{r}})v_\mathrm{ext}({\mathbf{r}})d{\mathbf{r } } & & \nonumber
\\ + \frac{1}{2}\int \frac{n({\mathbf{r}})n({\mathbf{r}}')}{\left| { \mathbf{r}}-{\mathbf{r}}'\right| } d{\mathbf{r}}d{\mathbf{r } } ' + e_{\mathrm{xc}}\left [ n^{\alpha } , n^{\beta } \right ] & .
\label{eq : e0}\end{aligned}\ ] ] ( effective atomic units are used through the paper : for gaas qds with the dielectric constant @xmath18 and effective electron mass @xmath19 , the values are 10.08 mev for energy and 10.95 nm for length . )
@xmath20 $ ] is the kinetic energy of the ks non - interacting reference system which has the same ground state spin density as the interacting one , and @xmath21 $ ] is the exchange - correlation energy functional . the spin densities @xmath22 satisfy the constraint @xmath23 with @xmath24 and @xmath25 . assuming that the ground state of the non - interacting reference system is non - degenerate , the non - interacting kinetic energy is given by @xmath26=\sum _
{ i,\sigma } \left\langle \psi _ { i}^{\sigma } \left| -\frac{1}{2}\nabla ^{2}\right| \psi _ { i}^{\sigma } \right\rangle $ ] , and the ground state spin density is uniquely expressed as @xmath27 here @xmath28 are the lowest single - particle orbitals which are obtained from @xmath29 + v_{\mathrm{xc}}^{\sigma } [ n^{\alpha } , n^{\beta } ; { \mathbf{r}}]\big\rbrace \psi _ { i}^{\sigma } ( { \mathbf{r}})=\varepsilon _ { i}^{\sigma } \psi _ { i}^{\sigma } ( { \mathbf{r}}),\ ] ] where @xmath30 $ ] and @xmath31 $ ] are the hartree and exchange - correlation potentials , respectively , @xmath32 & \equiv & \int \frac{n({\mathbf{r}}')}{|{\mathbf{r}}-{\mathbf{r}}'|}d^{3}{\mathbf{r } } ' , \\
v_{\mathrm{xc}}^\sigma[n^{\alpha } , n^{\beta } ; { \mathbf{r } } ] & \equiv & \frac{\delta e_\mathrm{xc } \left [ n^{\alpha } , n^{\beta } \right ] } { \delta n^{\sigma}({\mathbf{r}})}.\end{aligned}\ ] ] we have used the local spin - density approximation ( lsda)@xcite for the exchange - correlation energy functional , @xmath33\approx \int n({\mathbf{r}})\epsilon_\textrm{xc } \big(n({\mathbf{r}}),\zeta({\mathbf{r}})\big)d{\mathbf{r}},\ ] ] where @xmath34 is the spin polarization , and @xmath35 is the exchange - correlation energy per electron in the homogeneous electron gas .
we use tanatar - ceperley s parameterized form of @xmath35 for the spin - compensated case ( @xmath36 ) and fully spin polarized case ( @xmath37).@xcite @xmath35 for arbitrary spin polarization is obtained by the following interpolation,@xcite @xmath38,\ ] ] with @xmath39 which is exact for the exchange energy density , but only approximate for the correlation energy density .
recently , attaccalite , et al.@xcite parameterized a new lsda exchange - correlation functional form based on more accurate quantum monte carlo modeling data that includes spin polarization explicitly .
the new functional form was shown to be more accurate than tanatar - ceperley s in both zero and finite magnetic field when compared to quantum monte carlo data.@xcite numerical tests undertaken by us showed , however , that in the regime with which this paper is concerned , the difference between the two forms is minor . in this paper
, we will use tanatar - ceperley s form .
comparisons with quantum monte - carlo calculations for @xmath40 and interaction strength parameter @xmath41 have shown that lsda works well for both the ground state spin and energy in 2d parabolic quantum dots,@xcite as well as for many spin excitation energies.@xcite ( color online ) upper panel : contours of @xmath42 in qop system with parameters @xmath43 , @xmath44 , @xmath45 , and @xmath46 .
lower panel : electron density at @xmath47 and @xmath48 .
it is obvious that the electron density in the isolated dot is far from uniform even within the classically allowed area ; in the case shown here , the maximal density is 110% larger than that in the center.,title="fig:",width=192 ] ( color online ) upper panel : contours of @xmath42 in qop system with parameters @xmath43 , @xmath44 , @xmath45 , and @xmath46 .
lower panel : electron density at @xmath47 and @xmath48 .
it is obvious that the electron density in the isolated dot is far from uniform even within the classically allowed area ; in the case shown here , the maximal density is 110% larger than that in the center.,title="fig:",width=316 ] ( color online ) illustration of data analysis procedure . ( a )
the addition energy @xmath49 as a function of @xmath4 ( dotted ) , and its @xmath50 order polynomial fit that is related to increasing classical capacitance as the number of electron in the dot increases ( solid line ) .
( b ) the fluctuation of addition energy after removing the smooth trend .
( c ) mean - level - spacing - scaled fluctuation of addition energy , whose statistical properties are to be studied .
the data is from the qop system with parameters @xmath43 , @xmath44 , @xmath45 , and @xmath46 , width=307 ] . to investigate large electron number regime
, we developed efficient techniques for the implementation of ks - sdft method which have been reported elsewhere.@xcite essentially our algorithm includes the following components : ( 1 ) we use a particle - in - the - box basis for the representation of ks orbitals and for the action of the kinetic energy operator on the orbital wave functions ; ( 2 ) the kohn - sham total energy is directly minimized by a conjugate gradient method that is based on teter , payne and allan s method@xcite but with important modifications , which include a more efficient line minimizaiton scheme and a delayed update of the effective potential;@xcite ( 3 ) we use a fourier convolution method to calculate hartree potential efficiently;@xcite ( 4 ) a simple one - way multi - grid technique is used to accelerate the convergence.@xcite the high efficiency of our method makes the numerical study of statistical properties of large quantum dots with several hundred electrons become computationally feasible.@xcite
we first investigate statistical properties of quantum dots in a 2d isolated model system .
since experimental dots usually have irregular shapes that result in chaotic classical dynamics , we use the following quartic oscillator potential ( qop ) as the external confining potential , @xmath51 both the classical dynamics and the single - particle quantum mechanics at @xmath52 have been extensively studied:@xcite the system evolves continuously from integrable to fully chaotic as @xmath53 changes from 0 to 1 . the parameter @xmath54 is introduced to break the four - fold symmetry .
the prefactor @xmath55 is used to adjust the flatness of the confining potential and hence the electron density .
[ fig : qop ] shows the contours of a typical @xmath42 used in this study ( upper panel ) , and the corresponding electron density at @xmath56 ( lower panel ) .
a noticeable feature is that the electron density in isolated qd systems is not uniform : electrons form peaks near the boundary region because of classical coulomb repulsion between electrons . in real experimental systems , because of the coupling with the top gate , we expect to see an essentially uniform electron density distribution .
whether such a difference in electron density will result in significant deviation in statistical properties will be discussed in the next section . for a given @xmath42 , the ground state energy @xmath57 and spin @xmath58 as a function of @xmath2
are determined by calculating several spin configurations for each @xmath2 and selecting the one with minimal energy .
the addition energy is then calculated as the second difference of @xmath59 . as shown in fig .
[ fig : qop - data](a ) , @xmath60 decreases slowly as @xmath2 increases .
this is mainly a classical effect the increasing capacitance as the dot becomes large .
since we are mainly interested in the quantum mechanical effects , the classical smooth trend , denoted @xmath61 , is fit by a 4-th order polynomial of @xmath2 and then subtracted from @xmath60 . to compare with experiments ,
the addition energy is further scaled by the mean level spacing , @xmath62 , which is calculated by @xmath63 with @xmath64 estimated as the classical allowed area ; the resulting dimensionless spacing is denoted @xmath65 , @xmath66 to obtain good statistics , we choose five sets of ( @xmath67 ) ( 0.53 , 0.2 ) , ( 0.565 , 0.2 ) , ( 0.6,0.1 ) , ( 0.635 , 0.15 ) , and ( 0.67,0.1 ) and for each set of parameters , we calculate a range of electron numbers @xmath2 from 1 to 200 .
[ fig : qop - distr ] plots peak spacing distributions in isolated qds in different interaction regimes ; @xmath7 ranges from about @xmath68 to @xmath69 , and @xmath70-@xmath71 . in the large density or weak interaction regime , @xmath12 [ fig .
[ fig : qop - distr](a ) ] , the peak spacing distributions for even and odd @xmath2 are sharply different : the distribution for even @xmath2 is asymmetric and close to the wigner surmise while that for odd @xmath2 is a narrow sharp symmetric peak , which agrees very well with the ci - rmt model . as the interaction strength increases , such an even / odd effect is gradually reduced : the odd @xmath4 distribution becomes increasingly wider , and the even @xmath4 distribution becomes more symmetric . at @xmath72 [ fig .
[ fig : qop - distr](c ) ] , the only difference between the two is a weak tail in the even @xmath4 distribution at positive spacing . as @xmath7 is further increased to about @xmath69 [ fig . [
fig : qop - distr](d ) ] , which is the usual experimental regime , the even / odd effect is hardly perceptible , and the shape of the distribution is mainly gaussian - like .
the insets in fig .
[ fig : qop - distr ] show ground state spin distributions in the corresponding interaction regimes . at @xmath73 ,
the minimal spin configuration dominates in both even and odd @xmath2 cases , as one expects according to ci - rmt model .
but even in this weak interaction regime , the portion of @xmath74 for even @xmath4 is already significant , @xmath75 . the probability of high - spin ground state increases quickly as the e - e interaction becomes stronger . at @xmath76 ,
the high spin ground state is more probable than the minimal spin state.@xcite such a result is remarkable considering that @xmath77 is only a moderate interaction strength .
the boundary condition used in isolated quantum dots is very different from that in the experimental systems .
[ fig : rdot ] illustrates a typical experimental quantum dot structure , which is fabricated based on a gaas - algaas heterostructure .
numerical modeling of realistic quantum dots@xcite is still a great challenge in terms of computational efforts .
the difficulty mainly comes from the fact that although the volume where electrons are confined is small , the boundary conditions that couple the dot with the external confining gate are defined on a much larger scale .
previous numerical studies of realistic quantum dots considered only small dots with less than a few tens of electrons , except in ref . where by reducing schrdinger equations from 3d to multi - component 2d the author was able to study realistically a lateral quantum dot with @xmath2 up to 100 .
( color online ) schematic illustration of a realistic quantum dot studied in this paper .
the dot is formed based on a gaas - algaas heterostructure , which consists of ( from the bottom ) an undoped gaas substrate , an undoped algaas spacer layer , a n - doped algaas layer , and finally a gaas cap layer .
the widths for the spacer , doped and cap layers are denoted as @xmath78 , @xmath79 and @xmath80 , respectively . in this paper , @xmath81 , @xmath82 and @xmath83 are taken . , width=307 ] illustration of the shape of top confining gate used in this study .
a negative voltage is imposed on the shaded region , depleting the electrons underneath so that the motion of electrons is confined to a small region . to obtain enough statistics , 20 different irregular confining shapes are generated by changing @xmath84 , @xmath85 , @xmath86 and @xmath87 randomly .
, width=240 ] ( color online ) upper panel : contours of a sample @xmath88 . with the shape of the confining gate shown in fig .
[ fig : rdot - shape ] , the confining potential is very irregular ; its classical dynamics is believed to be chaotic .
lower panel : electron density at @xmath70 .
it is much more uniform that that in the isolated qop shown in fig .
[ fig : qop],title="fig:",width=192 ] ( color online ) upper panel : contours of a sample @xmath88 . with the shape of the confining gate shown in fig .
[ fig : rdot - shape ] , the confining potential is very irregular ; its classical dynamics is believed to be chaotic .
lower panel : electron density at @xmath70 .
it is much more uniform that that in the isolated qop shown in fig .
[ fig : qop],title="fig:",width=316 ] we note that the most important factor in experimental conditions that is absent in the isolated quantum dot model is the coupling between the dot and the top gate .
to take that into account , we introduce semi - infinite boundary conditions.@xcite using @xmath89 to denote the 3d spatial coordinate , @xmath89 where @xmath90 is the height in the growth direction and @xmath91 the coordinates perpendicular to the growth direction , we take @xmath92 where @xmath93 is the schottky barrier due to surface states,@xcite and @xmath94 is the gate voltage imposed on the top surface . in 2d cases , the inclusion of semi - infinite realistic boundary conditions is quite straightforward . for @xmath42 in eq .
( [ eq : ks ] ) , instead of an analytic form , it is calculated from @xmath95 where @xmath96 is the confining potential in @xmath97-direction due to the quantum well or heterojunction structure from which the quantum dot is fabricated . in the case of 2d quantum dots , @xmath97 is taken as the distance between the top surface and the mean electron height near the middle of the qd , @xmath98 . because of the coupling with the gate , the hartree potential has an additional image term,@xcite @xmath99 \,.\ ] ] this 2d quantum dot model will be denoted r2d to simplify the notation . besides the more realistic boundary conditions included in the r2d model
, its other advantage is that we can obtain statistics by changing the shape of the confining gate.@xcite in this study , we use a confining gate with the shape illustrated in fig . [
fig : rdot - shape ] . by taking @xmath84 , @xmath85 , @xmath86 , and @xmath87 randomly
, we get a series of irregular confining potentials that are statistically independent as indicated by correlation analysis . fig .
[ fig : rdot - sample ] shows the contour of a typical confining potential and a sample electron density for @xmath56 .
we see that indeed the electron density in r2d is much more uniform than that in the isolated case .
the availability of a lot of statistically independent irregular potentials enables us to investigate how spin and peak spacing distributions evolve as a function of @xmath2 and @xmath7 in a more systematic way .
( color online ) comparison of qop and r2d models : distributions of conductance peak spacing ( solid for even @xmath2 , dashed for odd @xmath2 ) and ground state spin ( insets ) in ( a ) qop at @xmath100 ; ( b ) r2d at @xmath101 ; ( c ) qop at @xmath102 ; ( d ) r2d at @xmath103 .
for the qop results @xmath104-@xmath105 , while for the r2d model the additional gate averaging allows a smaller range @xmath106-@xmath107 .
results for the two types of dots are in qualitative agreement .
, width=323 ] fig .
[ fig : qop - r2d ] shows spin and peak spacing distributions from isolated and realistic qd systems at two @xmath7 values ( @xmath108 and @xmath109 ) within comparable electron number ranges .
qualitatively , they have similar features : at @xmath110 , distributions from both qop and r2d have a weak tail on the positive side for even @xmath111 compared to odd @xmath2 , and at @xmath11 , the tail vanishes .
the spin distributions from two types of model systems are also the same within statistical uncertainty .
quantitatively there are indeed some observable differences in both spin and peak spacing distributions , but with the limited statistics available in this study , it is difficult to attribute physical significance to these subtle features .
our study therefore verifies the validity of using model systems to study statistical properties of quantum dots , and provides a direct demonstration of the universal feature of mesoscopic fluctuations .
( color online ) probabilities of different spin states as a function of @xmath7 for @xmath112-@xmath113 in realistic quantum dots ( circles : @xmath114 ; squares : @xmath115 ; diamonds : @xmath116 ; up triangles : @xmath117 ; down triangles : @xmath118 ) .
the error bars shown in the plot are calculated from the standard bootstrap method .
the ground state spin probabilities saturate at large @xmath7 . , width=316 ] ( color online ) probabilities of different spin states for different @xmath119 ranges at @xmath108 ( circles : @xmath114 ; squares : @xmath115 ; up triangles : @xmath117 ; down triangles : @xmath118 ) .
the error bars shown in the plot are calculated from the standard bootstrap method .
note the substantial change from small to large dots.,width=316 ] fig .
[ fig : spin - rs ] shows the spin distribution as a function of the interaction strength in the electron number range @xmath120-@xmath121 .
this study extends previous investigations for weak and moderate interaction strength in isolated dots to the stronger interaction regime . in agreement with previous studies ,
the fraction of high - spin ground states increases as the electron - electron interaction becomes stronger .
there are several features that are worth comment .
( i ) _ crossing : _ for even @xmath2 , the probability of @xmath114 crosses with that of @xmath115 at @xmath122 .
( ii ) _ saturation : _ probabilities of ground state spins for even @xmath2 saturate at about @xmath123 .
the spin distribution for odd @xmath2 is not saturated in the interaction strength range studied here , but the trend indicates that saturation is likely .
( iii ) _ higher spin states : _ the probabilities of @xmath116 and @xmath124 remain small for all interaction strengths , though @xmath125 becomes significant at @xmath126 . to investigate how statistical properties of quantum dots evolve as a function of electron number @xmath4 , we calculated peak spacing and spin distributions in different electron number ranges for @xmath2 up to 200 at @xmath108 fig .
[ fig : eadistr - ne ] shows peak spacing distributions in different @xmath2 ranges . for @xmath112-@xmath113 ,
the distributions for even and odd @xmath2 are obviously different .
the difference decreases as @xmath2 increases , and becomes barely observable for @xmath127-@xmath105 and @xmath128-@xmath71 .
our results show clearly that the conductance peak spacing distribution depends on electron number , and is significantly different between small ( @xmath129 ) and large ( @xmath130 ) dots .
[ fig : spin - ne ] shows the trend in ground state spin as @xmath4 increases . for this quantity ,
the dependence on @xmath2 is even more significant : the probability of high spin ground states increases as @xmath2 increases . in particular
, @xmath131 increases from 0.27 at @xmath112-@xmath113 to 0.45 at @xmath128-@xmath71 .
another notable feature is that the spin distribution saturates at large @xmath2 .
ks - sdft calculations reported above reveal some interesting features in conductance peak spacing and ground state spin distribution . to obtain a clearer picture of the underlying physics , and , in particular , to understand the main factors that determine the spin distribution in different regimes , we use a simple two - level model ( tlm ) to investigate the probability of singlet and triplet ground state spin for even @xmath2 cases .
the same model was used as an interpretive tool in hirose and wingreens@xcite work on spin distributions in small disordered parabolic quantum dots , and in doing that , they attribute saturation of the spin distribution to the effects of off - diagonal interaction matrix elements .
our analysis using single - particle wave functions in truly ballistic chaotic systems , however , leads to very different conclusions .
the basic idea of the two - level model is to consider a two - electron system with only two single - particle levels .
the interacting matrix elements ( ime ) @xmath132 are calculated using a screened interaction kernel @xmath133 to implicitly account for the effects of all other electrons . in particular
, we use the screened interaction obtained from the 2d random - phase approximation ( rpa ) , @xmath134 where @xmath135 is the fourier transform of the bare coulomb interaction , and @xmath136 is the susceptibility of the 2d electron gas .
it is straightforward to include the screening due to the top gate by using @xmath137 , but in the regime considered in this paper , the effects of the gate screening on the interacting matrix elements are negligible . in this tlm system
, one can obtain singlet and triplet state energies either by exact diagonalization ( ed ) or in the framework of hartree - fock ( hf ) theory ( ie . , first - order perturbation theory ) . in tlm - hf , the difference of energy between singlet and triplet states has the following simple form , @xmath138,\label{eq : tlm - hf}\ ] ] where @xmath139 is the single particle level spacing , @xmath140 are the direct coulomb interaction terms , and @xmath141 is the exchange term .
( color online ) two - level model analysis of spin distribution for @xmath112-@xmath113 as a function of interaction strength .
( a ) probability of triplet ground state as a function of @xmath7 from tlm - ed ( circles ) , tlm - hf ( down triangles ) and ks ( up triangles ) .
( b ) mean of diagonal interaction matrix elements , @xmath142 , @xmath143 , and @xmath144 , as well as the mean of the off - diagonal matrix element @xmath145 , all in units of @xmath62.,width=316 ] ( color online ) two - level model ( tlm ) analysis of spin distribution as a function of @xmath2 .
( a ) probability of triplet ground state as a function of @xmath2 from tlm - ed ( circles ) , tlm - hf ( down triangles ) and ks ( up triangles ) .
( b ) means of diagonal interaction matrix elements , @xmath142 , @xmath143 , and @xmath144 , as well as that of the off - diagonal matrix element @xmath145 , all in units of @xmath62.,width=316 ] the main difference between our tlm analysis and hirose and wingreen s is in the way to choose single - particle energies and wave functions . in our analysis , we use single - particle eigen - energies and wavefunctions in the extended thomas - fermi effective potential@xcite that is obtained within the same realistic confining gate structure as in ks - sdft calculations .
wave functions obtained in this way are guaranteed to be those of ballistic systems , and possible complications related with the disorder strength ( ie . ,
localization ) can therefore be avoided .
we note that the two - level model with this way of calculating interacting matrix elements can be regarded as a simplified version of the model proposed by ullmo , et al . in ref . .
[ fig : tlm - rs](a ) shows the probability of finding triplet ground states , @xmath131 , as a function of @xmath7 obtained from tlm - ed , tlm - hf , as well as ks - sdft calculations .
we see a striking agreement between tlm - ed and ks - sdft results , which is indeed remarkable considering the extreme simplicity of our two - level model compared to ks - sdft .
both tlm - hf and ed reproduce correctly the saturation of spin distribution at @xmath146 . to explore possible mechanisms for the saturation , fig . [
fig : tlm - rs](b ) plots the mean values of @xmath142 , @xmath143 , @xmath144 , and the off - diagonal term @xmath145 , ( the other off - diagonal term , @xmath147 , is close to @xmath145 , and is therefore not plotted here ) as a function of @xmath7 . it is clearly shown that off - diagonal terms are an order of magnitude smaller than the coulomb and exchange terms , indicating that the off - diagonal terms are negligible in terms of their contribution to the spin distribution .
this is verified by the general agreement between tlm - hf and tlm - ed . on the other hand ,
[ fig : tlm - rs](b ) shows that though @xmath148 and @xmath149 do saturate at high @xmath7 , @xmath150 does not .
this indicates that considering mean values of @xmath142 , @xmath143 , and @xmath144 alone can not explain the saturation in the spin distribution .
we also use the tlm to interpret the effect of finite size on the spin distributions , and the results are plotted in fig .
[ fig : tlm - ne ] .
the simple two - level model also grasps most of the physics : the probability of the triplet ground state calculated from both tlm - ed and hf increases as @xmath2 increases , but saturates at large @xmath2 , in agreement with the ks results .
we note that the difference between different @xmath2 ranges is mainly due to wave functions rather than eigenvalues ( which are always random - matrix like for our system ) ; for a @xmath2-electron system , the wave functions involved in the tlm are @xmath151 and @xmath152 .
our tlm results show clearly that the @xmath2-dependence of the spin distributions comes mainly from the changing statistics of the wave functions .
this is more directly demonstrated in fig .
[ fig : tlm - ne](b ) , where mean values of @xmath142 , @xmath143 , @xmath144 and @xmath145 are plotted .
we note that the @xmath2-dependence of wave - function statistics is usually neglected in most statistical theories of quantum dots , where the large @xmath2 limit is usually taken.@xcite since many experimental data were obtained with @xmath2 less than 200 , our studies show clearly the necessity of considering finite @xmath2 effects in interpreting those data .
in this paper , we use kohn - sham spin density functional theory to study electron - electron interaction effects in quantum dots . in particular , we calculated conductance peak spacing and ground state spin distributions in an isolated quantum dot with classically chaotic quartic external potential and in a realistic quantum dot with explicit consideration of the coupling between the dot and the external confining gate . in the former case , as an extension of our previous study , we investigated how the statistical properties evolve when the e - e interaction strength increases from the weak ( @xmath73 ) to moderate regimes ( @xmath153 ) for @xmath154-@xmath155 . in the case of the realistic quantum dots , we first made a direct comparison between isolated and realistic quantum dots , and found that in spite of the difference in the boundary conditions , their statistical properties are qualitatively similar .
we further investigated spin and peak spacing distributions in strong interaction regimes with @xmath156 up to 4.0 and their dependence on the electron number @xmath2 .
the even / odd effect in peak spacing distributions vanishes as @xmath156 increases and becomes weaker as @xmath4 increases .
we close with two comments on our results .
first , our study , as well as other theoretical investigations , have made it quite clear that the lack of even / odd pairing in experimental peak spacing distributions is due to the e - e interaction , and the ci - rmt model is therefore not valid for the description of mesoscopic fluctuations in coulomb blockade peak spacings . in theoretical condensed matter physics
, it is well accepted that the e - e interaction at @xmath11 can be accurately described by many - body perturbation theory , most notably the rpa model
. however , the results obtained from the rpa model combined with a rmt description of single particle properties disagree significantly from those obtained with numerical approaches . understanding the sources for such discrepancies will deepen our general physical picture of quantum dots .
second , our results were obtained under the local spin density approximation within ks - sdft , whose validity , though well established in 3d bulk material systems through decades of experience , is not necessarily well justified for 2d large quantum dots with irregular confining potential and/or in the strong interaction regime . in the small dots for @xmath2 up to about 30 , it is still computationally feasible to check the validity of lsda by comparing with more accurate quantum monte carlo modeling , but not for the large - dot regime ( @xmath157 ) . on the other hand ,
the statistical quantities investigated above are experimentally accessible .
comparing future experimental data with lsda predictions will provide a unique chance to check lsda in real systems . | we use kohn - sham spin - density - functional theory to study the statistics of ground - state spin and the spacing between conductance peaks in the coulomb blockade regime for both 2d isolated and realistic quantum dots .
we make a systematic investigation of the effects of electron - electron interaction strength and electron number on both the peak spacing and spin distributions .
a direct comparison between the distributions from isolated and realistic dots shows that , despite the difference in the boundary conditions and confining potential , the statistical properties are qualitatively the same .
strong even / odd pairing in the peak spacing distribution is observed only in the weak e - e interaction regime and vanishes for moderate interactions .
the probability of high spin ground states increases for stronger e - e interaction and seems to saturate around @xmath0 .
the saturated value is larger than previous theoretical predictions .
both spin and conductance peak spacing distributions show substantial variation as the electron number increases , not saturating until @xmath1 . to interpret our numerical results ,
we analyze the spin distribution in the even @xmath2 case using a simple two - level model . |
after the discovery of the d - brane , which is the typical nonperturbative effect in the string theory , the study of the nonperturbative effect is of great interest .
we can analyze the noncritical string theory , which is a simplified model of the string theory , nonperturbatively via the matrix model ( for reviews , see @xcite ) .
the nonperturbative effect in noncritical string theory can be derived via the string equation . on the other hand
, this effect appears as the instanton when we directly analyze the matrix model .
the string equation can not describe the whole of the nonperturbative effect .
there is an ambiguity that corresponds to the initial condition of the string equation .
studying the matrix model directly , we can fix it @xcite .
this analysis is generalized to various models @xcite .
one of the most interesting models is the @xmath0 noncritical string theory because it describes the dynamics of strings in two dimensional space - time . in this paper
, we consider the penner model , which corresponds to the @xmath0 noncritical string theory with one direction of the space - time compactified on the circle with self - dual radius @xcite .
the penner model is defined as a kind of the hermitian one - matrix model .
therefore , we can extend the analysis for the @xmath1 matrix model to our case with a small modification . and
, it is easy to see that the relation between the nonperturbative effect for the @xmath1 matrix model and that in the penner model . in this paper
, we generalize the analysis of the instanton in the @xmath1 matrix model to the case of the penner model .
and we show that some of the nonperturbative effect in the @xmath0 noncritical string theory can be understood as the effects of instantons in the penner model .
this paper is organized as follows .
first , in section [ sec : c=0 ] , we consider the instanton in the @xmath1 matrix model .
we also give the interpretation which makes it possible to extend the analysis to the case of the penner model . in section [ sec : penner ] , we summarize some of the known facts related to the correspondence between the penner model and the @xmath0 noncritical string theory . in section [ sec :
inst ] , we make the redefinition of the penner model to introduce the instantons , and calculate the contribution from the instantons to the free energy . in section [ sec : concl ] is devoted for the conclusions and discussions . in appendices ,
we show the details of calculations and the behaviors of orthogonal polynomials in large @xmath2 limit .
before studying the effect of instantons for the @xmath0 noncritical string theory , in this section , we review the results obtained in the case of the @xmath1 noncritical string theory @xcite .
we focus on the contribution of instantons to the free energy . in order to extend the instanton calculation to the @xmath0 case
, we also give the precise definition of instanton .
the @xmath1 noncritical string theory can be described by the hermitian one - matrix model .
its partition function is given by @xmath3 where @xmath4 is an @xmath5 hermitian matrix . by diagonalizing the matrix @xmath4
, this theory can be described by eigenvalues of the matrix @xmath4 , and the partition function can be expressed as @xmath6 vandermonde determinant is defined as @xmath7 . here , we concentrate on the @xmath2-th eigenvalue @xmath8 and represent it @xmath9 . integrating out the other @xmath10 eigenvalues , we obtain the effective potential @xmath11 of the @xmath2-th eigenvalue @xmath12 .
the partition function can be expressed using the effective potential as @xmath13 where the subscript `` @xmath10 '' indicate that the matrix @xmath4 is replaced by an @xmath14 matrix . in the large @xmath2 limit
, the system of an @xmath5 matrix gives the same results as one of an @xmath14 matrix .
hence , we can replace it by the expectation value of the standard @xmath5 matrix @xmath15 in the large @xmath2 limit , the effective potential @xmath11 can be expressed using the resolvent @xmath16 as @xmath17 at the leading order of the @xmath18 expansion .
now we consider the model corresponding to the @xmath1 noncritical string theory . taking the double scaling limit
, we obtain @xmath19 the first derivative of the effective potential vanishes in two regions .
one is the region where the resolvent has the cut .
the effective potential takes a constant value in this region , and it is local minimum .
the other is the point of @xmath20 where the effective potential has local maximum .
the region of the cut gives the dominant contribution , which corresponds to the perturbative expansion .
we call this region as `` dominant saddle point . ''
and the point of the local maximum of the effective potential gives sub - dominant contribution , which corresponds to the nonperturbative effect .
we call this point as `` sub - dominant saddle point , '' and this sub - dominant contribution , or the configuration in which one or more eigenvalues are at this `` sub - dominant saddle point , '' as `` instanton . ''
up to this point , we have restricted ourselves to the @xmath2-th eigenvalue
. however , there are @xmath2 eigenvalues and all of them can be possibly at the `` sub - dominant saddle point . ''
including these , we obtain @xmath21 we have divided the integration region , namely , inside the cut and outside the cut .
the `` dominant saddle point '' is identical to the region inside the cut , and the `` sub - dominant saddle point '' is included in the region outside the cut .
the superscript `` 0-inst '' indicates that all eigenvalues are inside the cut and `` @xmath22-inst '' indicates that @xmath22 eigenvalues are outside the cut . by neglecting the interaction between the eigenvalues ,
the free energy becomes @xmath23 .
\end{aligned}\ ] ] now , we consider why the nonperturbative correction of the free energy becomes imaginary .
the partition function of the matrix model is defined as an integration of a real function . by this definition ,
the free energy of this model should be real . to obtain the imaginary number ,
some change of the definition is needed .
the imaginary part of the free energy reflects the instability of the model . in the @xmath1 matrix model ,
all the eigenvalues distribute around one minimum of the potential .
this minimum corresponds to the `` dominant saddle point . ''
however , the potential of the @xmath1 matrix model has some local minima other than this , or it is unbounded from below . to prevent the eigenvalues from distributing around these unwanted minima , or the unbounded region
, we should deform the contour of the integration .
we integrate the effective potential over @xmath9 along the real axis around the minimum corresponding to the `` dominant saddle point , '' and the contour turns at right angle after reaching the `` sub - dominant saddle point.''(fig .
[ contour_of_c=0 ] ) then , we have the integration orthogonal to the real axis at the `` sub - dominant saddle . ''
it gives an imaginary number .
this modification of the curve of the integration is needed for all eigenvalues , and chosen not to include the unwanted vacua .
because we can define the partition function by the integration of the analytic function , the partition function depends only on the beginning and the end of the contour .
noncritical string theory . here
, we consider a cubic potential for instance .
we should put the beginning and the end of the contour at @xmath24 in the shaded region . ] for example , we take a cubic potential @xmath25 in this case , the potential is unbounded from below .
then , we should define the integration so as to avoid this divergence . on the other hand , the partition function
is defined as @xmath26 if we define the contour of the integration to be along the real axis , the integrand is diverges at @xmath27 .
generally , we assume that the contribution from boundary terms of integration vanishes in the @xmath1 matrix model .
it can be done by taking the beginning and the end of the contour to be @xmath28 in the shaded region in fig .
[ contour_of_c=0 ] | @xmath29 , @xmath30 or @xmath31 .
now , we put the beginning of the contour in the region @xmath28 and @xmath31 , and the end of the contour in @xmath28 and @xmath32 .
then , we obtain the wanted integration : the dominant contribution comes from the cut .
it correspond to the @xmath1 noncritical string theory . if we put the beginning and the end of the contour in the region @xmath33 and @xmath34
, integration does not include the contribution from the cut .
this definition of the integration does not correspond to the @xmath1 noncritical string theory .
thus , the choice of the beginning and the end of the contour determines which vacua are included in the theory .
we take the contour corresponding to the @xmath1 noncritical string theory .
the contributions from the perturbative expansion are divergent series , and the instantons give corrections to this series .
therefore , it is nt well - defined .
but if we restrict ourselves on the imaginary part of the free energy , the situation becomes different .
the contributions of a perturbative expansion are real because they come from the integration inside the cut .
hence , the dominant contribution of the imaginary part comes from the instanton .
therefore , the imaginary part of the nonperturbative effect is well - defined . in this way
, we have obtained the imaginary part of the nonperturbative effect .
in this section , we study the penner model .
the penner model is defined as a hermitian one - matrix model with a logarithmic potential .
it is noted that in the suitable double scaling limit , the free energy of the penner model is identical to that of the @xmath0 noncritical string theory with the target space compactified on the circle of self - dual radius @xmath35 @xcite . here , we review the relation between the penner model and @xmath0 , @xmath35 noncritical string theory . the penner model is defined as a hermitian one - matrix model with the following potential : @xmath36 we can derive the free energy via the method of orthogonal polynomials .
this method makes use of an infinite set of polynomials obeying the orthogonality condition : @xmath37 the normalization of orthogonal polynomials is given by having leading term @xmath38 .
the partition function of the hermitian one - matrix model is given by @xmath39 so , we should calculate @xmath40 .
orthogonal polynomials satisfy the following two recursion relations , [ recursionrelation ] @xmath41_{nm}p_m(x )
. \label{recursion2 } \end{aligned}\ ] ] using these relations , we can easily see that @xmath42 . for the case of the penner model ,
these recursion relations are exactly soluble .
solutions of these relations are @xmath43 the partition function of the model is given by @xmath44 , and the free energy is @xmath45 to obtain the free energy identical to the @xmath0 and @xmath35 noncritical string theory , we take the following double scaling limit @xmath46 and replace the summation over @xmath22 by integration with respect to @xmath47 via euler - maclaurin summation formula .
then , we obtain @xmath48 up to regular terms . taking @xmath49 , we obtain @xmath50 this expression is identical to the free energy of the @xmath0 noncritical string theory compactified on the circle with self - dual radius .
to see the precise behavior of the penner model , we consider the resolvent and the concrete expressions of orthogonal polynomials .
first , we consider the resolvent . in the large @xmath2 limit
, we can derive the resolvent @xmath51 from the loop equation , and we obtain @xmath52 here , the sign of `` @xmath53 '' is depends on the location of the cut .
the position of the cut is identical to the support of the eigenvalue density .
it is determined by the following function @xmath54 . @xmath55
the support of the eigenvalue density is an arc connecting two branch points of the resolvent satisfying the following two conditions : the real part of @xmath54 vanishes along this support and the real part of @xmath54 is negative around this support @xcite .
here , we put @xmath56 on one of branch points of the resolvent .
the penner model is defined with positive coupling constant @xmath57 in .
we also restrict the integration to positive eigenvalues , from @xmath58 to @xmath59 .
therefore there are contributions the boundaries at @xmath58 and @xmath60 .
these contributions vanish for @xmath57 .
however , to obtain the model corresponding to the @xmath0 noncritical string theory , @xmath61 should be negative .
we calculate the free energy , or some expectation value , with positive @xmath61 , and at the end , analytically continue the result to negative @xmath61 . in this process
, we use the fact that boundary terms of the integration vanish for positive @xmath61 . however , under the analytic continuation to negative @xmath61 , it is merely an assumption that boundary terms can be dropped .
in fact , if we use the contour of the integration which is used in the case of positive @xmath61 , boundary terms do not vanish for negative @xmath61 .
therefore we should deform the contour to drop boundary terms . as is the case with the model corresponding to the @xmath1 noncritical string theory
, we can determine the contour of the integration by specifying the position of its both ends .
for the penner model , boundary terms vanish when we put the boundary in the region where @xmath28 with real part of eigenvalues negative .
because the beginning and the end of the contour are located at the same point , the contour should go around the singularity of the logarithmic potential of the penner model to obtain a nontrivial result .
then , the contour of the integration becomes as indicated in fig .
[ fig : contour_of_penner ] .
the position of the cut changes depends on the choice of the contour .
for positive @xmath61 , the cut is located on the positive real axis and the contour can be taken to be along the cut . for negative @xmath61 ,
both ends of the cut have the imaginary part , and the cut intersect the real axis on its positive part.(fig .
[ fig : contour_of_penner ] ) the result of the integration does nt change under the deformation of the contour which does nt move the both ends and does nt cross singularity . and
the position of the cut has been determined when we have specified the position of the both ends of the contour .
then , we can take the contour to be along the cut . .
we should put both ends of the contour at @xmath28 in the shaded region .
the support of the eigenvalue distribution is along the contour , and no longer on the real axis . ]
next , we consider the concrete expression of orthogonal polynomials . we show only the result here , and
the detailed calculations are given in appendix [ app : orth_poly ] .
in large @xmath2 limit , orthogonal polynomials can be written as @xmath62,\end{aligned}\ ] ] where , @xmath63 is @xmath54 with @xmath61 replaced by @xmath64 .
in this section , we consider the effect of the instanton in the penner model . in the case of the @xmath1 matrix model ,
the model is a hermitian one - matrix model , and its dynamics can be reduced to that of the eigenvalues .
the instanton is defined as the configuration in which one of the eigenvalues is located at the local maximum of the effective potential .
the penner model is also a hermitian one - matrix model .
we can use the same definition of the instanton as that for the case of the @xmath1 theory .
the effective potential of an eigenvalue of a hermitian one matrix model is defined as @xmath65 the effective potential can be expressed in terms of orthogonal polynomials as @xmath66 where the orthonormal functions @xmath67 are build from the orthogonal polynomials .
the `` dominant saddle '' of this effective potential gives the contribution to the partition function that corresponds to the perturbative expansion .
the orthonormal functions can be approximated in the large @xmath2 limit as @xmath68 .
\label{largenorthfunc}\ ] ] therefore , the `` dominant saddle '' of @xmath69 is located at the position of the cut of the resolvent with @xmath61 replaced by @xmath64 . for the model corresponding to the @xmath1 noncritical string theory , the position of the cut of @xmath70 for @xmath71 is included in the position of the cut of @xmath72 .
therefore , we can divide the integration into two parts : the inside the cut of and outside of the cut of @xmath72 . on the other hand , for penner model
, the position of the cut is varies with @xmath22 .
but , the contour of the integration can be deformed because of the analyticity of @xmath70 .
so , we first decompose the integration into that of the individual @xmath73 , and then deform the contour to be along the cut . in this way
, we can pick up the contribution from the `` dominant saddle '' of all of @xmath73 .
these contributions reproduce the perturbative expansion of the @xmath0 noncritical string theory . and @xmath74 , and the end of the contour in the region @xmath75 .
the integration picks up all cuts in the region @xmath76 for @xmath77 .
the cut with @xmath78 gives the contribution corresponding to the perturbative series of the @xmath0 noncritical string theory .
the cuts with @xmath79 correspond to the nonperturbative effect . ]
there is no local maximum in the effective potential for the penner model except for what comes from the cut of the resolvent .
however , the effective potential is multivalued on the complex plane of the @xmath2-th eigenvalue @xmath9 , due to the logarithmic term of the potential .
so , we use a polar coordinate on which the effective potential becomes single - valued .
limit , the effective potential is expressed in terms of the resolvent .
because the resolvent has the cut , the effective potential seems to be multivalued .
however , this multivaluedness is due to the approximation that is justified only in the large @xmath2 limit .
$ ] can be expressed in terms of the polynomials , and therefore the effective potential does nt have the multivaluedness except what comes from the potential of the model . ] when we extend the range of @xmath81 to @xmath82 , there is a cut for each additional @xmath83 of @xmath84 .
these infinite number of the cuts can be counted as the `` saddle point '' of the effective potential . for the penner model
, we usually assume that boundary terms of the integration with respect to the eigenvalues vanish . in the case for negative @xmath61
, each end of the contour should be in the region in which @xmath74 and @xmath85 .
when , the range of @xmath84 is extended to @xmath86 , each end of contour can be at any of @xmath74 in shaded region of fig [ contour_of_c=1 ] ( @xmath87 with @xmath88 ) . to make the model correspond to the @xmath0 noncritical string theory , we should add a small imaginary part to @xmath61 , as we have seen in previous section .
then , boundary terms vanish at @xmath89 .
we put the beginning of the contour at @xmath74 in one of the shaded region of fig [ contour_of_c=1 ] , and the end of the contour at @xmath89 ( fig [ contour_of_c=1 ] ) .
this contour picks up the cut which lies in the region where @xmath84 is greater than the beginning of the contour .
the dominant contribution comes from the cut with the lowest value of the effective potential , which is located at the smallest @xmath84 of all the cuts along the contour , and the other cuts give sub - dominant contributions .
now , we calculate the dominant contribution and nonperturbative corrections to the free energy of the penner model .
first , integrating the effective potential for the @xmath2-th eigenvalue , we calculate the nonperturbative correction from the configuration in which one eigenvalue is instanton . the effective potential can be written as @xmath90 where @xmath91 is omitted because it does not depend on @xmath9 . in the large @xmath2 limit ,
the orthonormal function @xmath70 is constant on the cut up to sub - leading order , because @xmath70 can be expressed as .
we call the cut located in the region @xmath92 @xmath93-th cut and denote it by @xmath94 .
we define the value of @xmath63 on these cuts as @xmath95 because the orthogonal polynomials @xmath96 are single valued on the complex plane of @xmath9 , the orthonormal functions @xmath70 become multivalued via the potential of the penner model @xmath97 .
therefore , the next to leading order corrections in @xmath18 of @xmath70 is single valued .
in fact , we can see that the @xmath98 contribution of the orthogonal polynomials are single valued , via their concrete expression in the large @xmath2 limit ( see appendix ) . then , integrating the effective potential , we obtain , @xmath99 } \int_{\mathcal c_0}{\mathrm d}x\,\psi_n^2(x ) .
\end{aligned}\ ] ] here , we define the normalization of the orthonormal functions in such a way that @xmath100 , namely , @xmath101 to obtain the result corresponding to the @xmath0 noncritical string theory , we should take the suitable double scaling limit .
we perform the rest of the calculation in this limit .
we take the following limit , @xmath102 in this limit , the difference of @xmath63 s between two neighboring cuts becomes , @xmath103 for the penner model to correspond to the @xmath0 noncritical string theory , we should take @xmath49 , and it becomes @xmath104 .
we choose the contour in such the way that we pick up the @xmath93-th cut from @xmath105 to @xmath106 in the integration .
the dominant contribution comes from the @xmath107-th cut , because the @xmath108 is smallest for smallest @xmath93 .
contributions from other cuts of @xmath109 are contributions from instantons .
then , the dominant contribution , or the contribution from the configuration without the instanton , of the integration of the effective potential becomes @xmath110 by replacing the summation over n by an integration , the contributions from instantons are @xmath111 here , the summation over @xmath22 is replaced by the integration over @xmath112 .
we also introduced some regularization to suppress the contribution from the terms with @xmath113 , because the double scaling limit does not work well in the region @xmath113 and the terms with @xmath113 is expected to be nonuniversal .
using , and @xmath49 , we obtain @xmath114 but there is further correction to which comes from the fact that we replace the summation over @xmath22 by an integration over @xmath112 .
this correction can be derived via the euler - mclaurin summation formula . including this correction , prefactors of these terms diverge . to see this divergence
, we introduce a regularization .
when integrating over @xmath112 in , we have introduced some regularization , for example , @xmath115 we first take @xmath116 , and after the integration , we take the limit of @xmath117 .
now , we return the integration over @xmath112 to the summation . then , @xmath112 can be regarded as integer , and it can be written as @xmath118 then , taking @xmath119 , we obtain @xmath120 therefore , in the limit of @xmath121 , the prefactors are diverge .
we discuss this point later . up to now
, we have simply considered the correction to the free energy from instantons to be @xmath122 .
this is because we have ignored interactions between instantons , and we obtain @xmath123 then , the partition function can be written as @xmath124 therefore , the contribution from instantons to the free energy can be written as @xmath122 approximately . to be exact , however , there are interactions between instantons , and we should include these effect .
we consider multi - instanton effects in more detail . to derive the effective potential of multiple eigenvalues
, we use a slightly different method from that in @xcite .
the partition function of the matrix model can be written as @xmath125 here , the vandermonde determinant @xmath126 can be expressed in terms of the orthogonal polynomials as @xmath127 using this expression we can obtain the effective potential .
first , we consider the effective potential of one eigenvalue
. integrating out the other eigenvalues than the @xmath2-th eigenvalue @xmath12 , we obtain @xmath128 this agrees with .
next , we consider the case in which two of eigenvalues are instantons
. integrating out @xmath129 eigenvalues , excluding @xmath12 and @xmath130 , we obtain @xmath131 here , the second term can be dropped because this term vanishes due to the orthogonality of the orthogonal polynomials .
matrix model , the integration around the saddle point of the instanton does not vanish .
this contributes to the effective potential .
then , we can not simply neglect this term .
on the other hand , integrations inside individual cuts vanish separately in the case of the penner model .
this is because the orthogonal polynomials do not depend on the contour .
the orthogonality condition defined on the contour which is along some of cuts is satisfied on the contour which is along only one cut . then , there are no contributions to the effective potential
. therefore we can drop this term in the case of the penner model .
] then , the contribution to the partition function from the configuration in which two of eigenvalues are the instanton becomes @xmath132 where we include the factor of @xmath133 , which reflects the number of way of specifying two isolated eigenvalues .
we can calculate the contribution from the configuration in which @xmath134 eigenvalues are instanton in a similar way .
it can be written as @xmath135 by summing up these contributions , the partition function can be written as @xmath136 then , we calculate this using concrete expressions of the orthonormal functions .
we also use the regularization which is introduced in . using ,
we obtain @xmath137 the free energy can be obtained by exponentiating this partition function : @xmath138 \notag\\ & = \sum_{\eta=0}^\infty\log \left[\frac{1}{1-{\mathrm{e}}^{2\pi i(t+\lambda\eta)}}\right ] \notag\\ & = \sum_{\eta=0}^\infty\sum_{l=0}^\infty \frac{1}{l}{\mathrm{e}}^{2\pi i(t+\lambda\eta)l } \notag\\ & = \sum_{l=1}^\infty \frac{1}{l\left(1-{\mathrm{e}}^{2\pi i \lambda l}\right)}{\mathrm{e}}^{2\pi itl } .
\end{aligned}\ ] ] taking @xmath49 , we obtain @xmath139 furthermore , we take @xmath119 . then , @xmath140 can be written as @xmath141 in the limit of @xmath121 , the prefactors are diverge .
it is known that nonperturbative corrections of the @xmath0 noncritical string theory compactified on the circle with radius @xmath142 are given by terms of the form @xmath143 and @xmath144 with positive integer @xmath93 .
our result agrees with these forms .
another special feature of @xmath140 is that prefactors of these terms diverge . to make the comparison of prefactors with those of the nonperturbative effect of the @xmath0 noncritical string theory , we use not the worldsheet formulation ( the liouville theory ) but the matrix quantum mechanics formulation .
the free energy of the @xmath0 noncritical string theory with radius @xmath142 can be obtained via the matrix quantum mechanics and expressed as @xcite @xmath145 taking the contour to be the whole real axis , we obtain @xmath146 these terms come from the residues at the two series of poles @xmath147 and @xmath148 .
there are two types of nonperturbative effects .
the first term which comes from the poles @xmath147 does nt depend on the radius @xmath142 at leading order of the perturbative expansion .
this term corresponds to the effect due to the d - instanton . on the other hand , the second term which comes from the poles @xmath148 depends on the radius @xmath142 at the leading order of the perturbative expansion
this term corresponds to the effect due to the d - particle which is not localized in the direction compactified on the circle with radius @xmath142 .
for the self - dual radius of @xmath35 , prefactors of these terms diverge individually . however , the whole is regular because these divergences cancel each other .
therefore the instantons of the penner model correspond to only one of these two types of the nonperturbative effect .
in fact , the expression with regularization is quite similar to the individual terms of .
these two terms can not be distinguished for @xmath35 .
someone may think that it is more similar to the first term of , but it depends on how to take the regularization .
we can not see which corresponds to the effect of instantons in the penner model .
in this paper , we have studied the correspondence between the effects of the instantons in the penner model and the nonperturbative effects in the @xmath0 noncritical string theory with self - dual radius .
we have calculated the contribution from the instantons to the free energy of the penner model . in the cases of the model corresponding to the @xmath1 noncritical string theory and kazakov series in the @xmath149 noncritical string theory ,
the instantons in hermitian one - matrix model are studied in recent works .
the effects of the instantons agree with the nonperturbative effects of the noncritical string theory in these cases . in this paper , we have compared the effect of the instantons in the penner model with the nonperturbative effect of the @xmath0 noncritical string theory compactified on the circle with self - dual radius of @xmath35 . in the case of the penner model , especially in the double scaling limit corresponding to the @xmath0 noncritical string theory ,
the eigenvalues can no longer regarded as real . to make the model well - defined
, we should introduce an analytic continuation .
we have extended the definition of the penner model , and defined the partition function as a multiple contour integral with respect to the @xmath2 eigenvalues .
we have also found the suitable choice of the both ends of the contour that gives the effect of instantons corresponding to the nonperturbative effect of the @xmath0 noncritical string theory . in the case of the @xmath1 matrix model
, the instanton is defined as the configuration with an eigenvalue on the local maximum of the effective potential . in the case of the penner model , there are no local maxima of the effective potential .
there are infinite number of cuts . in this case
, the instanton is defined as an eigenvalue is located on the other cut along the contour than that corresponding to the perturbative series . in this way , we have generalized the calculation for the instantons to the penner model .
thus , we have seen the correspondence between the effects of the instanton and nonperturbative effects of the @xmath0 noncritical string theory at the leading order .
the evaluation of the next - to - leading order fixes the prefactors of the contribution from the instantons . in the penner model , these prefactors diverge . on the other hand , in the @xmath0 noncritical string theory with self - dual radius of @xmath35
, the prefactors of the nonperturbative effect , which is calculated in the matrix quantum mechanics , do not diverge .
however , the nonperturbative effects of the @xmath0 noncritical string theory is constructed from the effects that come from the d - instantons and d - particles .
if we pick up the effects of the d - instantons only or d - particles only , the prefactor diverges .
therefore , the divergence of the prefactor does not mean the inconsistency of the correspondence .
furthermore , using the expression in which we introduce a regularization , we can find a similarity between our result and nonperturbative corrections which is calculated in the matrix quantum mechanics .
for the self - dual radius of @xmath35 , we can not distinguish these effects that come from d - instantons and those from d - particles .
although we can not specify which of d - instanton or d - particle corresponds to the instantons in the penner model , there should be the effect corresponding to the other . and including this , we will obtain the prefactor which does nt diverge .
this is left for future studies .
the authors are grateful to h. kawai for valuable advice and discussions .
we would also like to thank m. hanada for fruitful discussions .
this work is supported in part by the grant - in - aid for scientific research ( 17@xmath1501647 ) and the grant - in - aid for the 21st century coe `` center for diversity and universality in physics '' from the ministry of education , culture , sports , science and technology ( mext ) of japan .
the work of y.m . is supported in part by jsps research fellowships for young scientists .
the method of orthogonal polynomials is a powerful tool to study the hermitian one - matrix model . in this section ,
we consider the orthogonal polynomials in the large @xmath2 limit for the penner model .
orthogonal polynomials @xmath151 is defined to obey the following orthogonal condition : @xmath152 the partition function of the hermitian one - matrix model is obtained via the recursion relation @xmath153_{nm}p_m(x ) , \tag{\ref{recursion2 } } \end{aligned}\ ] ] where @xmath42 . in the case of the penner model , we take the potential @xmath154 , and using , we obtain @xmath155 where @xmath156 . using the ratio of the orthogonal polynomials @xmath157 ,
can be expressed as @xmath158 in the large @xmath2 limit , the rescaled index @xmath159 becomes a continuous variable @xmath160 and @xmath161 becomes continuous function @xmath162 .
then , we can expand @xmath93 as [ largenexpand ] @xmath163 in a similar fashion , we expand @xmath164 and @xmath165 as @xmath166 substituting these relation to , we obtain @xmath167 using these relation , we obtain [ sol_of_k ] @xmath168 . \end{aligned}\ ] ] the sign of @xmath53 is taken to be @xmath169 at @xmath74 .
the orthogonal polynomials can be expressed as @xmath170.\ ] ] by using the euler - maclaurin summation formula to convert a summation into an integral , it becomes @xmath171.\ ] ] the orthonormal functions @xmath70 is convenient to describe the @xmath22-dependence of these functions .
they can be expressed as @xmath172.\ ] ] then , substituting , we obtain [ sol_of_p ] @xmath173^{\frac14 } \exp\left[-\frac{n\xi}{2}g_\xi(x)\right ] , \\
g\xi g_\xi(x ) & = -(1 + 2g\xi)\log \left[\frac{1}{2}\left(x-(1 + 2g\xi)+\sqrt{x^2 - 2(1 + 2g\xi)x+1}\right)\right ] \notag\\&\qquad -\log\left[\frac{1}{2}\left(x+1+\sqrt{x^2 - 2(1 + 2g\xi)x+1}\right)\right ] \notag\\&\qquad + \log\left[\frac{1}{2}\left(x-1+\sqrt{x^2 - 2(1 + 2g\xi)x+1}\right)\right ] \notag\\&\qquad + \sqrt{x^2 - 2(1 + 2g\xi)x+1 } -2g\xi .
\end{aligned}\ ] ] the orthogonal polynomials @xmath96 are single valued on the complex plane of @xmath9
. however , its asymptotic form in the large @xmath2 limit has a cut .
the presence of a cut indicate that the asymptotic form of the orthogonal polynomials is not single valued .
this cut is only the consequence of the large @xmath2 limit .
we should take the branch which reproduce the suitable form of the orthogonal polynomials in the large @xmath2 limit .
the cut in the asymptotic form of the orthogonal polynomials is identical to that of the resolvent .
we should take the physical sheet on which the resolvent becomes @xmath174 at @xmath27 to obtain the suitable branch of the asymptotic behavior of the orthogonal polynomials .
the orthogonal polynomials can be expressed in terms of the expectation value of the matrix model as @xmath175 .
\end{aligned}\ ] ] here , the subscript `` @xmath22 '' indicates that the expectation value is taken in the @xmath176 matrix model , and the subscript @xmath177 indicates connected part .
then , in the large @xmath2 limit , the orthogonal polynomials can be expressed in terms of the resolvent @xmath178 as @xmath179 , \label{orth_and_resolvent}\ ] ] where @xmath180 is the resolvent with the coupling of the matrix model @xmath61 replaced by @xmath64 .
in fact , shows that @xmath96 behaves as @xmath181\ ] ] at the leading order , and using the relation @xmath182
= \xi r_\xi(x ) , \end{aligned}\ ] ] we can see the agreement with .
the sign included in the expression of @xmath93 should be taken to be the same as the resolvent . with @xmath9 in some region
does nt agree with the sign of the resolvent .
this comes from the fact that the suitable sign of this term depends on @xmath160 .
if we assume that @xmath162 is continuous function in @xmath160 and take the suitable sign for each @xmath160 , it agrees with the resolvent . ]
the resolvent becomes @xmath183 in the limit @xmath27 on the physical sheet .
then , the orthogonal polynomials behave as @xmath184 = x^n.\ ] ] it reproduces the suitable behavior of the orthogonal polynomials in the limit of @xmath27 .
it also corresponds to that the sign of @xmath53 is taken to be @xmath169 at @xmath74 in . on the cut
, the orthogonal polynomials can be obtained as a linear combination of two branches corresponding to , for instance , the @xmath185 and @xmath186 in the sign of @xmath53 in the . it is explained as follows .
when we consider the orthonormal function @xmath187 , in the large @xmath2 limit , it becomes @xmath188 \sin[\frac{n\xi}{2}{\mathop{\mathrm{im}}}g_\xi(x)+\frac{\pi}{4}].\ ] ] since the imaginary part of the @xmath54 is the density of eigenvalues and @xmath63 is obtained from @xmath54 by replacing @xmath61 with @xmath64 , the orthonormal functions and orthogonal polynomials have the point where @xmath189 on the cut .
the orthogonal polynomials can be expressed in terms of the expectation value of the matrix model as @xmath190 then , in the large @xmath2 limit , it becomes @xmath191 at the point where an eigenvalue is classically located . because the cut of the resolvent comes from the support of the eigenvalue distribution , the point where @xmath191 is on the cut . to obtain the point where @xmath189
, we should take the linear combination of two branches .
meanwhile , the oscillatory behavior of @xmath192 can be approximated by the average of @xmath193 and we can regard it as a constant inside the cut .
we can see how to take the linear combination of two branches of the asymptotic form of the orthogonal polynomials by studying the behavior near the endpoint of the cut . in this region ,
the orthogonal polynomials can be approximated by the airy function . in the double scaling limit ,
the orthonormal functions become @xmath194.\ ] ] the recursion relation that the orthogonal polynomials satisfy gives the differential equation which the orthonormal functions satisfy in the double scaling limit . the differential equation for the orthonormal functions at the leading order of the perturbative expansion of
the noncritical string is @xmath195 the both ends of the cut is located on @xmath196 . introducing the new variable @xmath197 and @xmath84 defined by @xmath198
, we obtain a differential equation , @xmath199 then , for @xmath200 , we can make use of a standard analysis of the airy function .
this implies that the cut is on the line of @xmath200 .
it corresponds to the condition that the cut is on the line on which @xmath54 is pure imaginary .
the other condition that the cut is embedded on the region where @xmath201 determines that the cut is on the line of the @xmath202 .
the airy function can be expressed via the integration on the complex plane of @xmath93 as @xmath203 .
\label{airy}\ ] ] the asymptotic form of the airy function can be obtained by the method of the steepest descent .
there are two saddle points in the @xmath93 plane , corresponding to the two branches of the asymptotic form of the orthogonal polynomials . for @xmath202 and @xmath204 , we take the contour which picks up the both of two saddle points .
this corresponds to the linear combination of the two branches .
then , for @xmath205 , this contour picks up only one of the saddle point .
so we should nt take the linear combination in this region .
the contour of the integration is determined by the choice of the boundary of the contour .
we should use the same choice of the boundary for @xmath206 .
since the saddle points are placed at the different point from the case of @xmath207 , this contour picks up only one of the saddle point for both @xmath204 and @xmath205 . in this way
, we should take the linear combination of two branches only inside the cut .
the orthonormal functions @xmath208 obey the following orthonormality condition : @xmath209 here , the integration with respect to @xmath9 can be approximated by the integration inside the cut in the large @xmath2 limit .
we check satisfies this orthonormality condition , especially the normalization .
first , the orthonormal function given by becomes @xmath210\ ] ] at the leading order of the @xmath211 expansion .
when we take the average of the oscillatory behavior of the square of the orthonormal functions @xmath192 , it can be regarded as a constant on the cut . examining on the end of the cut @xmath212 ,
we find @xmath213 . here
, the position of the cut depends on the index of @xmath22 of the orthogonal polynomials .
then , @xmath22-dependence of the @xmath192 becomes @xmath214 the norm of the orthonormal function @xmath70 does not depend on @xmath22 at the leading order of the @xmath18 expansion .
second , on the correction at the order of @xmath18 , we have the following expression for the orthonormal function @xmath70 : @xmath215 here , the @xmath216 contribution can be neglected , because it does not depend on @xmath22 as we have seen earlier .
using this expression and integrating inside the cut , we obtain @xmath217 therefore , @xmath70 obeys suitable normalization condition defined by the integration inside the cut .
generally , in the double scaling limit , the quantities describing the physics of the noncritical string do not depend on the details of the potential of the matrix model .
this universality reflects the fact that the physical quantities do not depend on the cut - off .
conversely speaking , the universal quantity which does not depend on the details of the potential is physical quantity of the noncritical string theory .
we can introduce the universality into the penner model . in the case of the penner model
, we can add arbitrary polynomials into the potential .
generally , we can consider the following potential : @xmath218 when we consider the penner model as the model corresponding to the @xmath0 noncritical string theory , we restrict ourselves to the critical point on which two ends of the cut merge . on such critical point , @xmath164 in takes the value @xmath219 . in the double scaling limit , that is the @xmath220 with @xmath221 @xmath164 and @xmath165
defined in becomes the resolvent can be described in terms of the orthogonal polynomials as @xmath223 then , @xmath224 in this case becomes @xmath225 it agrees with the result for the usual penner model .
we can do the rest of the calculation in the same way as the usual penner model .
then , we obtain the same result for the contribution from the instantons .
p. di francesco , p. h. ginsparg and j. zinn - justin , `` 2-d gravity and random matrices , '' phys .
* 254 * ( 1995 ) 1 [ arxiv : hep - th/9306153 ] .
h. ginsparg and g. w. moore , `` lectures on 2-d gravity and 2-d string theory , '' arxiv : hep - th/9304011 .
i. r. klebanov , `` string theory in two - dimensions , '' arxiv : hep - th/9108019 .
m. hanada , m. hayakawa , n. ishibashi , h. kawai , t. kuroki , y. matsuo and t. tada , `` loops versus matrices : the nonperturbative aspects of noncritical string , '' prog .
phys . * 112 * , 131 ( 2004 ) [ arxiv : hep - th/0405076 ] .
h. kawai , t. kuroki and y. matsuo , `` universality of nonperturbative effect in type 0 string theory , '' nucl .
b * 711 * ( 2005 ) 253 [ arxiv : hep - th/0412004 ] .
a. sato and a. tsuchiya , `` zz brane amplitudes from matrix models , '' jhep * 0502 * * 0502 * ( 2005 ) 032 [ arxiv : hep - th/0412201 ] .
n. ishibashi , t. kuroki and a. yamaguchi , `` universality of nonperturbative effects in c < 1 noncritical string theory , '' jhep * 0509 * ( 2005 ) 043 [ arxiv : hep - th/0507263 ] .
j. distler and c. vafa , `` a critical matrix model at c = 1 , '' hutp-90-a062 s. chaudhuri , h. dykstra and j. d. lykken , `` the penner matrix model and c = 1 strings , '' fermilab - pub-91 - 049-t j. ambjorn , c. f. kristjansen and y. makeenko , `` generalized penner models to all genera , '' phys .
d * 50 * ( 1994 ) 5193 [ arxiv : hep - th/9403024 ] .
s. y. alexandrov , v. a. kazakov and d. kutasov , `` non - perturbative effects in matrix models and d - branes , '' jhep * 0309 * ( 2003 ) 057 [ arxiv : hep - th/0306177 ] . s. y. alexandrov and i. k. kostov , `` time - dependent backgrounds of 2d string theory : non - perturbative effects , '' jhep * 0502 * ( 2005 ) 023 [ arxiv : hep - th/0412223 ] . | we derive the effect of instantons in the penner model .
it is known that the free energies of the penner model and the @xmath0 noncritical string at self - dual radius agree in a suitable double scaling limit . on the other hand
, the instanton in the matrix model describes a nonperturbative effect in the noncritical string theory .
we study the correspondence between the instantons in the penner model and the nonperturbative effect in @xmath0 noncritical string at self - dual radius .
[ cols= " < " , ] * nonperturbative effect + in @xmath0 noncritical string theory + and penner model * yoshinori matsuo _ department of physics , kyoto university , kyoto 606 - 8502 , japan _ |
in 1990s turaev introduced the notion of _ shadows _ of @xmath3- and @xmath2-manifolds with the intention of studying quantum invariants of knots and @xmath3-manifolds .
a simple polyhedron @xmath0 is called a _ shadow _ of a @xmath2-manifold @xmath4 if @xmath4 collapses onto @xmath0 and @xmath0 is embedded properly and locally - flat in @xmath4 . as it is well - known ,
@xmath4 can be reconstructed from @xmath0 and an appropriate decoration of its regions with some half - integers called _ gleams_. it is called turaev s reconstruction .
shadows provide many geometric properties of @xmath3- and @xmath2-manifolds .
we refer the reader to costantino @xcite for studies of stein structures , spin@xmath5 structures and complex structures of 4-manifolds . in @xcite costantino and thurston established the relation between shadows and stein factorizations of stable maps from @xmath3-manifolds into @xmath6 .
consequently they observed the relation between hyperbolicity of @xmath3-manifolds and their _ shadow complexities _ , which was strengthened by ishikawa and koda later @xcite . here
the _ shadow complexity _ of @xmath4 is defined as the minimum number of true vertices of a shadow of @xmath4 .
this notion is introduced by costantino in @xcite , in which he studied closed @xmath2-manifolds with shadow complexities @xmath7 and @xmath8 in a special case . in @xcite martelli completely classified the closed @xmath2-manifolds with shadow complexity @xmath7 in the general case . in this paper
we study acyclic @xmath2-manifolds with shadow complexity zero by observing the structure of simple polyhedra .
to state the first theorem , we introduce a canonical way to equip internal regions of a subpolyhedron of a shadowed polyhedron with gleams .
let @xmath9 be a shadowed polyhedron , and @xmath1 a simple polyhedron with @xmath10 .
let @xmath11 be an internal region of @xmath1 .
we observe that @xmath12 and that @xmath11 might be split by @xmath13 into some internal regions @xmath14 of @xmath0 ( @xmath15 ) .
then we assign a gleam to @xmath11 by @xmath16 suppose that there exist a triangulation @xmath17 of the pair @xmath18 and a sequence of elementary simplicial collapses from @xmath19 onto @xmath20 .
then we say that @xmath0 _ collapses _ @xmath1 .
moreover , this removal is called a _ polyhedral collapse _ ( or simply _ collapse _ ) and denoted by @xmath21 .
we denote by @xmath22 the @xmath2-manifold obtained from a shadowed polyhedron @xmath9 by turaev s reconstruction .
now we state our first theorem .
[ thm : s - collapse ] let @xmath9 be a shadowed polyhedron , and @xmath1 a simple polyhedron collapsed from @xmath0 .
assign a gleam to each internal region of @xmath1 by formula ( [ sum of gleams ] ) .
then we have @xmath23 . in the second theorem
, we will study acyclic simple polyhedra and their collapsibility .
martelli introduced a way to convert a simple polyhedron whose singular set is a disjoint union of circles to a graph in @xcite .
we will use his notations and introduce some moves on a graph that correspond to collapsings .
[ thm : no vertex ] any acyclic simple polyhedron whose singular set is a disjoint union of circles collapses onto @xmath24 .
we have an important consequence of theorem [ thm : s - collapse ] and theorem [ thm : no vertex ] as follows : [ cor : shadow complexity 0 ] every acyclic @xmath2-manifold with shadow complexity zero is diffeomorphic to @xmath25 .
this paper consists of 4 sections . in section 1
we review the definitions of simple polyhedra and shadows . in section 2 and in section 3 , we give the proofs of theorem [ thm : s - collapse ] and theorem [ thm : no vertex ] respectively . in section 4 , we discuss some consequences of our theorems . throughout this paper , we work in smooth category unless otherwise mentioned .
the author would like to thank his supervisor , masaharu ishikawa , for his useful comments and encouragement .
he would also like to thank the referee for careful reading and helpful suggestions .
a compact topological space @xmath0 is called a _ simple polyhedron _ if any point @xmath26 of @xmath0 has a regular neighborhood @xmath27 homeomorphic to one of the five local models shown in figure [ local model ] .
true vertex _ is a point whose regular neighborhood has a model of type ( iii ) .
we note that the model of type ( iii ) is homeomorphic to a cone over the complete graph @xmath28 with @xmath2 vertices .
the point @xmath29 ) / ( k_4 \times \{0 \})$ ] is just a true vertex
. a connected component of the set of points whose regular neighborhoods are of type ( ii ) or ( v ) is called a _
triple line_. the union of all true vertices and triple lines is called the _ singular set _ of @xmath0 and denoted by @xmath13 .
the boundary @xmath30 of @xmath0 is the set of points whose regular neighborhoods are of type ( iv ) or ( v ) .
each component of @xmath31 is called a _ region _ of @xmath0 . if a region @xmath11 contains points of type ( iv )
then @xmath11 is called a _ boundary region _ , and otherwise it is called an _
internal region_. let @xmath4 be a compact oriented @xmath2-manifold and let @xmath32 be a ( possibly empty ) trivalent graph in the boundary @xmath33 of @xmath4 .
a simple polyhedron @xmath0 in @xmath4 is called a _ shadow _ of @xmath34 if the following hold : * @xmath4 collapses onto @xmath0 , * @xmath0 is locally flat in @xmath4 , that is , for each point @xmath26 of @xmath0 there exists a local chart @xmath35 of @xmath4 around @xmath26 such that @xmath36 , and * @xmath37 .
the following theorem by turaev is very important and is called _ turaev s
reconstruction_. [ turaev @xcite ] let @xmath0 be a shadow of a @xmath2-manifold @xmath4 .
then there exists a canonical way to equip each internal region of @xmath0 with a half - integer .
conversely , we can reconstruct @xmath4 uniquely from @xmath0 and the half - integers . each half - integer in the above is called a _
gleam_. a simple polyhedron @xmath0 whose internal regions are equipped with a gleam is called a _ shadowed polyhedron _ and denoted by @xmath9 ( or simply @xmath0 ) .
.,width=136 ] [ rmk ; gleam ] as pointed out by turaev @xcite , a gleam generalizes the euler number of closed surfaces embedded in oriented 4-manifolds .
we can interpret the gleam as follows .
let @xmath11 be an intenal region of a shadow @xmath0 in a 4-manifold @xmath4 and let @xmath38 be a point of @xmath39 .
by the locally flatness there exists a 3-ball @xmath40 around @xmath38 such that @xmath40 contains @xmath41 .
we take interval which passes through @xmath38 and is transverse to @xmath11 after giving an auxiliary euclidean metric for @xmath40 as shown in figure [ fiber ] . by taking intervals for each point of @xmath39 continuously
, we get an interval bundle over @xmath39 .
we note that the interval bundle is a subbundle of the normal bundle over @xmath39 in @xmath4 .
let @xmath42 be a small perturbation of @xmath11 such that @xmath43 is lying in the interval bundle .
if @xmath42 is generically chosen , then @xmath11 and @xmath42 have only isolated intersections . counting them with signs , we have @xmath44 for this formula we refer the reader to carrega and martelli @xcite .
we close this section with the definition of the shadow complexity .
let @xmath4 be a compact oriented @xmath2-manifold having a shadow .
we define the _ shadow complexity _ @xmath45 of @xmath4 to be the minimum number of true vertices of a shadow of @xmath4 .
costantino defined the shadow complexity for `` closed '' 4-manifolds in @xcite .
a shadow of a closed 4-manifold @xmath4 is defined as a shadow of a 4-manifold to which @xmath4 is obtained by attaching 3- and 4-handles .
in this section we introduce a proposition on pl topology and provide a lemma for the proof of theorem [ thm : s - collapse ] . [ prop ; nbd are plhomeo ] let @xmath4 be an @xmath46-dimensional compact pl manifold and fix a triangulation @xmath19 of @xmath4 .
let @xmath47 be a subcomplex of a double barycentric subdivision of @xmath19 for @xmath48 .
if @xmath49 , then @xmath50 and @xmath51 are pl - homeomorphic . for the proof of this proposition we refer the reader to (
* lemma 3.25 , theorem 3.26 ) .
a pl manifold has a unique smoothing in dimension @xmath52 @xcite .
in our case @xmath53 , proposition [ prop ; nbd are plhomeo ] with `` pl - homeomorphic '' replaced by `` diffeomorphic '' also holds .
[ lem : emb shadow ] let @xmath9 be a shadowed polyhedron , and @xmath1 a simple subpolyhedron of @xmath0 . assign a gleam to each internal region of @xmath1 by formula ( [ sum of gleams ] ) .
then we have @xmath54 .
let @xmath19 be the second barycentric subdivision of a given triangulation of @xmath1 , and set @xmath55 then @xmath56 is a subcomplex of @xmath19 , and @xmath57 .
we have @xmath58 by proposition [ prop ; nbd are plhomeo ] , where @xmath59 .
note that @xmath1 is proper and locally flat in @xmath60 .
hence @xmath1 is a shadow of @xmath60 . by turaev s reconstruction
, there should exist gleams for the internal regions of @xmath1 so that the 4-manifold reconstructed from them is diffeomorphic to @xmath60 .
it suffices to show that such gleams coincide with ones given by formula ( [ sum of gleams ] ) .
let @xmath11 be an internal region of @xmath1 , and set @xmath61 . if @xmath62 , the region @xmath11 is also an internal region for @xmath0 .
hence it is obvious that their gleams coincide by remark [ rmk ; gleam ] .
we turn to the case @xmath63 . as mentioned above , the region @xmath11 is split into internal regions @xmath64 of @xmath0 . .
the two right pictures are @xmath41.,width=377 ] let @xmath38 be a point contained in @xmath65 . then @xmath41 can be described in either of the two right parts of figure [ smooth emb ] , where the colored areas indicate @xmath66 . note that @xmath67 .
we assume that these pictures are drawn in @xmath68 and consider the regular neighborhood of @xmath41 . carrying out turaev s reconstruction with such 3-dimentional blocks ,
we get @xmath22 in which @xmath11 is smoothly embedded .
next we consider the interval bundle over @xmath69 as mentioned in remark [ rmk ; gleam ] for @xmath70 . by the smoothness of @xmath11 , if @xmath71 , the restrictions of the interval bundles of @xmath69 and @xmath72 to @xmath73 coincide .
hence the union of the interval bundles is regarded as an interval bundle over @xmath74 , and we denote it by @xmath75 .
let @xmath76 be a generic small perturbation of @xmath65 in @xmath75 such that the images of the true vertices of @xmath0 do not lie in the zero section .
then let @xmath42 ba a generic small perturbation of @xmath11 such that @xmath77 and @xmath43 lies in the restriction of the interval bundle @xmath75 to @xmath39 . by restricting @xmath42 we get a small perturbation @xmath78 of @xmath79 as in remark [ rmk ; gleam ] for @xmath70 .
note that each point @xmath80 except for a true vertex is sandwiched between @xmath79 and @xmath81 for some @xmath82 , in other words , the point @xmath38 belongs to both @xmath69 and @xmath72 . by formula ( [ sum of intersection ] ) , we have the following : @xmath83 and the proof is completed .
[ proof of theorem [ thm : s - collapse ] ] there exist diffeomorphisms @xmath84 by turaev s reconstruction , proposition [ prop ; nbd are plhomeo ] and lemma [ lem : emb shadow ] . in this section
we study simple polyhedra and its graph notation introduced by martelli , and give the proof of theorem [ thm : no vertex ] .
we first introduce convenient lemmas .
we note that the first one was shown by ikeda in ( * ? ? ?
* lemma 12 ) .
[ ikeda @xcite ] [ lem : no genus ] any region of a simple polyhedron @xmath0 is orientable and has no genus if @xmath85 .
the proof of lemma [ lem : no genus ] is given by using mayer vietoris exact sequence .
for any region @xmath11 of @xmath0 , he considered a closed surface @xmath86 obtained from @xmath11 by capping off the all boundary components of @xmath11 by disks and constructed a new simple polyhedron that contains @xmath86 .
he checked that @xmath87 must vanish .
[ lem : scc splits p ] let @xmath0 be an acyclic simple polyhedron and @xmath88 be a simple closed curve in @xmath31
. then @xmath88 splits @xmath0 into two parts such that one of them is acyclic and the other is a homology-@xmath89 .
moreover the first homology of the latter is generated by @xmath88 .
since any region of @xmath0 is orientable by lemma [ lem : no genus ] , @xmath90 is homeomorphic to an annulus .
set @xmath91 and @xmath92 .
using the mayer vietoris exact sequence for the decomposition @xmath93 , we have the isomorphism @xmath94 for @xmath95 .
hence @xmath96 for @xmath97 and @xmath98 .
moreover @xmath99 holds from the following equality @xmath100 hence @xmath101 , that is , @xmath102 has two connected components : one of them is acyclic and the other is a homology-@xmath89 .
let @xmath1 be the acyclic connected component of @xmath102 and @xmath103 the other component .
we regard @xmath0 as the union @xmath104 . from the mayer vietoris exact sequence for this decomposition
, it follows that @xmath88 generates @xmath105 .
let @xmath0 be a simple polyhedron whose singular set is a disjoint union of circles . in @xcite
, martelli introduced a graph encoded from @xmath0 and classified closed @xmath2-manifolds with shadow complexity zero .
-bundle over @xmath89 .
these pictures indicate the only boundaries of them . , width=245 ] a regular neighborhood of @xmath106 has a structure of @xmath107-bundle over @xmath89 , where @xmath107 is a cone of three points .
there are three topological types @xmath108 and @xmath109 , and they are shown in figure [ nbd_of_s1 ] .
each connected component of @xmath110 is a surface homeomorphic to a region of @xmath0 .
any surface is decomposed into disks , pairs of pants and mbius strips .
hence we have the following .
[ martelli @xcite ] [ prop : decomposition of p ] any simple polyhedron whose singular set is a disjoint union of circles is decomposed along simple closed curves into pieces homeomorphic to @xmath24 , a pair of pants , the mbius strip , @xmath111 , @xmath112 or @xmath109 .
a decomposition of @xmath0 as in proposition [ prop : decomposition of p ] provides a graph @xmath113 consisting of some edges and vertices ( b ) , ( d ) , ( p ) , ( 2 ) , ( 111 ) , ( 12 ) or ( 3 ) as in figure [ vertices ] . the vertices of type ( d ) , ( p ) ,
( 2 ) , ( 111 ) , ( 12 ) and ( 3 ) denote some portions homeomorphic to @xmath24 , a pair of pants , the mbius strip , @xmath111 , @xmath112 and @xmath109 respectively .
a vertex of type ( b ) denotes a boundary component of @xmath0 .
note that each edge encodes a simple closed curve along which @xmath0 decomposes except the edges adjoining a vertex of type ( b ) .
we also note that we distinguish the two edges adjoining the vertex of type ( 12 ) : the edge marked with two lines corresponds to a simple closed curve winding twice along the circle in @xmath13 .
as martelli said , we can uniquely reconstruct the simple polyhedron @xmath0 from a pair consisting of a graph @xmath113 and a map @xmath114 .
it is necessary to choose homeomorphisms that glues polyhedral pieces at each edge of @xmath113 since there are two self - homeomorphisms of @xmath89 , orientation - preserving and -reversing , up to isotopy .
it is encoded by a map from @xmath115 to @xmath116 .
the graph @xmath113 that describes @xmath0 is not unique since a surface decomposes into disks , pairs of pants and mbius strips in several ways .
there are some local moves as two examples shown in figure [ ih- and yv- moves ] that do not change the topological type of the polyhedron .
we call the two moves in figure [ ih- and yv- moves ] _ ih - move _ and _ yv - move_. if there exists an edge connecting two vertices @xmath117 and @xmath118 of a graph , then @xmath117 and @xmath118 are said to be _
let @xmath0 be an acyclic simple polyhedron whose singular set consists of circles .
we note that we only need to consider the case @xmath119 since there is no acyclic closed simple polyhedron without true vertices ( * ? ? ?
* theorem 1 ) .
let @xmath113 be a graph obtained from @xmath0 .
our goal is to transform @xmath113 into a 1-valent graph whose vertices are type ( b ) and type ( d ) as shown in figure [ goal graph ] .
it is obvious that this graph corresponds to the polyhedron @xmath24 .
[ assrt : no sigma ] there is no embedded @xmath109 in any acyclic simple polyhedron .
assume that there exists @xmath109 in an acyclic simple polyhedron @xmath0 .
by lemma [ lem : scc splits p ] , a simple closed curve @xmath120 splits @xmath0 into @xmath109 and an acyclic subpolyhedron but does not generate @xmath121 .
it is a contradiction .
[ assrt : tree ] the graph @xmath113 is a tree .
this follows readily from lemma [ lem : scc splits p ] .
the subgraph @xmath122 contains the root of @xmath113 .
, width=52 ] @xmath123 .
the subgraph @xmath122 contains the root of @xmath113 .
, width=94 ] if @xmath0 is decomposed into some pieces as in proposition [ prop : decomposition of p ] , there must be at least one piece homeomorphic to @xmath24 by applying iteratively lemma [ lem : scc splits p ] .
choose one of such pieces and let the corresponding vertex of type ( d ) be the root of @xmath113 .
we assume that @xmath113 has at least one vertex of type ( 111 ) .
we transform @xmath113 into a tree having no vertex of type ( 111 ) .
consider the farthest vertex of type ( 111 ) from the root .
let us denote it by @xmath124 .
if we remove the corresponding piece @xmath111 from @xmath0 , it is decomposed into three subpolyhedra .
let @xmath103 be one of them such that it contains @xmath24 corresponding to the root and @xmath125 and @xmath126 be the remaining two subpolyhedra .
let @xmath122 , @xmath127 and @xmath128 be the subgraphs of @xmath113 corresponding to @xmath103 , @xmath125 and @xmath126 respectively as shown in figure [ around i d ] . by lemma [ lem : scc splits p ] ,
at least one of @xmath125 and @xmath126 is homologically @xmath89 .
assume that @xmath125 is so .
let @xmath88 be the simple closed curve that cuts @xmath125 off from @xmath0 .
by lemma [ lem : scc splits p ] , @xmath88 generates @xmath129 .
if @xmath125 has no boundary except @xmath88 , the simple polyhedron obtained from @xmath125 by capping off the boundary component by a disk is closed , acyclic and without true vertices , contrary to ( * ? ? ?
* theorem 1 ) .
hence @xmath125 has some boundary components other than @xmath88 . in other words
, there exists a vertex of type ( b ) in @xmath127 .
[ assrt : ( 12)-(p)-(b ) ] let @xmath130 be a vertex of type ( p ) such that it is adjacent to a vertex of type ( b ) . if @xmath130 is adjacent to a vertex of type ( 12 ) , the edge between them is marked with two lines .
assume that there is an edge , denoted by @xmath131 , adjoining @xmath130 and a vertex of type ( 12 ) such that the edge is not marked with two lines .
let @xmath132 be the vertex of type ( 12 ) . along the simple closed curve corresponding to @xmath131
, @xmath0 is decomposed into two subpolyhedra : one contains a region @xmath11 corresponding to @xmath130 and another contains @xmath112 corresponding to @xmath132 .
since @xmath11 has two boundary components , the former subpolyhedron has a 1-cycle .
by lemma [ lem : scc splits p ] , the latter subpolyhedron should be acyclic .
however the subpolyhedron collapses so that it contains a mbius strip in a region .
this contradicts lemma [ lem : no genus ] .
.,width=340 ] .,width=491 ] [ assrt : moves of collapsing ] the moves of @xmath113 described in figure [ moves of collapsing(graph ) ] ( a ) , ( b ) and ( c ) are realized by collapsings of @xmath0 .
\(a ) ( resp .
( b ) ) the corresponding part of @xmath0 is shown in figure [ moves of collapsing(poly ) ] ( a1 ) ( resp .
it can collapse along the boundary component in the direction of the arrows described in the figure , and the resulting polyhedron is shown in figure [ moves of collapsing(poly ) ] ( a2 ) ( resp .
( c ) figure [ moves of collapsing(poly ) ] ( c1 ) shows the corresponding part of @xmath0 .
after we glue the two pieces , a pair of pants and @xmath112 as in figure [ moves of collapsing(poly ) ] ( c1 ) , it turns out that @xmath0 can be described as in figure [ moves of collapsing(poly ) ] ( c2 ) .
let @xmath0 collapse along a part of the boundary component as indicated by the arrows .
then the resulting polyhedron is shown in figure [ moves of collapsing(poly ) ] ( c3 ) .
[ assrt : ( 111 ) disappear ] the graph @xmath113 can change into a subgraph that does not contain @xmath124 and @xmath127 only by the moves ( a),(b),(c ) in figure [ moves of collapsing(graph ) ] , the ih - move and the yv - move .
let @xmath118 be a vertex of type ( b ) in @xmath127
. then @xmath118 is adjacent to a unique vertex @xmath133 of type ( 111 ) , ( 12 ) or ( p ) .
if @xmath133 is of type ( 111 ) , that is @xmath134 , we apply the move ( a ) and the proof is completed .
we consider the case where @xmath133 is type ( 12 ) or ( p ) .
if @xmath133 is of type ( 12 ) , the edge between @xmath118 and @xmath133 must be marked with two lines as shown in figure [ possible_cases ] ( 1 ) by the same reason of assertion [ assrt : ( 12)-(p)-(b ) ] .
if @xmath133 is of type ( p ) , the two vertices adjacent to @xmath133 other than @xmath118 can not be of type ( b ) by the acyclicness of @xmath0 .
therefore the possible cases are as shown in figure [ possible_cases ] .
set @xmath135 . in each case of ( 1)(7 ) in figure [ possible_cases ] , we apply some moves to @xmath136 as follows and denote the resulting graph by @xmath137 ( @xmath138 ) . * in this case we apply the move ( b ) . then the number of vertices in @xmath137 is less than the one in @xmath136 . * in these cases
we apply the move ( c ) . then the number of vertices in @xmath137 is less than the one in @xmath136 . * in this case we apply yv - move .
then the number of vertices in @xmath137 is less than the one in @xmath136 .
* let @xmath139 be the vertex adjacent to @xmath118 in @xmath136 .
we define a set @xmath140 of vertices of type ( p ) as follows : @xmath141 if and only if there is a path between @xmath139 and @xmath117 in @xmath136 such that it contains vertices only of type ( p ) .
then there exists at least one vertex in @xmath140 which is adjacent to a vertex different from @xmath118 and of type other than ( p ) or ( 111 ) since @xmath113 is a tree .
choose such a vertex @xmath117 in @xmath140 and a path between @xmath139 and @xmath117 as above .
we apply ih - moves along the path , and then the resulting graph @xmath137 is as in one of the cases ( 1)(5 ) . in all cases ( 1)(7 )
we can decrease the number of vertices in @xmath136 but this is finite .
hence it comes down to the case where @xmath118 is adjacent to one of type ( 111 ) , and the proof is completed .
we iterate assertion [ assrt : ( 111 ) disappear ] until all the vertices of type ( 111 ) in @xmath113 disappear and denote the resulting graph by @xmath142 .
there still exists a vertex of type ( b ) in @xmath142 by ( * ? ? ?
* theorem 1 ) . we denote it by @xmath118 .
the vertex adjacent to @xmath118 is of type ( d ) , ( 12 ) or ( p ) .
if it is of type ( d ) , @xmath142 is as in figure [ goal graph ] , which corresponds to @xmath24 .
if the vertex adjacent to @xmath118 is of type ( 12 ) or ( p ) , the possible cases are ( 1)(3),(5),(7 ) in figure [ possible_cases ] . as in the proof of assertion [ assrt :
( 111 ) disappear ] we apply some moves to @xmath142 and finally obtain a graph such that @xmath118 is adjacent to one of type ( d ) .
this completes the proof of theorem [ thm : no vertex ] .
in this section , we disscuss applications of our results .
[ proof of corollary [ cor : shadow complexity 0 ] ] let @xmath4 be an acyclic @xmath2-manifold with shadow complexity zero , and let @xmath0 be a shadow of @xmath4 without true vertices
. then each connected component of @xmath13 is @xmath89 or a closed interval . if @xmath13 consists of only circles , then @xmath0 collapses onto @xmath24 by theorem [ thm : no vertex ] . by theorem [ thm : s - collapse ] ,
a disk @xmath24 is a shadow of @xmath22 , and then @xmath22 is diffeomorphic to @xmath25 .
assume that @xmath13 has a closed interval component .
we decompose @xmath0 into subpolyhedra @xmath143 along all closed interval components of @xmath13 . from turaev
s reconstruction @xcite , the boundary connected sum @xmath144 and @xmath145 are diffeomorphic . for @xmath146 ,
the 4-manifold @xmath147 is diffeomorphic to @xmath25 as mentioned above since @xmath148 consists of only circles .
it follows that @xmath149 .
we have another application to the study of 3-manifolds .
let @xmath150 be a closed connected @xmath3-manifold .
a shadow of @xmath150 is defined as a shadow of a @xmath2-manifold whose boundary is @xmath150 .
costantino and thurston indicated in @xcite that the stein factorization of a stable map on @xmath150 to @xmath6 can be seen as a shadow of @xmath150 with a certain modification if necessary .
the notion of complexity of stable maps was introduced by ishikawa and koda in @xcite .
especially , the complexity of a stable map is zero if and only if the map has no singular fiber of type @xmath151 and type @xmath152 .
see @xcite for the precise definitions .
these observations and our results immediately yield the following : [ rmk : stein fac ] let @xmath150 be a closed connected @xmath3-manifold .
then @xmath150 admits a stable map with complexity zero and acyclic stein factorization if and only if @xmath150 is homeomorphic to @xmath153 . 99 a. carrega and b. martelli , _ shadows , ribbon surfaces , and quantum invariants _ , arxiv preprint arxiv:1404.5983 , to appear in quantum topology . f. costantino , _ complexity of 4-manifolds _ , experiment . math . *
15 * ( 2006 ) , 237-249 .
f. costantino , _ stein domains and branched shadows of 4-manifolds _ , geom .
dedicata * 121 * ( 2006 ) , 89-111 . f. costantino , _ branched shadows and complex structures on 4-manifolds _ , j. knot theory ramifications * 17 * ( 2008 ) , 14291454 .
f. costantino and d. thurston , _
3-manifolds efficiently bound 4-manifolds _ , j. topol . * 1 * ( 2008 ) , 703745 . a. hatcher , _ pants decompositions of surfaces _
, arxiv:9906084 .
m. hirsch and b. mazur , _ smoothings of piecewise linear manifolds _ , ann .
studies 80 , princeton university press , princeton nj 1974 .
h. ikeda , _ acyclic fake surfaces _ , topology * 10 * ( 1971 ) , 936
. m. ishikawa and y. koda , _
stable maps and branched shadows of 3-manifolds _ , arxiv preprint arxiv:1403.0596 , to appear in math .
b. martelli , _ four - manifolds with shadow - complexity zero _ , int .
res . not .
imrn * 2011 * , 12681351 . c. rourke and b. sanderson , _ introduction to piecewise - linear topology _ , ergebnisse der mathematik , band 69 , springer - verlag , new york - heidelberg , 1972 .
o. saeki , _ topology of singular fibers of differentiable maps _ , lecture notes in mathematics 1854 , springer - verlag , berlin , 2004 .
v. g. turaev , _ quantum invariants of knots and 3-manifolds _ , de gruyter studies in math . , 18 , walter de gruyter & co. , berlin , 1994 . | our purpose is to classify acyclic 4-manifolds having shadow complexity zero . in this paper , we focus on simple polyhedra and discuss this problem combinatorially .
we consider a shadowed polyhedron @xmath0 and a simple polyhedron @xmath1 that is obtained by collapsing from @xmath0 .
then we prove that there exists a canonical way to equip internal regions of @xmath1 with gleams so that two 4-manifolds reconstructed from @xmath1 and @xmath0 are diffeomorphic .
we also show that any acyclic simple polyhedron whose singular set is a union of circles can collapse onto a disk . as a consequence of these results
, we prove that any acyclic 4-manifold having shadow complexity zero with boundary is diffeomorphic to a @xmath2-ball . |
the photon collider @xcite option of the next generation linear colliders ( lc ) @xcite offers the opportunity to study with high precision the physics of gauge bosons of the standard model with sensitivity to effects coming from new physics .
high energy photons are produced by compton back - scattering of laser photons from high energy electron ( or positron ) beams : they will not be monochromatic but will have an energy spectrum .
the high energy part of this spectrum will mainly include photons with definite helicity @xmath5 .
the sm cross section of @xmath6 at energies greater than @xmath7 gev is about @xmath8 pb and remains constant up to higher energies of interest @xcite .
this large cross section will ensure very high event rates .
when initial photons are circularly polarized one expects different momentum distributions between the positively and negatively charged leptons from the decay of the @xmath9 gauge bosons which is referred to as a _ charge asymmetry_. a qualitative discussion of the mechanism originating charge asymmetries is given in sec .
[ sec2 ] .
one expects that a study of charge asymmetries will be a sensitive tool to study new physics effects .
therefore , all mechanisms leading to this asymmetry should be investigated in detail . in the following
we study the charge asymmetry in the standard model ( sm ) and make preliminary considerations on how these asymmetries change due to some possible effects of new physics .
numerical results ( including all plots ) have been obtained with the comphep package @xcite in a version which allows to set the initial photons in a definite helicity state and to implement realistic photon energy spectra .
the following cuts are applied below : ( 1 ) an angular cut on the muons scattering angles given by @xmath10 , with @xmath11 mrad ( corresponding to the tesla detector angular limitations ) ; ( 2 ) a cut on muons transverse momentum @xmath12 gev , both on each muon and on the couple of muons .
these two cuts help to suppress the background .
we discuss two effects of charge asymmetry : + @xmath13 asymmetry of @xmath0 and @xmath1 momenta in each event , averaged over events in the process @xmath14 where both muons are recorded .
+ @xmath13 difference in distributions of @xmath0 and @xmath1 in the processes @xmath15 , @xmath16 , where we assume @xmath9 reconstruction via effective mass of two jets ( @xmath17 mode ) or its lepton decay .
most of the numerical results presented concern this last approach .
@xmath18 in the following we mark polarization of an initial state , e.g. @xmath19 , that means @xmath20 , @xmath21 , where @xmath22 and @xmath23 are helicities of photons moving in the positive and negative direction of the collision axis ( @xmath24 axis ) respectively .
@xmath13 in sm , at the tree level , the process @xmath25 is described by 19 diagrams .
we subdivide them in 5 types , shown in fig . 1 ( a similar classification was given also in ref . @xcite ) .
the collection of diagrams within each type is obtained from those shown in the figure with the exchange @xmath26 , @xmath27 and permutations of photons . in the estimates of cross sections
we denote @xmath28 and @xmath29 .
.,width=264,height=302 ] * 3 double - resonant diagrams ( drd ) of fig . [ diag](1 ) describe @xmath30 production and decay .
their contribution to the total cross section can be evaluated as @xmath31 , numerically it accounts for @xmath32% of the cross section . * 4 single - resonant diagrams of fig .
[ diag](2 ) with @xmath9 exchange in @xmath33channel .
the corresponding contribution is estimated as @xmath34 . * 4 single resonant diagrams with @xmath35 exchange in @xmath33channel ( gauge boson bremsstrahlung ) , fig .
[ diag](3 ) .
their contribution to the total cross section is estimated as : @xmath36 ; numerically they provide less then 1% . * 6 diagrams with radiation of @xmath37 boson in the process @xmath38 , fig .
[ diag](4 ) give asymptotic contribution @xmath39 ; numerically they amount less then 1% . *
2 non - resonant diagrams fig .
[ diag](5 ) with @xmath40 ; numerically they amount less then 1% .
our analytical estimates are based on the equations for @xmath41 processes at @xmath42 , assuming for the sm gauge couplings @xmath43 .
numerical estimates are obtained at @xmath44 gev with comphep in the feynman gauge for the squared contributions of each group .
the value of the interference with the dominant drd is roughly the same , since drd are large only in two regions of the final phase space corresponding to @xmath9 resonances .
the other contributions do not have these two peaks .
the asymmetry is present in the double - resonant contribution .
there are no reasons to expect the other diagrams to contribute to the asymmetry more than their relative contributions to the cross section .
note that the contribution of diagrams of group 5 is at the level of radiative corrections to the dominant contribution .
the precision provided by including this contribution is beyond the accuracy of the tree approximation .
@xmath13 the process @xmath15 is described by only the first 3 groups of diagrams .
_ therefore , in the following qualitative discussion we refer to the double resonant contribution while all numerical results and plots are obtained for the complete set of 19 diagrams .
_ let us describe the expected picture for double resonant contribution .
( _ i _ ) at @xmath4 gev the cross section practically does not depend on photon polarizations .
( _ ii _ ) in the process , the @xmath9 gauge bosons are distributed mainly around the forward and backward directions with mean transverse momentum @xmath45 ( @xmath46 @xcite ) .
( _ iii _ ) in this process helicity conservation holds approximately ( see helicity amplitudes in @xcite ) .
the helicity of @xmath47 moving in the positive direction of @xmath24 axis @xmath48 , while @xmath49 , is independent on the charge sign of @xmath9 .
( _ iv _ ) let the @xmath50axis be directed along the @xmath9 momentum ( @xmath51 ) and @xmath52 and @xmath53 are the energy and the longitudinal momentum of @xmath35 in the @xmath9 rest frame .
it is straightforward to show that the distribution of muons from the decay of the @xmath9 gauge boson with charge @xmath54 and helicity @xmath55 in its rest frame is : momentum within the interval @xmath56 . ]
@xmath57 . in other words ,
the distribution of muons from the decay of @xmath47 has a peak along @xmath51 if @xmath58 and opposite to @xmath51 when @xmath59 .
these distributions are boosted easily to the @xmath60 collision frame .
for example when photons have a @xmath61 helicity state , produced @xmath0 are distributed around the upper value of their longitudinal momentum ( both in the forward and backward direction ) , while produced @xmath1 are concentrated near the zero value of their longitudinal momentum .
this boost makes the distribution in @xmath62 wider in the first case and narrower in the second case .
we present here the first results , reporting the distributions of muons @xmath63 ( with @xmath64 the component of @xmath65 parallel to the collision axis ( taken to be the @xmath24 axis ) and @xmath66 the transverse momentum ) .
[ energyvar ] shows such muon distributions in the @xmath67 plane for a @xmath61 initial photon polarization state with monochromatic photons at different values of the energy .
[ energyvar ] clearly shows the charge asymmetry : _ a strong difference in the distributions of @xmath1 and @xmath0 _ in accordance with qualitative description given above .
the absolute value of charge asymmetry effect decreases with energy due to the increasing importance of the applied cuts .
plane for @xmath68 helicity of colliding photons : @xmath1 on the left , @xmath0 on the right : @xmath69 gev ( top ) and @xmath70 gev ( bottom ) .
monochromatic beams.,title="fig:",width=264,height=264 ] plane for @xmath68 helicity of colliding photons : @xmath1 on the left , @xmath0 on the right : @xmath69 gev ( top ) and @xmath70 gev ( bottom ) .
monochromatic beams.,title="fig:",width=264,height=264 ] plane for @xmath68 helicity of colliding photons : @xmath1 on the left , @xmath0 on the right : @xmath69 gev ( top ) and @xmath70 gev ( bottom ) .
monochromatic beams.,title="fig:",width=264,height=264 ] plane for @xmath68 helicity of colliding photons : @xmath1 on the left , @xmath0 on the right : @xmath69 gev ( top ) and @xmath70 gev ( bottom ) .
monochromatic beams.,title="fig:",width=264,height=264 ] * in all figures and tables below we consider @xmath71 gev for the monochromatic case and @xmath72 gev , @xmath73 for `` realistic '' spectra . * in fig
. [ mm ] we show two - dimensional level lines , for different initial helicity states and monochromatic photons . plane ( pb / bin ) . left
@xmath1 , right
@xmath0 ; from the top : @xmath74 , @xmath75 helicities of colliding photons .
monochromatic beams . , title="fig:",width=264,height=264 ] plane ( pb / bin ) . left
@xmath1 , right @xmath0 ; from the top : @xmath74 , @xmath75 helicities of colliding photons .
monochromatic beams . , title="fig:",width=264,height=264 ] plane ( pb / bin ) . left
@xmath1 , right
@xmath0 ; from the top : @xmath74 , @xmath75 helicities of colliding photons .
monochromatic beams . , title="fig:",width=264,height=264 ] plane ( pb / bin ) . left
@xmath1 , right
@xmath0 ; from the top : @xmath74 , @xmath75 helicities of colliding photons .
monochromatic beams . ,
title="fig:",width=264,height=264 ] note that the scales of effect in the graphs with @xmath1 for @xmath61 case differ from other cases by a factor of seven . due to cp conservation ,
the @xmath76 distributions for @xmath61 case coincide with @xmath77 distribution for @xmath78 case .
the obtained distributions in the @xmath79 plane have the form which corresponds to the qualitative picture outlined in sec .
[ sec2 ] . to make a more transparent analysis
, we consider the normalized mean values of longitudinal and transverse momenta of @xmath1 and @xmath0 , determined considering the longitudinal momentum of @xmath35 in the forward hemisphere ( @xmath80 ) , and take their relative difference as a measure of the charge asymmetry : & p_l+^= , p_t+^= , & [ plt ] & _ l= , _
t=.&[chalt ] we have calculated these quantities for the process @xmath81 at @xmath44 gev @xmath82 gev for monochromatic case and for @xmath72 gev @xmath83 gev for the `` realistic '' photon spectra . in tables below
the parameter n takes two values n = l ( longitudinal ) and n = t ( transverse ) .
table 1 presents the values of the asymmetries for different photon polarizations in the monochromatic case .
the @xmath78 case coincides with the @xmath61 one with the exchange @xmath84 . for @xmath19 initial state
the pictures in forward and backward hemispheres are different .
the latter coincide with that for the forward hemisphere for @xmath85 initial state with the exchange @xmath86 .
& @xmath87 & @xmath88 @xmath61 n = l & 0.605 & 0.191 & 0.520n = t & 0.353 & 0.160 & 0.374@xmath19
n = l & 0.229 & 0.571 & -0.427n = t & 0.167 & 0.262 & -0.221@xmath85 n = l & 0.233 & 0.631 & -0.461n = t & 0.179 & 0.283 & -0.223 [ tabmm500 ] one can see that the values of the asymmetry is typically 20 - 50% . that is _ a huge and easily observable effect . _
note that the scale of longitudinal distributions is given by the initial photon energy while the scale for transverse distributions is given by the @xmath9 mass .
thus , the transverse asymmetries are smaller than the longitudinal ones .
@xmath13 * effect of photon non monochromaticity .
* photon beams at the photon collider will be non - monochromatic .
the @xmath61 n = l & 0.499&0.150&0.536n = t & 0.382&0.178&0.364@xmath19 n = l & 0.187&0.451&-0.415n = t & 0.198&0.272&-0.160@xmath85 n = l & 0.205&0.514&-0.436n = t&0.219 & 0.300&-0.158 [ tabss500 ] energy spectra are well known for high energies where the approximation of ref .
@xcite works well but the low - energy part will differ depending on technological details .
fortunately , these photons are unpolarized and do not contribute to the asymmetry .
so that , in all tables 24 we used `` realistic '' photon spectra with @xmath89 from ref .
@xcite for the high energy part , and from ref .
@xcite with @xmath90 for the low energy part ( similar to that it was done in ref .
we assume here @xmath73 for @xmath91 conversion which corresponds to the maximal value of photon energy @xmath92 gev at @xmath72 gev .
the resulting asymmetries are shown in the table 2 .
the longitudinal asymmetry changes only weakly in comparison with the monochromatic case , while the transverse asymmetry changes strongly .
we verified that in the @xmath67 plane , relative to the monochromatic case , one can see a noticeable deviation in the phase space regions responsible for the asymmetry .
@xmath13 next , we consider the effect of * possible anomalous interactions of @xmath9bosons * , parameterized by the usual anomalous magnetic moment @xmath93 and the quadrupole moment @xmath94 in the trilinear and quartic vertices ( with the same & & @xmath61 l & 0.498&0.150&0.537&&t & 0.379&0.182&0.351 0.1&0&@xmath19 l & 0.186&0.451&-0.416 & & t & 0.200&0.272&-0.152 & & @xmath85 l & 0.204&0.510&-0.429&&t & 0.221&0.300&-0.152 & & @xmath61 l & 0.491&0.151&0.457&&t & 0.388&0.177&0.378 0&0.1&@xmath19 l & 0.190&0.457&-0.413&&t & 0.192&0.270&-0.169 & & @xmath85 l & 0.206&0.522&-0.438&&t&0.212 & 0.301&-0.173 [ tabano500 ] normalization as in ref .
we calculated the asymmetries defined in eq .
( 1 ) for large enough values of @xmath93 and @xmath94 .
the results are shown in the table 3 .
comparing with table 2 , one can see that the anomalous magnetic moment @xmath95 changes the quantities @xmath96 only weakly .
one should study , whether any other variable is better to investigate the effects of this anomaly . on the other hand ,
the quadrupole moment @xmath94 changes @xmath97 significantly at least in the @xmath61 case . therefore
, this asymmetry can be useful for the study of the @xmath94 anomaly .
@xmath13 * possible new particles .
* to mimic the effect of new interactions and/or new particles , we consider a toy model with a `` muon '' having a mass of 40 gev .
the results are shown in table 4 . comparing with table 2 ,
we conclude that the study of charge asymmetry can be a useful tool for the discovery of new particles .
@xmath61 n = l & 0.519 & 0.287 & 0.289n = t & r 0.385&0.236&0.240@xmath19 n = l & 0.325&0.569&-0.271n = t & 0.222&0.310&-0.201@xmath85 n = l & 0.295&0.545&-0.250n = t & 0.216&0.308&-0.175 [ tabmuon ]
the charge asymmetry in relative distributions of positive and negative muons can be a more useful instrument to hunt for new physics .
the first problem here is to find some representative variables in the 5dimensional @xmath98 phase space . for this purpose
we consider the @xmath99 process with monochromatic photons . ,
monochromatic photons,,width=264,height=264 ] fig .
[ kdistr ] presents the level curves of the distribution in @xmath100 , in its longitudinal and transverse component . in the _ absence _ of charge asymmetry this distribution
would be centered around the point @xmath101 .
clearly the charge asymmetry is quite large . and @xmath102 for the ( - - ) case .
bottom panels : distribution in @xmath103 for ( - - ) and ( -+ ) initial helicities.,title="fig:",width=264,height=264 ] and @xmath102 for the ( - - ) case .
bottom panels : distribution in @xmath103 for ( - - ) and ( -+ ) initial helicities.,title="fig:",width=264,height=264 ] and @xmath102 for the ( - - ) case .
bottom panels : distribution in @xmath103 for ( - - ) and ( -+ ) initial helicities.,title="fig:",width=264,height=264 ] and @xmath102 for the ( - - ) case .
bottom panels : distribution in @xmath103 for ( - - ) and ( -+ ) initial helicities.,title="fig:",width=264,height=264 ] besides , we study various `` natural '' dimensionless variables . in particular , we find useful to present distributions in following variables , figure [ vvn ] .
c v=,u= , + vvn=. fig .
[ mm ] shows that in the cases @xmath61 , @xmath78 distributions in the forward and backward hemispheres are identical . therefore , in this case the variables @xmath104 and @xmath102 are useful while @xmath103 can not be useful .
vice versa , for the @xmath19 case the variable @xmath103 is useful while @xmath104 and @xmath102 are not .
the choice of some other variables and sensitivity to new physics effects will be studied elsewhere .
@xmath13 the effects considered so far are identical for electrons and muons .
so that , absolutely the same asymmetry will be observed in @xmath105 , @xmath106 , @xmath107 distributions .
therefore , all these contributions should be gathered for a complete analysis .
this will enhance the value of cross section for @xmath108 from 0.9 to 3.7 pb and for @xmath15 , etc . to 23.5 pb ( millions of events per year ) .
@xmath13 the contribution to the observable final state is given not only by the process @xmath108 , but also by processes @xmath109 pairs .
the most important of them will be , for example , @xmath110 , etc . with subsequent decay @xmath111 .
the cross section of this process is 17% of those discussed above ( @xmath112 @xcite ) + 17% for the case with the change @xmath113 , etc .
+ 3% for @xmath114 .
unfortunately , at the moment there are no regular methods for precise calculation of such processes like comphep or whizard ( 6 or 8 particles in final state and very narrow intermediate state ) .
we hope that the corresponding algorithm will be developed in a reasonable time using the fact that @xmath115 is a very narrow particle .
this work is partially supported by grants rfbr 02 - 02 - 17884 , nsh-2339.2003.2 , intas 00 - 00679 and grant 015.02.01.16 russian universities and by the european union under contract n. hpmf - ct-2000 - 00752 .
i. f. ginzburg acknowledges support from the landau network
centro volta ( como ) italy , and from infn sezione di perugia .
i. f. ginzburg , g. l. kotkin , v. g. serbo and v. i. telnov , nucl .
instrum .
meth . 205 ( 1983 ) 47 ;
+ i. f. ginzburg , g. l. kotkin , s. l. panfil , v. g. serbo and v. i. telnov , nucl .
. meth . a 219 ( 1984 ) 5 .
b. badelek _ et al . _
[ ecfa / desy photon collider working group collaboration ] , hep - ex/0108012 .
ginzburg , g.l .
kotkin , s.l .
panfil , v.g .
serbo , nucl .
b 228 ( 1983 ) 285 . | it is shown that the difference in the distributions of positive ( @xmath0 ) and negative charged leptons ( @xmath1 ) in reactions @xmath2 and @xmath3 at @xmath4 gev leads to observable _ charge asymmetry _ of muons which is sensitive to new physics effects . |
let @xmath0 be a nonsingular compact connected real algebraic surface , i.e. @xmath0 is a nonsingular compact connected real algebraic subset of some @xmath9 of dimension @xmath10 . recall that @xmath0 is _ rational _ if the field of rational functions @xmath11 of @xmath0 is a purely transcendent field extension of @xmath12 of transcendence degree @xmath10 .
more geometrically , @xmath0 is rational if there are nonempty zariski open subsets @xmath13 and @xmath14 of @xmath15 and @xmath0 , respectively , such that there is an isomorphism of real algebraic varieties|in the sense of @xcite|between @xmath13 and @xmath14 . loosely speaking
, @xmath0 is rational if a nonempty zariski open subset of @xmath0 admits a rational parametrization by a nonempty zariski open subset of @xmath15 ( the last condition imposes a priori only that @xmath0 is unirational but in dimension 2 , unirationality implies rationality ) .
a typical example of a rational compact real algebraic surface is the unit sphere @xmath16 in @xmath17 .
a rational parametrization in that case is the inverse stereographic projection .
it has recently been shown that any rational nonsingular compact real algebraic surface is isomorphic either to the real algebraic torus @xmath18 , or to a real algebraic surface obtained from the real algebraic sphere @xmath16 by blowing up a finite number of points @xcite . in the sequel
, it will be convenient to identify the real algebraic surface @xmath0 with the affine scheme @xmath19 , where @xmath20 denotes the @xmath12-algebra of all algebraic|also called regular|functions on @xmath0 @xcite .
a real - valued function @xmath21 on @xmath0 is _ algebraic _ if there are real polynomials @xmath22 and @xmath23 in @xmath24 such that @xmath23 does not vanish on @xmath0 and such that @xmath25 on @xmath0 .
the algebra @xmath20 is the localization of the coordinate ring @xmath26/i$ ] with respect to the multiplicative system of all polynomials that do not vanish on @xmath0 , where @xmath27 denotes the vanishing ideal of @xmath0 .
it is the subring of @xmath11 of all rational functions on @xmath0 that do not have any poles on @xmath0 .
thanks to the above convention , we can define a _
curvilinear subscheme _ of @xmath0 to be a closed subscheme @xmath28 of @xmath0 that is isomorphic to @xmath29/(x^e)$ ] , for some nonzero natural integer @xmath30 .
we call @xmath30 the _ length _ or the _ order _ of @xmath28
. let @xmath28 be a curvilinear subscheme of @xmath0 .
the reduced scheme @xmath31 is an ordinary point of @xmath0 .
a curvilinear subscheme of @xmath0 is also called a _ curvilinear infinitely near point over @xmath31 _ or an _ infinitesimal arc _ based on @xmath31 .
a curvilinear infinitely near point of length @xmath32 is an ordinary point of @xmath0 .
a curvilinear infinitely near point of length @xmath10 is a pair @xmath33 , where @xmath28 is a point of @xmath0 and @xmath34 is @xmath32-dimensional subspace of the tangent space @xmath35 of @xmath0 at @xmath28 .
equivalently , a curvilinear infinitely near point @xmath28 of @xmath0 of length @xmath10 is a point of the exceptional divisor of the real algebraic surface obtained by blowing up @xmath0 in an ordinary point . by induction
, a curvilinear infinitely near point of @xmath0 of length @xmath30 is a curvilinear infinitely near point of length @xmath36 on the exceptional divisor @xmath37 of a blow - up of @xmath0 ( cf .
@xcite ) .
let @xmath28 and @xmath38 be curvilinear subschemes of @xmath0 .
we say that @xmath28 and @xmath38 are _ distant _ if the points @xmath31 and @xmath39 of @xmath0 are distinct .
let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , where @xmath5 is some natural integer . denote by @xmath4 the set of @xmath5-tuples @xmath6 of mutually distant curvilinear subschemes @xmath40 of @xmath0 of orders @xmath41 , respectively .
recall that an _ algebraic automorphism _ of @xmath0 is a bijective map @xmath21 from @xmath0 into itself such that all coordinate functions of @xmath21 and @xmath42 are algebraic functions on @xmath0 @xcite .
denote by @xmath1 the group of algebraic automorphisms of @xmath0 into itself .
equivalently , @xmath1 is the group of @xmath12-algebra automorphisms of @xmath20 .
one has a natural action of @xmath1 on @xmath4 .
one of the main results of the paper is the following .
[ thtrans ] let @xmath0 be a nonsingular rational compact real algebraic surface .
let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , for some natural integer @xmath5
. then the group @xmath1 acts transitively on @xmath4 . roughly speaking , theorem [ thtrans ]
states that the group @xmath1 acts @xmath5-transitively on curvilinear infinitely near points of @xmath0 , for any @xmath5 .
the statement generalizes earlier work where @xmath43-transitivity was proved for ordinary points only , i.e. , in case of the trivial partition @xmath8 $ ] ( cf .
@xcite ) .
the statement of theorem [ thtrans ] motivates the following question .
let @xmath0 be a nonsingular rational compact real algebraic surface .
is the subset @xmath1 of algebraic automorphisms of @xmath0 dense in the set @xmath44 of all diffeomorphisms of @xmath0 ?
equivalently , can any diffeomorphism of @xmath0 be approximated by algebraic automorphisms ?
this question is studied in the forthcoming paper @xcite .
the problem of approximating smooth maps between real algebraic varieties by algebraic maps has been studied by numerous authors @xcite .
it should be noted that theorem [ thtrans ] does not seem to follow from the known @xmath2-transitivity of @xmath1 on ordinary points .
the difficulty is that if two @xmath2-tuples @xmath28 and @xmath38 of ordinary distinct points of @xmath0 tend to two @xmath5-tuples of mutually distant curvilinear infinitely near points with lengths @xmath41 , then the algebraic diffeomorphisms mapping @xmath28 to @xmath38 do not necessarily have a limit in @xmath1 .
our proof of theorem [ thtrans ] goes as follows .
first we show that the statement of theorem [ thtrans ] is valid for the real algebraic surfaces @xmath16 and @xmath18 by explicit construction of algebraic automorphisms ( see theorems [ thsphere ] and [ thtorus ] ) .
this step does use the earlier work mentioned above . then
, we use the fore - mentioned fact that an arbitrary nonsingular rational compact real algebraic surface is either isomorphic to @xmath18 , or to a real algebraic surface obtained from @xmath16 by blowing up a finite number of distinct ordinary points ( * ? ? ?
* theorem 4.3 ) . in order to give an application of theorem [ thtrans ]
, we need to recall the following .
let @xmath0 be a nonsingular rational compact connected real algebraic surface , and let @xmath28 be a curvilinear infinitely near point of @xmath0 .
the _ blow - up _ of @xmath0 at @xmath28 is the blow - up @xmath45 of @xmath0 at the sheaf of ideals defined by the closed subscheme @xmath28 .
explicitly , if @xmath28 is defined by the ideal @xmath46 on the real affine plane @xmath15 , then the blow - up of @xmath15 at @xmath28 is the real algebraic sub - variety of @xmath47 defined by the equation @xmath48 , where @xmath49 are homogeneous coordinates on the real projective line @xmath50 .
the blow - up @xmath45 is also called a _ weighted blow - up _ , for obvious reasons . if @xmath51 , the blow - up @xmath45 is the ordinary blow - up of @xmath0 at @xmath28 .
if @xmath52 then the blow - up @xmath45 has a singular point .
a local equation of the singularity is @xmath53 in @xmath17 .
this is often called a singularity _ of type @xmath54 _ ( see e.g. ( * ? ? ?
* definition 2.1 ) ) .
note that weighted blow - ups recently turned out to have several applications in real algebraic geometry ( see @xcite ) .
we apply our results , and study singular real rational surfaces that are obtained from nonsingular ones by performing weighted blow - ups .
the latter surfaces get horns and appear as rather diabolic to us , and will be referred to , for brevity , by the following term .
a singular real compact surface @xmath0 is _ dantesque _ if it is obtained from a nonsingular real algebraic surface @xmath55 by performing a finite number of weighted blow - ups .
the following statement implies that any rational dantesque surface is obtained from @xmath18 or @xmath16 by blowing - up a finite number of mutually distant curvilinear subschemes of @xmath18 or @xmath16 , respectively .
note that in particular , such a surface is connected .
[ thnonsuccessive ] let @xmath0 be a rational dantesque surface .
then * there are mutually distant curvilinear subschemes @xmath40 on @xmath18 such that @xmath0 is isomorphic to the real algebraic surface obtained from @xmath18 by blowing up @xmath40 , or * there are mutually distant curvilinear subschemes @xmath40 on @xmath16 such that @xmath0 is isomorphic to the real algebraic surface obtained from @xmath16 by blowing up @xmath40 . on a singular surface ,
a curvilinear infinitely near point @xmath28 is _ nonsingular _ if @xmath31 is a nonsingular point .
let @xmath0 be a rational dantesque surface .
let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , where @xmath5 is some natural integer . denote by @xmath4 the set of @xmath5-tuples @xmath6 of mutually distant nonsingular curvilinear subschemes @xmath40 of @xmath0 of orders @xmath41 , respectively .
denote again by @xmath1 the group of algebraic automorphisms of the possibly singular real algebraic surface @xmath0 .
note that the definition of algebraic automorphism above makes perfectly sense for singular varieties .
alternatively , one can define an algebraic automorphism of @xmath0 to be an automorphism of @xmath19 .
anyway , one has a natural action of @xmath1 on @xmath4 .
we have the following generalization of theorem [ thtrans ] above . [ thtranssing ]
let @xmath0 be a rational dantesque surface .
let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , for some natural integer @xmath5
. then the group @xmath1 acts transitively on @xmath4 . as an application of theorem [ thtrans ] and theorem [ thnonsuccessive ] , we prove the following statement .
[ thiso ] let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , for some natural integer @xmath5 .
let @xmath0 and @xmath55 be two rational dantesque surfaces .
assume that each of the surfaces @xmath0 and @xmath55 contains exactly one singularity of type @xmath56 for each @xmath57
. then @xmath0 and @xmath55 are isomorphic as real algebraic surfaces if and only if they are homeomorphic as singular topological surfaces .
theorem [ thiso ] generalizes to certain singular real rational surfaces an earlier result for nonsingular ones ( ( * ? ? ?
* theorem 1.2 ) and ( * ? ? ? * theorem 1.5 ) ) .
we will show by an example that the statement of theorem [ thiso ] does not hold for the slightly more general class of real rational compact surfaces that contain singularities of type @xmath58 ( see example [ exiso ] ) .
we want to thank l. evain , j. kollr , and d. naie for useful discussions .
the object of this section is to prove theorem [ thtrans ] in the case of the real algebraic torus : [ thtorus ] let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , for some natural integer @xmath5 .
the group @xmath59 acts transitively on @xmath60 .
the above statement is a generalization of the following statement , that we recall for future reference .
[ thntorus ] let @xmath2 be a natural integer .
the group @xmath59 acts @xmath2-transitively on @xmath18 . for the proof of theorem [ thtorus ]
, we need several lemmas .
it will turn out to be convenient to replace @xmath61 by the isomorphic real projective line @xmath50 .
[ lephitorus ] let @xmath62 $ ] be real polynomials in @xmath63 of the same degree .
suppose that @xmath23 does not have any real roots .
define @xmath64 by @xmath65 then @xmath66 is an algebraic automorphism of @xmath67 into itself .
it suffices to prove that @xmath66 is an algebraic map .
we write @xmath66 in bi - homogeneous coordinates : @xmath68,[y_0:y_1])= \left([x_0:x_1 ] , [ \overline{q}(x_0,x_1)y_0+\overline{p}(x_0,x_1)y_1 : \overline{q}(x_0,x_1)y_1 ] \right),\ ] ] where @xmath69 and @xmath70 are the homogenizations of @xmath22 and @xmath23 , respectively .
since @xmath23 has no real zeros , the homogeneous polynomial @xmath70 does not vanish on @xmath50 .
therefore , if @xmath71 then @xmath72 and @xmath73 .
it follows that @xmath66 is a well defined algebraic map from @xmath67 into itself .
let @xmath28 be a curvilinear infinitely near point of a nonsingular compact real algebraic surface @xmath0 .
if @xmath74 is an ordinary point of @xmath0 , then @xmath28 is said to be _ infinitely near _ to @xmath74 if @xmath75 .
let @xmath28 be a curvilinear infinitely near point of @xmath18 .
we say that @xmath28 is _ vertical _ if @xmath28 is tangent to a vertical fiber @xmath76 , for some @xmath77 , i.e. if the scheme - theoretic intersection @xmath78 is not reduced .
[ lenonvertical ] let @xmath40 be mutually distant curvilinear subschemes of @xmath79 .
then there is an algebraic diffeomorphism @xmath66 of @xmath67 such that * @xmath80 is infinitely near to the point @xmath81 of @xmath67 , and * @xmath80 is not vertical , for all @xmath82 . by theorem [ thntorus ] , we may assume that @xmath40 are infinitely near to the points @xmath83 of the real algebraic torus @xmath67 , respectively .
let @xmath84 be a tangent vector to @xmath67 at @xmath85 that is tangent to @xmath85 .
this means the following .
if @xmath85 is an ordinary point then @xmath86 . if @xmath85 is not an ordinary point then @xmath87 and the @xmath88-dimensional subscheme of @xmath67 of length @xmath10 defined by @xmath89 is contained in the closed subscheme @xmath28 of @xmath67 .
let @xmath90 $ ] be real polynomials in @xmath91 of the same degree such that * @xmath23 does not have any real roots , * @xmath92 , @xmath93 , @xmath94 and * @xmath95 whenever @xmath87 .
define @xmath96 by @xmath97 according to lemma [ lephitorus]|exchanging @xmath63 and @xmath91|the map @xmath66 is an algebraic automorphism of @xmath67 .
since @xmath92 , one has @xmath98 .
it follows that @xmath80 is also infinitely near to @xmath81 .
the jacobian of @xmath66 at @xmath81 is equal to @xmath99 by construction , @xmath100 has first coordinate non zero whenever @xmath87 . therefore , @xmath80 is not vertical , for all @xmath82 .
let @xmath40 be mutually distant curvilinear subschemes of the real algebraic torus @xmath67 of orders @xmath41 , respectively .
let @xmath101 be the curvilinear infinitely near point of @xmath67 defined by the ideal @xmath102 in @xmath103 .
it suffices to show that there is an algebraic automorphism @xmath66 of @xmath67 such that @xmath104 for all @xmath82 . by lemma [ lenonvertical ]
, we may assume that the curvilinear infinitely near point @xmath85 is infinitely near to @xmath81 and that @xmath80 is not vertical .
it follows that @xmath85 is defined by an ideal of the form @xmath105 where @xmath106 $ ] .
let @xmath62 $ ] be of the same degree such that * @xmath23 does not have any real roots , * @xmath107 modulo @xmath108 for all @xmath82 .
such polynomials abound by the chinese remainder theorem . by lemma [ lephitorus ] , the polynomials @xmath22 and @xmath23
give rise to an algebraic automorphism @xmath66 of @xmath67 defined by @xmath65 in order to show that @xmath104 for all @xmath82 , we compute @xmath109 and @xmath110 modulo @xmath108 .
indeed , @xmath23 is invertible modulo @xmath108 , and @xmath107 modulo @xmath108 , by construction of @xmath22 and @xmath23 .
it follows that @xmath104 .
the object of this section is to prove theorem [ thtrans ] in the case of the real algebraic sphere @xmath16 : [ thsphere ] let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 , for some natural integer @xmath5 .
the group @xmath111 acts transitively on @xmath112 .
the above statement is a generalization of the following statement , that we recall for future reference .
[ thnsphere ] let @xmath2 be a natural integer .
the group @xmath111 acts @xmath2-transitively on @xmath16 . for the proof of theorem [ thsphere ]
, we need several lemmas .
[ lephisphere ] let @xmath113 $ ] be such that * @xmath114 does not have any roots in the interval @xmath115 $ ] , and * @xmath116 .
define @xmath117 by @xmath118 then @xmath66 is an algebraic automorphism of @xmath16 .
let @xmath28 be infinitely near to a point of the equator @xmath119 of @xmath16 .
we say that @xmath28 is _ vertical _ if @xmath28 is tangent to the great circle of @xmath16 passing through the north pole .
as for the torus above , we need some standard points on @xmath16 .
let @xmath120 for @xmath57 .
note that @xmath121 and @xmath122 for all @xmath82 .
[ lenonhorizontal ] let @xmath40 be mutually distant curvilinear subschemes of @xmath16 .
then there is an algebraic automorphism @xmath66 of @xmath16 such that * @xmath80 is infinitely near to @xmath123 , and * @xmath80 is not vertical , for all @xmath82 . by theorem [ thnsphere ] , we may assume that @xmath85 is infinitely near to the point @xmath123 , for all @xmath82 .
let @xmath124 be a tangent vector to @xmath16 at @xmath85 that is tangent to @xmath85 .
let @xmath113 $ ] be such that * @xmath114 does not vanish on @xmath115 $ ] , * @xmath116 , * @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , and * @xmath130 or @xmath131 , whenever @xmath87 .
such polynomials abound .
take , for example , @xmath132 where @xmath133 is any real number such that @xmath134 or @xmath135 whenever @xmath87 .
define @xmath117 by @xmath136 according to lemma [ lephisphere]|permuting the roles of @xmath137|the map @xmath66 is an algebraic automorphism of @xmath16 . since @xmath125 , @xmath126 and @xmath127 , the curvilinear infinitely near point @xmath80 is again infinitely near to @xmath123 , for all @xmath82 .
the jacobian of @xmath66 at @xmath123 is equal to @xmath138 by construction , @xmath139 has first or second coordinate non zero whenever @xmath87 .
therefore , @xmath80 is not vertical , for all @xmath82 .
[ lemod ] let @xmath30 be a nonzero natural integer , and let @xmath140 .
let @xmath141/(x - x_i)^e$ ] be such that @xmath142 in @xmath143/(x - x_i)^e$ ] .
assume , moreover , that @xmath144 .
then there is @xmath145/(x - x_i)^e$ ] such that @xmath146 in @xmath143/(x - x_i)^e$ ] .
moreover , there is such an element @xmath147 such that @xmath148 is invertible in @xmath143/(x - x_i)^e$ ] . if @xmath149 then @xmath150 , and one can take @xmath151
therefore , we may assume that @xmath152 .
let @xmath153 be the valuation of @xmath154 , i.e. , @xmath155 , where @xmath156/(x - x_i)^e$ ] is invertible .
since @xmath157 , one has @xmath158 , i.e. , @xmath154 is not invertible in @xmath143/(x - x_i)^e$ ] and @xmath159 . by hensel
s lemma , there are lifts @xmath160 in @xmath143/(x - x_i)^{e+2d}$ ] of @xmath161 , respectively , satisfying the equations ( [ eqmod ] ) in the ring @xmath143/(x - x_i)^{e+2d}$ ] .
note that @xmath162 is invertible in @xmath143/(x - x_i)^{e+2d}$ ] , and that @xmath163 has valuation @xmath153 . in order to simplify notation
, we denote again by @xmath161 the elements @xmath160 , respectively .
let @xmath164/(x - x_i)^{e+2d}$ ] be the inverse of @xmath165 .
let @xmath166 .
we verify that equations ( [ eqmod1 ] ) hold and that @xmath148 is invertible in @xmath143/(x - x_i)^e$ ] .
the element @xmath148 is clearly invertible in @xmath143/(x - x_i)^{e+2d}$ ] since @xmath154 is not invertible .
since @xmath167 one has @xmath168 it follows that @xmath169 in @xmath143/(x - x_i)^{e+2d}$ ] , and therefore also in @xmath143/(x - x_i)^e$ ] . in order to prove that the other equation of ( [ eqmod1 ] ) holds as well ,
observe that @xmath170 by what we have seen above .
substituting @xmath171 , one obtains @xmath172 in @xmath143/(x - x_i)^{e+2d}$ ] .
since the valuation of @xmath154 is equal to @xmath153 , one deduces that @xmath173 in @xmath143/(x - x_i)^e$ ] .
hence , @xmath174 , as was to be proved .
let @xmath40 be mutually distant curvilinear subschemes of @xmath16 of orders @xmath41 , respectively .
let @xmath101 be the curvilinear infinitely near point of @xmath16 defined by the ideal @xmath175 in @xmath176 $ ] , where @xmath177 is the taylor polynomial in @xmath178 of @xmath179 at @xmath180 of order @xmath181 .
note that @xmath101 is infinitely near to @xmath123 for all @xmath82 .
we show that there is an algebraic automorphism @xmath66 of @xmath16 such that @xmath104 for all @xmath82 . by lemma [ lenonhorizontal ]
, we may assume that @xmath40 are infinitely near to the points @xmath182 of @xmath16 , respectively , and that they are not vertical .
it follows that @xmath85 is defined by an ideal of the form @xmath183 where @xmath184 $ ] are of degree @xmath185 .
moreover , since @xmath85 is a curvilinear infinitely near point of @xmath16 , we have @xmath186 by lemma [ lemod ] , there is @xmath187/(x - x_i)^{e_i}$ ] such that @xmath188 in @xmath143/(x - x_i)^{e_i}$ ] , and , moreover , @xmath189 is invertible . by the chinese remainder theorem , there is a polynomial @xmath145 $ ] such that @xmath190 , for all @xmath82 . then @xmath191 and @xmath148 is invertible in @xmath143/(x - x_i)^{e_i}$ ] , for all @xmath82 . put @xmath192 then @xmath193 and @xmath114 is invertible in @xmath143/(x - x_i)^{e_i}$ ] , for all @xmath82 .
moreover , * @xmath114 does not have any roots in the interval @xmath115 $ ] , and * @xmath116 . by lemma
[ lephisphere ] , the polynomials @xmath194 give rise to an algebraic automorphism @xmath66 of @xmath16 defined by @xmath118 in order to show that @xmath104 for all @xmath82 , we compute @xmath195 so that @xmath196 is the curvilinear infinitely near point of @xmath16 defined by the ideal @xmath197 .
we have @xmath198 and @xmath199 it follows that @xmath104 .
the proof of theorem [ thtrans ] is now similar to the proof of ( * ? ? ?
* theorem 1.4 ) . we include it for convenience of the reader .
let @xmath0 be a nonsingular real rational surface , and let @xmath200 and @xmath201 be two @xmath5-tuples in @xmath4 . as mentioned before , @xmath0 is isomorphic to @xmath18 or to the blow - up of @xmath16 at a finite number of distinct points @xmath202 .
if @xmath0 is isomorphic to @xmath18 then @xmath1 acts transitively on @xmath4 by theorem [ thtorus ] .
therefore , we may assume that @xmath0 is isomorphic to the blow - up @xmath203 of @xmath16 at @xmath202 .
moreover , we may assume that the points @xmath204 do not belong to any of the exceptional divisors , by ( * ? ? ?
* theorem 3.1 ) .
thus we can consider the @xmath205 as curvilinear subschemes of @xmath16 .
it follows that @xmath206 and @xmath207 are two @xmath208-tuples in @xmath209 , where @xmath210 $ ] . by theorem [ thsphere ]
, there is an automorphism @xmath211 of @xmath16 such that @xmath212 , for all @xmath82 , and @xmath213 , for all @xmath214 .
the induced automorphism @xmath66 of @xmath0 has the property that @xmath215 , for all @xmath214 .
the object of this section is to prove theorem [ thnonsuccessive ] that asserts that a real rational weighted blow - up surface is isomorphic to a real algebraic surface obtained from @xmath16 or @xmath18 by blowing up a finite number of mutually distant curvilinear subschemes .
the following is a particular case of ( * ? ? ?
* theorem 3.1 ) .
[ lekleinbottle ] let @xmath0 be a nonsingular rational real algebraic klein bottle let @xmath216 be a finite subset of @xmath0 .
then there is an algebraic map @xmath217 such that 1 .
@xmath21 is the blow - up of @xmath16 at @xmath10 distinct real points @xmath218 , and 2 .
@xmath219 , for @xmath220 .
[ leblow - uptorus ] let @xmath28 be a curvilinear infinitely near point of @xmath18 , and let @xmath221 be a real algebraic curve in @xmath18 such that there is a nonsingular projective complexification @xmath222 of @xmath18 having the following properties : 1 .
the zariski closure @xmath223 of @xmath221 in @xmath222 is nonsingular and rational , 2 .
the self - intersection of @xmath223 in @xmath222 is even and non - negative , 3 .
@xmath224 , and 4 .
@xmath223 is not tangent to @xmath28 , i.e. , the scheme - theoretic intersection @xmath225 is of length @xmath32 . then , there is an algebraic map @xmath226 that is a blow - up at a curvilinear infinitely near point @xmath38 whose exceptional curve @xmath227 is equal to the strict transform of @xmath221 in @xmath228 , where @xmath229 is either the real algebraic torus @xmath18 , or the rational real algebraic klein bottle @xmath230 .
let @xmath55 be the blow - up of @xmath18 at @xmath28 .
let @xmath231 be the blow - up of @xmath222 at @xmath28 .
it is clear that @xmath232 is a nonsingular projective complexification of @xmath55 .
let @xmath233 be the length of the curvilinear infinitely near point @xmath28 , where @xmath234 .
let @xmath235 be the minimal resolution of @xmath232 . if @xmath28 is a point of length @xmath32 , then @xmath236 , of course .
the morphism @xmath237 is a repeated blow - up of @xmath222 .
more precisely , there is a sequence of morphisms of algebraic varieties over @xmath12 @xmath238^{f_{m}}&{\mathcal{x}}_{m}\ar[r]^{f_{m-1 } } & \cdots\ar[r]^{f_0}&{\mathcal{x}}_0={\mathcal{x}}\ , , } \ ] ] with the following properties .
each morphism @xmath177 is an ordinary blow - up of @xmath239 at a nonsingular ordinary real point @xmath85 of @xmath239 , for all @xmath82 .
one has @xmath240 , i.e. , @xmath241 is the blow - up of @xmath222 at @xmath31 .
moreover , denoting by @xmath242 the exceptional curve of @xmath177 in @xmath243 , the center of blow - up @xmath244 belongs to @xmath242 but does not belong to the strict transform of any of the curves @xmath245 in @xmath243 for all @xmath246 .
denote again by @xmath242 and @xmath223 the strict transforms of @xmath242 and @xmath223 in @xmath247 , respectively .
the curves @xmath248 have self - intersection @xmath249 , the curve @xmath250 has self - intersection @xmath251 .
since @xmath223 is not tangent to @xmath28 , the curve @xmath223 in @xmath247 has odd self - intersection @xmath252 .
the curves @xmath253 form a chain of curves over @xmath12 in @xmath232 , intersecting in real points only ( see figure [ fichain ] ) .
the morphism @xmath235 is the contraction of the curves @xmath248 .
the morphism @xmath231 is the contraction of @xmath254 , i.e. , @xmath255 .
let @xmath256 be the self - intersection of @xmath223 in @xmath247 .
since @xmath257 and @xmath258 , the integer @xmath259 is even and non - negative .
let @xmath260 be pairwise complex conjugate of @xmath223 .
denote by @xmath261 the blow - up of @xmath262 in @xmath260 .
the algebraic variety @xmath261 is again defined over @xmath12 .
the strict transform of @xmath223 in @xmath261 is a nonsingular rational curve of self - intersection @xmath251 .
denote again by @xmath242 the strict transform of @xmath242 in @xmath261 .
the self - intersection of @xmath242 is equal to @xmath249 , if @xmath263 , the self - intersection of @xmath250 is equal to @xmath251 .
let @xmath264 be the algebraic surface defined over @xmath12 obtained from @xmath261 by contracting the union of the curves @xmath265 to a point , and let @xmath266 be the contracting morphism .
let @xmath267 be the algebraic surface defined over @xmath12 obtained from @xmath264 by contracting @xmath268 to a point , and let @xmath269 be the contracting morphism .
since @xmath270 is a repeated blow - down of @xmath251-curves , the algebraic surface @xmath267 is nonsingular .
moreover , the morphism @xmath271 is a blow - up of @xmath267 at a nonsingular curvilinear infinitely near point @xmath38 of @xmath267 .
denote again by @xmath223 the curve @xmath268 in @xmath264 .
the curve @xmath223 in @xmath264 is the exceptional curve of @xmath271 .
now take the associated real algebraic varieties , denoted by the corresponding roman characters .
since the points @xmath260 are non real , one has @xmath272 , i.e. , @xmath273 , the minimal resolution of @xmath55 .
it follows that @xmath274 , and that the induced algebraic map @xmath275 is the blow - up of the curvilinear infinitely near point @xmath38 of the nonsingular compact connected real algebraic surface @xmath276 .
the exceptional curve of @xmath277 is equal to the strict transform of @xmath221 in @xmath55 .
the only thing that is left to prove is the fact that the real algebraic surface @xmath276 is isomorphic to @xmath18 or to the rational real algebraic klein bottle @xmath230 . in order to establish this ,
observe that @xmath278 , as an @xmath279-fold blow - up of @xmath280 , is homeomorphic to the connected sum of @xmath18 and @xmath233 copies of @xmath281 .
since @xmath282 also is homeomorphic to the connected sum of @xmath276 and @xmath233 copies of @xmath281 , it follows that @xmath276 is homeomorphic to a torus or a klein bottle . by (
* theorem 1.3 ) , or ( * ? ? ?
* theorem 1.2 ) , or ( * ? ? ?
* theorem 1.5 ) , @xmath276 is isomorphic to @xmath18 or the rational real algebraic klein bottle @xmath230 . a similar , but easier , argument applies and proves the following lemma .
[ leblow - upsphere ] let @xmath28 be a curvilinear infinitely near point of @xmath16 , and let @xmath221 be a real algebraic curve in @xmath16 such that there is a nonsingular projective complexification @xmath222 of @xmath16 having the following properties : 1 .
the zariski closure @xmath223 of @xmath221 in @xmath222 is nonsingular and rational , 2 .
the self - intersection of @xmath223 in @xmath222 is even and non - negative , 3 .
@xmath224 , and 4 .
@xmath223 is not tangent to @xmath28 , i.e. , the scheme - theoretic intersection @xmath225 is of length @xmath32 .
then , there is an algebraic map @xmath283 that is the blow - up of @xmath16 at a curvilinear infinitely near point @xmath38 .
moreover , the exceptional curve @xmath284 is equal to the strict transform of @xmath221 in @xmath285 .
there is a nonsingular real rational compact surface @xmath55 such that @xmath0 is isomorphic to the real algebraic surface obtained from @xmath55 by repeatedly blowing up a nonsingular curvilinear infinitely near point . since @xmath55 is a nonsingular rational compact real algebraic surface , @xmath55
is obtained either from @xmath16 or from @xmath18 , by repeatedly blowing up an ordinary point ( cf .
* theorem 3.1 ) or ( * ? ? ?
* theorem 4.1 ) ) .
hence , there is a sequence of algebraic maps @xmath286^{f_n}&x_{n-1}\ar[r]^{f_{n-1}}&\cdots\ar[r]^{f_1}&x_0=z\ , , } \ ] ] where @xmath287 or @xmath288 , and each map @xmath177 is a blow - up at a nonsingular curvilinear infinitely near point @xmath101 of @xmath289 , possibly of length @xmath32 , for @xmath290 .
denote by @xmath291 the exceptional divisor @xmath292 of @xmath177 in @xmath293 .
let @xmath294 be the set of the curvilinear infinitely near points @xmath101 .
define a partial ordering on @xmath294 by @xmath295 if the composition @xmath296 maps @xmath297 to @xmath298 .
it is clear that @xmath294 is a forest , i.e. , a disjoint union of trees .
let @xmath299 be the number of edges in the forest @xmath294 .
we show the statement of the theorem by induction on @xmath299 .
the statement is clear for @xmath300 .
suppose , therefore , that @xmath301 .
we may assume that @xmath302 is the root of a tree of @xmath294 of nonzero height .
let @xmath221 be a real algebraic curve in @xmath229 satisfying the conditions of lemma [ leblow - uptorus ] if @xmath288 , and of lemma [ leblow - upsphere ] if @xmath287 , with @xmath303 .
such curves abound : one can take a bi - degree @xmath304 in @xmath18 , or a euclidean circle in @xmath16 , respectively .
moreover , we may assume that the strict transform of @xmath221 in @xmath293 does not contain @xmath305 , for all @xmath306 .
applying lemma [ leblow - uptorus ] and lemma [ leblow - upsphere ] , respectively , one obtains a sequence @xmath286^{f_n}&x_{n-1}\ar[r]^{f_{n-1}}&\cdots\ar[r]^{f_2 } & x_1\ar[r]^{f_1'}&x_0'=z'\ , , } \ ] ] where @xmath307 contracts the strict transform of @xmath221 in @xmath308 to a point @xmath309 .
the real algebraic surface @xmath310 is either the real algebraic sphere @xmath16 , or the real algebraic torus @xmath18 , or the rational real algebraic klein bottle @xmath230 . by construction , the number of edges in the forest @xmath311 associated to the latter sequence of blow - ups is equal to @xmath312 .
therefore , if @xmath313 or @xmath314 , we are done . if @xmath229 is the real algebraic klein bottle @xmath230 , then , according to lemma [ lekleinbottle ] , there is a sequence of blow - ups @xmath315^{f_0}&x_{-1}'\ar[r]^{f_{-1}}&x_{-2}=s^2\ , } \ ] ] at ordinary points such that the images of the centers @xmath316 in @xmath317 and @xmath318 are distinct from the centers of the blow - ups @xmath241 and @xmath319 .
we conclude also in this case by the induction hypothesis . a close inspection of the above proof reveals that the following slightly more technical statement holds .
[ thnonsuccessiveprime ] let @xmath0 be a rational dantesque surface , and let @xmath320 be a finite subset of nonsingular points of @xmath0 .
then there is an algebraic map @xmath217 or @xmath321 with the following properties : 1 .
there are mutually distant curvilinear subschemes @xmath40 on @xmath16 or @xmath18 , respectively , such that @xmath21 is the blow - up at @xmath40 , and 2 .
@xmath322 , for all @xmath82 .
the object of this section is to prove theorem [ thtranssing ] .
let @xmath6 and @xmath323 be two elements of @xmath4 .
we prove that there is an algebraic automorphism @xmath66 of @xmath0 such that @xmath324 , as curvilinear infinitely near points .
let @xmath216 be the set of ordinary points @xmath325 of @xmath0 .
since @xmath216 is a finite set of nonsingular points of @xmath0 , there is , by theorem [ thnonsuccessiveprime ] , an algebraic map @xmath217 or @xmath321 with the following properties : * there are mutually distant curvilinear subschemes @xmath202 on @xmath16 such that @xmath21 is the blow - up at @xmath202 , and * @xmath326 , for all @xmath82 .
since @xmath21 is an isomorphism at a neighborhood of @xmath216 , the image @xmath327 is a curvilinear infinitely near point of @xmath16 of length @xmath328 , and the same holds for @xmath329 , for all @xmath82 . by theorems [ thtorus ] and [ thsphere ]
, there is an algebraic automorphism @xmath211 of @xmath16 or @xmath18 , respectively , such that @xmath330 for @xmath57 , and @xmath212 for @xmath331 .
then , @xmath211 induces an algebraic automorphism @xmath66 of @xmath0 with the required property .
let @xmath0 and @xmath55 be rational dantesque surfaces , such that each of the surfaces @xmath0 and @xmath55 contains exactly one singularity of type @xmath56 , for each @xmath57 .
conversely , suppose that @xmath0 and @xmath55 are homeomorphic
. by theorem [ thnonsuccessive ] , there are nonsingular real rational surfaces @xmath276 and @xmath282 , and @xmath5-tuples @xmath332 and @xmath333 such that @xmath0 is the blow - up of @xmath276 at @xmath40 and @xmath55 is the blow - up of @xmath282 at @xmath334 . since @xmath0 and @xmath55 are homeomorphic , @xmath276 and @xmath282 are homeomorphic .
it follows that @xmath276 and @xmath282 are isomorphic . by theorem [ thtrans ]
, there is an isomorphism @xmath335 such that @xmath324 for @xmath57 .
the isomorphism @xmath66 induces an isomorphism between @xmath0 and @xmath55 .
the following example shows that the statement of theorem [ thiso ] does not hold for the slightly more general class of rational compact connected real algebraic surfaces that contain singularities of type @xmath58 .
[ exiso ] let @xmath0 be the real algebraic surface obtained from the real algebraic torus @xmath18 by contracting a fiber @xmath336 to a point . then @xmath0 is a rational compact connected real algebraic surface containing only one singular point .
its singularity is of type @xmath337 .
let @xmath28 be a point of @xmath281 .
the real algebraic surface @xmath230 obtained from @xmath281 by blowing up @xmath28 is a real algebraic klein bottle .
let @xmath55 be the real algebraic surface obtained from the real algebraic klein bottle @xmath230 by contracting to a point the strict transform of a real projective line in @xmath281 that passes through @xmath28 .
then @xmath55 is a rational compact connected real algebraic surface containing only one singular point .
its singularity is of type @xmath337 . | let @xmath0 be a singular real rational surface obtained from a smooth real rational surface by performing weighted blow - ups .
denote by @xmath1 the group of algebraic automorphisms of @xmath0 into itself .
let @xmath2 be a natural integer and let @xmath3 $ ] be a partition of @xmath2 . denote by @xmath4 the set of @xmath5-tuples @xmath6 of distinct nonsingular curvilinear infinitely near points of @xmath0 of orders @xmath7 .
we show that the group @xmath1 acts transitively on @xmath4 .
this statement generalizes earlier work where the case of the trivial partition @xmath8 $ ] was treated under the supplementary condition that @xmath0 is nonsingular . as an application we classify singular real rational surfaces obtained from nonsingular surfaces by performing weighted blow - ups . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ msc 2000 : _ 14p25 , 14e07 _ keywords : _ real algebraic surface , rational surface , geometrically rational surface , weighted blow - up singularity , algebraic automorphism , transitive action _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
in four years of data taking , icecube has observed high energy starting events ( hese ) with deposited energies between 60 tev and 2 pev @xcite .
the scientific debate about the origin of these events is extremely lively .
there is little doubt that cosmic neutrinos have been seen , but their origin is not yet understood . in this work ,
we focus our attention on two specific classes of events , not yet observed , that can give us precious information on the extra terrestrial component of neutrinos flux : the so - called double pulse events , due to tau - neutrinos @xcite , and the cascades above 2 pev that include events due to electron antineutrinos interacting at the glashow resonance .
the @xmath0 are not expected to be produced in astrophysical sources ( nor in the atmosphere ) but they are predicted to be a non negligible component of the cosmic neutrino flux due to flavor oscillations @xcite and thus represent a distinctive signature of a cosmic population . at low energy , it is impossible to distinguish cascades produced by charged current ( cc ) interactions of @xmath0 from those produced by cc - interactions of @xmath7 and neutral current ( nc ) interactions of all neutrino flavors .
the only way to tag @xmath0 is to observe a double pulse in the detector @xcite , which is produced by the cc - interaction of @xmath0 , when @xmath8 is produced , followed by a second energy release , when the @xmath8 decays . a very recent analyses from icecube @xcite , dedicated to the search of these events with different topology with respect to tracks and cascades , reported a null result .
we discuss the implications of this result and the perspective for future @xmath0 detection . the second class of events considered in this paper are cascades with deposited energy above 2 pev ; these events can be produced by deep inelastic scattering ( dis ) of high energy @xmath7 and @xmath0 and by @xmath1 interacting with electrons through glashow resonance @xcite . as already discussed in @xcite , the rate of these events
depends on the neutrino production mechanisms .
in particular , since glashow resonance is only possible for @xmath1 , a larger signal is expected if neutrinos are produced by @xmath3 collisions with respect to the case of @xmath9 interactions ( see @xcite for a review on the spectra of secondary particles produced in @xmath3 and @xmath4 interaction ) , being indeed the antineutrino fraction larger in the first case .
the possibility to discriminate among the two mechanisms depends on the relative contributions of events produced by dis and glashow resonance .
we perform a realistic calculation of these contributions . differently from previous work on the subject @xcite
, we discuss the role of leptonic channels in glashow resonance that can be correctly evaluated only if the difference between the incoming neutrino energy and the energy deposited in the detector is taken into account .
the expected rates of both classes of events depend on the assumed neutrino energy distribution .
our nominal hypothesis is that the cosmic neutrino spectra are described by single power law that extends until 10 pev .
we consider the neutrino spectral index as a free parameter and we fix the normalization of fluxes by requiring that they produce the events observed by icecube at low energies ( i.e. below 2 pev ) .
we thus obtain the expected rates of double pulse and cascades above 2 pev as a function of the slope of the neutrino spectrum .
this permits us to discuss the relevance of the assumed neutrino energy distribution for future @xmath0 detection , for the discrimination of @xmath3 or @xmath9 production mechanism and/or for the observation of high energy cutoff , automatically implementing the present information provided by icecube at low energy .
the plan of the paper is the following : in sec.[sec1 ] we describe our assumptions on the cosmic neutrino flux , in sec.[sec2 ] we calculate the expected number of double pulse events in icecube and in sec.[sec3 ] the expected number of cascades with deposited energy avove 2 pev . in sec.[sec4 ]
we made a comparison between our results and previous works on these subjects and finally , in sec.[sec5 ] , we draw our conclusions .
we assume that the total flux of cosmic neutrinos ( and antineutrinos ) has an isotropic distribution and that the spectrum can be described by a power law @xmath10 that extends till @xmath11 . recalling that neutrinos take about 1/20 of the energy of the parent proton in cosmic ray interactions
, this means that we are considering protons with energies up to 200 pev in their sources .
it is generally expected that , due to flavor oscillations , a cosmic neutrino population is characterized by a flavor content @xmath12 independently on the specific production mechanism . in reality , a certain imprint of the neutrino production mechanism does remain , as it is discussed e.g. in @xcite .
the fluxes divided per flavors can be generally given as : @xmath13 where @xmath14 and @xmath15 are ( small ) parameters described in @xcite that are determined by the neutrino flavor content at the source ( i.e. before oscillations ) .
in the following , we consider the case of neutrino produced by charged pion decays for which @xmath16 , @xmath17 ; the errors are obtained by propagating uncertainties in neutrino oscillation parameters .
the normalization of the flux @xmath18 is obtained by requiring that the number of events , due to cosmic neutrinos , reproduces the results obtained by icecube at low energies ( i.e. between 60 tev and 2 pev ) . in three years of data taking
, icecube has observed @xmath19 events against an expected background of @xmath20 events from atmospheric muons and neutrinos .
we require that the number of events from astrophysical neutrinos , calculated as : @xmath21 where @xmath22 is the observation time and @xmath23 are the _ effective areas _ for the various neutrino flavors given in @xcite , is equal to @xmath24 .
we introduced an upper integration limit to mimic the effect of the icecube observation threshold at 2 pev . by following the above procedure
, we obtain the flux normalization : @xmath25 \label{fa}\ ] ] note that the coefficient @xmath18 determines the flux of cosmic neutrinos at 1 pev . in the power law assumption , see eq.([pl ] ) , this quantity is relatively well constrained being equal to @xmath26 for @xmath27 and @xmath28 for @xmath29 .
as stated in the introduction , one of the goals of this work is to discuss the detection of the @xmath0 component of the high energy ( he ) neutrino flux providing the proof of existence of a cosmic population .
we are interested to investigate the dependence of the expected number of double pulse events from the energy distribution of cosmic neutrinos and from the icecube observation time . in order to perform this calculation
we need the effective area for double pulse events , @xmath30 , recently published by the icecube collaboration @xcite .
following the icecube prescription , the expected number of double pulse events in the observation time @xmath22 is @xmath31 where the differential flux of the @xmath0 component , @xmath32 is normalized to reproduce the hese events observed by icecube ( see previous section ) . in order to discuss the dependence of @xmath33 from the spectral index @xmath34 , it is useful to give an analytical description of the icecube effective area . considering that double pulse events are a _ subset _ of the events caused by cc tau neutrino interactions , we describe the effective area as , @xmath35 where @xmath36 is the effective area for @xmath0 calculated in @xcite , the factor @xmath37 gives the fraction of @xmath0 interactions that are due to cc processes and the constant @xmath38 describes the effect of geometrical and quality cuts implemented by icecube for the search of these events .
the function @xmath39 describes the probability that a neutrino with energy @xmath40 produces a tau traveling more than @xmath41 before it decays , where @xmath41 is the minimum distance to give rise to an observable double pulse in the detector .
we expect that @xmath41 is of the order of tens of meters with @xmath42 km ; we checked that the changes are not conspicuous in the range of energies of interest .
] , that is the typical distance between the doms @xcite .
the taus produced in cc - dis have an average energy equal to : @xmath43 where @xmath44 is the mean inelasticity which is nearly constant in the energy range that we are considering @xcite . if we neglect @xmath8 energy dispersion and assume the one - to - one relationship between @xmath45 and @xmath46 expressed by eq.([etau ] ) , the probability @xmath39 is given by : @xmath47\ ] ] where @xmath48 represents the minimum neutrino energy which is necessary to produce a tau with decay length larger than @xmath41 .
this can be calculated as : @xmath49 with @xmath50 and @xmath51 . using the expression in eq.([2p ] ) , we find that the effective area of icecube , in the energy region from 0.1 to 10 pev , is reasonably well described setting @xmath52 that corresponds to @xmath53 m. in other words the following parameterized expression for the effective area can be used : @xmath54 \beta=0.455\\[1ex ] e_{\mbox{\tiny min}}=0.5\mbox { pev}\end{array } \right\},\ ] ] as showed in fig.[2bangeff ] for a direct comparison of the icecube effective area .
using the previous expression it is possible to obtain an accurate analytical formula for the expected number of double pulse events @xmath55 where @xmath56 is the incomplete gamma function and the normalization of @xmath57 flux is assumed to be @xmath58 , as it is expected for neutrinos produced by charged pions decays with few @xmath2 uncertainty due to errors in the neutrino oscillation parameters ( see previous section for details ) . to check our result , we compare with icecube calculations in @xcite finding agreement at the level of few percents .
the above expression allows us to investigate the dependences of the expected number of double pulse events on the spectral index @xmath34 and on the high energy cutoff @xmath59 of the neutrino spectrum .
in particular , it permits us to show that our knowledge of the neutrino spectrum is already sufficient to make significative predictions .
the number of double pulse events expected in 4 years of data taking is 0.66 , 0.53 , 0.41 , 0.31 for @xmath60 and @xmath61 , so it is not surprising that icecube have not seen double pulse events so far .
our calculations are done by adopting the nominal cutoff energy @xmath62 pev .
however , the predicted values are not strongly dependent on the assumed high energy cutoff . for @xmath63 ,
the counting rate varies indeed by only @xmath64 when the cutoff energy is varied within the decade @xmath65 pev . for larger values of @xmath34 ,
the dependence of @xmath66 on @xmath59 is considerably weaker .
the dependence of @xmath66 on the spectral index mainly arises from the normalization @xmath18 of the cosmic neutrino flux at @xmath67 , see eq.([pl ] ) .
the residual dependence on @xmath34 is relatively weak and affects the final results at the few @xmath2 level when @xmath68 .
a good approximation for the predicted number of double pulses is thus given by : @xmath69 we remind that the normalization @xmath18 is constrained within a factor of @xmath70 for @xmath71 . this can be better appreciated from fig.[parental ] , where we show with a yellow line the integrand @xmath72 of calculated for @xmath73.the function @xmath74 is peaked around 0.5 pev and approximately one half of the double pulse signal is due to neutrinos with initial energy below 2 pev , i.e. to the energy region already probed by hese observations in icecube .
finally , we show in figure [ 2bang ] the probability to observe at least one double pulse as a function of spectral index and number of years . to observe a double pulse in icecube with a probability greater than 90 @xmath2 we must wait about 10 years in the most favorable case ( @xmath27 ) , about 15 years in the case @xmath73 and much more if the spectral index is close to @xmath29 .
in this section we estimate the number of cascade events with deposited energy above @xmath75 in icecube .
two types of cascade events are described : those from deep inelastic scattering ( dis ) and those produced by the glashow resonance @xcite .
cascades from dis are mostly given by cc - interactions of @xmath7 and @xmath0 with a negligible contribution from nc interactions of neutrinos of all flavors , as discussed in the following .
the expected number of cascades from dis , , is given by : @xmath76 \label{dis}\ ] ] where : + - @xmath77 and @xmath78 are the effective area for dis of @xmath7 and @xmath0 , which are calculated in sect . [ disentangling ] ; + - the factor @xmath79 is given in the previous section ; + - the function @xmath80 represents the probability that a cc - dis event produced by neutrino @xmath81 of energy @xmath82 has a visible energy above @xmath83 . .__branching ratio of @xmath84 decay _ _ [ cols="^,^",options="header " , ] [ default ] the ratios of the last two columns decreases with @xmath34 , because glashow resonance and deep inelastic scattering are more affected by a change of the slope with respect to the number of double pulse events , as we can see in the eqs.([prox1],[prox2],[prox3 ] ) . in order to clarify what is the relevant energy region of the parental neutrino flux giving a contribution to expected events discussed in this work , we show in fig.[parental ] the integrands @xmath85 of eqs.([tau1]-[dis]-[gr ] ) multiplied by an extra factor @xmath46 for the three channels analyzed in this paper for a spectral index @xmath73 .
it is interesting to remark that also @xmath0 of about 0.5 pev can give rise to a `` double pulse '' event , because there is a compensation between the flux power law decrease and the probability to not decay that increases exponentially with the energy .
for the three types of processes analyzed in the paper , under the hypothesis @xmath73 in the neutrino spectra .
_ _ ] the difference between @xmath3 and @xmath9 interaction at the source is already discussed in other works .
we compare here with some of them , that are important and well - known , namely @xcite , @xcite , @xcite and @xcite .
we point out the reasons why our analysis is different and show that , although results could seem to be incompatible , there are no contradictions .
our analysis of the events with deposited energy above 2 pev shows that the rate in the case of @xmath9 interaction at the source is about 75 @xmath2 of the rate given by @xmath3 interaction .
in fact , the difference is due to events produced by glashow resonance , but in this region of energy , the rate of dis events is large and this reduces the difference between the two types of mechanism .
this is not in contradiction with what is written in the paper @xcite , in which the ratio between resonant and non resonant event above 2 pev , is about one half . in that analysis , in fact , _ the energy of incoming neutrino and not the deposited energy into the detector _
is considered . when the deposited energy is considered instead , the neutral current interaction of all type of neutrinos and the charged current interaction of @xmath86 give a negligible contribution to cascade above 2 pev : this explain the difference between the results of the two papers .
our analysis uses the deposited energy , since this is observable , and it is not simple ( or possible ) to reconstruct the energy of incoming neutrinos .
that explain also the difference with respect to @xcite . the comparison with @xcite concerns another important remark . even in this work , the difference between @xmath3 and @xmath9 interaction at the source is discussed and their conclusion is that the number of glashow events , in the case of @xmath3 mechanism , is about 6 times greater than what expected in the case of @xmath9 mechanism . again , this statement is not in contradiction with our statement , because _ they compare models with the same flux of protons at the source _ whereas we take into account the observed flux of neutrinos at the earth . in other words ,
we ask ourselves what is the fraction of electron antineutrinos with respect to the _ observed _ total flux of neutrinos and , as a consequence , only the difference due to neutrinos oscillations is relevant for these considerations .
the oscillations produce a difference of about a factor of 2 between the two mechanisms of production , and the additional difference found by @xcite is due to the different fraction of energy that neutrinos receive in the @xmath3 and @xmath9 production mechanism .
the last remark concerns a hypothesis proposed in @xcite and @xcite , where it was suggested that the events observed above pev could be due to glashow resonance .
if this was true , were caused by leptonic decay , since the energy resolution is incompatible with a visible energy being 6.32 pev .
now , the ratio between hadronic and leptonic branching ratios is , @xmath87 and the same neutrino flux leads to leptonic and hadronic decays . to date
there are three events with energy below 2 pev , so in the above assumption , we expect to have also 6 hadronic events .
the probability to observe none is given by the poissonian pdf , @xmath88 , which is disfavored at 3 @xmath89 .
in this paper we have considered two categories of very high energy events in icecube that can help to diagnose cosmic neutrinos : the double pulse events , that may allow us to clearly discriminate the cosmic component of @xmath0 ; the cascades with deposited energy above 2 pev , including events produced by @xmath1 at the glashow resonance , that can be used to investigate the cosmic neutrino production mechanisms .
as stated in the introduction , we estimated the rate of these high energy events with the important constraint provided by the data already observed by icecube , i.e. we used the data collected in the low - energy region below @xmath70pev to normalize our calculations . in this way
, we obtained the expected rates of high energy events as a function of the neutrino spectral index @xmath34 , that we varied in the range @xmath90 .
we found that the non - observation of double pulse events does not contradict the hyphotesis of a cosmic neutrino population .
this conclusion is only marginally dependent on the assumed cosmic neutrino spectrum .
in fact , we have shown that : + _ i ) _ one half of the expected signal is due to neutrinos with energy below @xmath91 pev , i.e. _ from a spectral region that is already observed in the hese data _ ( see fig . [ parental ] and discussion in the section [ sec2 ] ) + _ ii ) _ in the most favorable case , with spectral index @xmath27 _ we need to wait about 10 more years to observe a double pulse with a probability greater than 90 @xmath2_. concerning the cascades with deposited energy above 2 pev , we have shown that : + _ i ) _ due to the difference between the energy deposited in the detector and the energy of the interacting neutrino , the contribution of leptonic channels to glashow resonance events is suppressed by 75@xmath2 .
this implies that hadronic modes account for 90@xmath2 of the total signal produced by glashow resonance above 2 pev ; + _ ii ) _ this class of events can be used to probe the high energy tail of the cosmic neutrino spectrum .
the absence of cascades above 2 pev disfavours a neutrino spectral index @xmath27 ( with a cutoff @xmath92pev ) at about 2@xmath89 considering that 3 events are expected in the worst scenario ( @xmath9 interaction ) .
an unbroken power law with @xmath29 is instead still compatible at 1@xmath89 with the present results .
the absence of events close to the glashow resonance energy @xmath93pev is not problematic under this hypothesis , since only 0.4 events are expected due to glashow resonance after 4 years ( see the table [ sommario ] ) ; + _ iii ) _ the difference between the event rates produced by @xmath3 or @xmath4 neutrino production mechanisms is not large enough to distinguish among the two options , even if we assume that the parent neutrino spectrum is known .
an observation time @xmath94 years would be required to obtain a @xmath5 discrimination , if the neutrino spectral index is @xmath73 .
99 m. g. aartsen _ et al . _
[ icecube collaboration ] , `` evidence for high - energy extraterrestrial neutrinos at the icecube detector , '' science * 342 * ( 2013 ) 1242856 [ arxiv:1311.5238 [ astro-ph.he ] ] .
m. g. aartsen _ et al . _
[ icecube collaboration ] , `` observation of high - energy astrophysical neutrinos in three years of icecube data , '' phys .
lett . * 113 * ( 2014 ) 101101 [ arxiv:1405.5303 [ astro-ph.he ] ]
. m. g. aartsen _ et al . _
[ icecube collaboration ] , `` a combined maximum - likelihood analysis of the high - energy astrophysical neutrino flux measured with icecube , '' astrophys .
j. * 809 * ( 2015 ) 1 , 98 [ arxiv:1507.03991 [ astro-ph.he ] ] .
f. vissani , g. pagliaroli and f. l. villante , `` the fraction of muon tracks in cosmic neutrinos , '' jcap * 1309 * ( 2013 ) 017 [ arxiv:1306.0211 [ astro-ph.he ] ] .
a. palladino , g. pagliaroli , f. l. villante and f. vissani , `` what is the flavor of the cosmic neutrinos seen by icecube ?
, '' phys .
* 114 * ( 2015 ) 17 , 171101 [ arxiv:1502.02923 [ astro-ph.he ] ] . m. g. aartsen _ et al . _
[ icecube collaboration ] , `` flavor ratio of astrophysical neutrinos above 35 tev in icecube , '' phys .
* 114 * ( 2015 ) 17 , 171102 [ arxiv:1502.03376 [ astro-ph.he ] ] .
a. c. vincent , s. p. ruiz and o. mena , `` the flavour composition of the high - energy icecube neutrinos , '' arxiv:1505.03355 [ astro-ph.he ] .
a. palladino and f. vissani , `` the natural parameterization of cosmic neutrino oscillations , '' eur .
j. c * 75 * ( 2015 ) 433 [ arxiv:1504.05238 [ hep - ph ] ] .
m. bustamante , j. f. beacom and w. winter , `` theoretically palatable flavor combinations of astrophysical neutrinos , '' arxiv:1506.02645 [ astro-ph.he ] .
g. pagliaroli , a. palladino , f. vissani and f. l. villante , `` testing neutrino decay scenarios with icecube data , '' arxiv:1506.02624 [ hep - ph ] .
j. g. learned and s. pakvasa , `` detecting tau - neutrino oscillations at pev energies , '' astropart .
phys . * 3 * ( 1995 ) 267 [ hep - ph/9405296 , hep - ph/9408296 ] .
d. f. cowen [ icecube collaboration ] , `` tau neutrinos in icecube , '' j. phys .
ser . * 60 * ( 2007 ) 227 .
r. abbasi _ et al . _
[ icecube collaboration ] , `` a search for uhe tau neutrinos with icecube , '' phys .
d * 86 * ( 2012 ) 022005 [ arxiv:1202.4564 [ astro-ph.he ] ] .
s. l. glashow , `` resonant scattering of antineutrinos , '' phys .
* 118 * ( 1960 ) 316 .
v. s. berezinsky and a. z. gazizov , `` cosmic neutrinos and possibility to search for w bosons having 30-gev-100-gev masses in underwater experiments , '' jetp lett .
* 25 * ( 1977 ) 254 [ pisma zh .
eksp . teor .
* 25 * ( 1977 ) 276 ] .
l. a. anchordoqui , h. goldberg , f. halzen and t. j. weiler , `` neutrinos as a diagnostic of high energy astrophysical processes , '' phys .
b * 621 * ( 2005 ) 18 [ hep - ph/0410003 ] .
a. bhattacharya , r. gandhi , w. rodejohann and a. watanabe , `` the glashow resonance at icecube : signatures , event rates and @xmath3 vs. @xmath4 interactions , '' jcap * 1110 * ( 2011 ) 017 [ arxiv:1108.3163 [ astro-ph.he ] ] .
v. barger , l. fu , j. g. learned , d. marfatia , s. pakvasa and t. j. weiler , `` glashow resonance as a window into cosmic neutrino sources , '' phys .
d * 90 * ( 2014 ) 121301 [ arxiv:1407.3255 [ astro-ph.he ] ] .
s. r. kelner , f. a. aharonian and v. v. bugayov , `` energy spectra of gamma - rays , electrons and neutrinos produced at proton - proton interactions in the very high energy regime , '' phys .
d * 74 * ( 2006 ) 034018 [ erratum - ibid .
d * 79 * ( 2009 ) 039901 ] .
s. r. kelner and f. a. aharonian , `` energy spectra of gamma - rays , electrons and neutrinos produced at interactions of relativistic protons with low energy radiation , '' phys .
d * 78 * ( 2008 ) 034013 [ erratum - ibid .
d * 82 * ( 2010 ) 099901 ] .
f. l. villante and f. vissani , `` how precisely neutrino emission from supernova remnants can be constrained by gamma ray observations ?
. rev .
d * 78 * ( 2008 ) 103007 [ arxiv:0807.4151 [ astro - ph ] ] .
s. hummer , m. ruger , f. spanier and w. winter , `` simplified models for photohadronic interactions in cosmic accelerators , '' astrophys . j. * 721 * ( 2010 ) 630 [ arxiv:1002.1310 [ astro-ph.he ] ] .
s. hummer , m. maltoni , w. winter and c. yaguna , `` energy dependent neutrino flavor ratios from cosmic accelerators on the hillas plot , '' astropart .
* 34 * ( 2010 ) 205 [ arxiv:1007.0006 [ astro-ph.he ] ] .
r. gandhi , c. quigg , m. h. reno and i. sarcevic , `` neutrino interactions at ultrahigh - energies , '' phys .
d * 58 * ( 1998 ) 093009 [ hep - ph/9807264 ] .
p. lipari , `` hadronic interactions modeling , and the calculation of the inclusive fluxes of atmospheric muons and neutrinos , '' nucl .
* 175 - 176 * ( 2008 ) 96 .
g. battistoni _ et al .
_ , `` the hadronic models for cosmic ray physics : the fluka code solutions , '' nucl . phys .
* 175 * ( 2008 ) 88 [ hep - ph/0612075 ] .
a. bhattacharya , r. gandhi , w. rodejohann and a. watanabe , `` on the interpretation of icecube cascade events in terms of the glashow resonance , '' arxiv:1209.2422 [ hep - ph ] .
v. barger , j. learned and s. pakvasa , `` icecube pev cascade events initiated by electron - antineutrinos at glashow resonance , '' phys .
d * 87 * ( 2013 ) 3 , 037302 [ arxiv:1207.4571 [ astro-ph.he ] ] .
f. vissani and f. aharonian , `` galactic sources of high - energy neutrinos : highlights , '' nucl .
instrum .
a * 692 * , 5 ( 2012 ) .
z. z. xing and s. zhou , `` the glashow resonance as a discriminator of uhe cosmic neutrinos originating from p - gamma and p - p collisions , '' phys .
d * 84 * ( 2011 ) 033006 .
c. moura , `` detection of ultra - high energy tau - neutrino by double - bang event , '' aip conf .
* 809 * ( 2006 ) 190 .
a. m. dziewonski and d. l. anderson , `` preliminary reference earth model , '' phys .
earth planet .
interiors * 25 * ( 1981 ) 297 .
l. yacobi , d. guetta and e. behar , `` implication of the non - detection of neutrinos above 2pev , '' arxiv:1510.01244 [ astro-ph.he ] . | icecube collaboration has seen an unexpected population of high energy neutrinos compatible with an astrophysical origin .
we consider two categories of events that can help to diagnose cosmic neutrinos : double pulse , that may allow us to clearly discriminate the cosmic component of @xmath0 ; cascades with deposited energy above 2 pev , including events produced by @xmath1 at the glashow resonance , that can be used to investigate the neutrino production mechanisms .
we show that one half of the double pulse signal is due to the neutrinos spectral region already probed by icecube . by normalizing to hese data
, we find that 10 more years are required to obtain 90@xmath2 probability to observe a double pulse .
the cascades above 2 pev provide us a sensitive probe of the high energy tail of the neutrino spectrum and are potentially observable , but even in this case , the dependence on type of the source is mild .
in fact we find that @xmath3 or @xmath4 mechanisms give a difference in the number of cascades above 2 pev of about 25 @xmath2 that can be discriminated at @xmath5 in @xmath6 years of data taking . |
we thank prof .
m. bttiker for helpful communications and discussions .
we gratefully acknowledge support by a rgc grant from the government of hong kong under grant number hku 261/95p , the natural sciences and engineering research council of canada and le fonds pour la formation de chercheurs et laide la recherche de la province du qubec .
we thank the computer center of the university of hong kong for computational facilities .
m. bttiker , a. prtre and h. thomas , phys .
lett . * 70 * , 4114 ( 1993 ) ; m. bttiker and h. thomas , in _ quantum - effect physics , electronics and applications _ , eds .
k. ismail _ et .
( institute of physics conference series number 127 , bristol , 1992 ) , pp .
19 .
al . _ have investigated dynamic magnetoconductance of a quasi-1d dirty wire using a phenomenological constant to account for the self - consistent internal potential response , rather than using the full theory by computing the local density of states .
liu , ben yu - kuang hu , c.a .
stafford and s. das sarma , phys .
b * 50 * , 5799 ( 1994 ) .
near the second subband threshold , there seems to be a small but systematic deviation to the equal sign of eq .
( [ final ] ) .
this deviation could be due the fact that transmission probability has sharp changes in its value near the onset of the second subband and eq .
( [ final ] ) does not include multi - propagating modes . | the frequency dependent conductance of a two - dimensional quantum wire is computed using a current conserving formalism .
the correction to the dc - conductance due to a time - dependent potential is related to the local partial density of states which we compute numerically .
the current conservation is explicitly confirmed by computing the global density of states and comparing it with a quantity which is related to the electron dwell time .
our calculation clearly reveals the physical meaning of the various partial density of states .
= 10000 the physics associated with quantum conduction in various low dimensional systems has been a major focus in current solid state research . due to advances in controlled crystal growth and lithographic techniques , it is now possible to fabricate various two - dimensional submicron structures where accurate experimental measurements can be made@xcite . on the theoretical side ,
the understanding of quantum transport in these very small systems has been advanced by the landauer theory@xcite of one - dimensional transport . in cases where multi - probes are attached to the conductor
, bttiker has provided the useful formula for computing various conductances@xcite .
the landauer - bttiker formalism has been applied to a large class of quantum transport problems in the mesoscopic and ballistic regimes where the external potential is static . on the other hand ,
when external potential has a time - dependent oscillating component , the landauer - bttiker formalism can not be directly applied . as shown by bttiker and his co - workers@xcite that a direct application of the original approach of landauer - bttiker formalism , where the internal electric potential distribution is not needed , can not yield electric current and charge conservation .
hence the coherent quantum transport theory for mesoscopic and ballistic conductors must be extended when time - varying external potentials are to be included . clearly
the problem of computing frequency dependent conductance for mesoscopic and ballistic conductors is a very important subject from both theoretical and application points of view .
recently in a series of articles , bttiker and co - workers@xcite have developed a theoretical formalism for investigating frequency - dependent coherent quantum transport for mesoscopic and ballistic conductors when electric potential in the leads far away from the scattering region of the conductor is time varying .
this is an important advance in the theory of mesoscopic physics and warrant further development of the theory and application . in
simple one dimensional systems such as a @xmath0function potential , a perfect quantum wire , or an 1d quantum well , the frequency dependent admittance can be explicitly calculated since the scattering matrix can be obtained analytically@xcite .
however , in truly 2d systems the application of the formalism is less straightforward . this is due partly to technical difficulties of computing certain quantities such as the various partial density of states ( pdos , see below ) .
these quantities , which appear naturally in the theoretical formulation of the problem@xcite , have deep physical meaning which is not obviously clear and needs to be revealed .
to the best of our knowledge , there are so far no studies and direct calculations@xcite of the various partial density of states in 2d .
but these quantities must be obtained if the frequency dependent admittance is to be computed from first principles .
the purpose of this paper is to report our investigations on the dynamic transport properties of a 2d quantum wire system .
in particular , we consider a t - shaped quantum wire structure as shown in fig .
( 1 ) .
the two probes extend to @xmath1 while the scattering region is provided by the t - shaped junction as shown being bounded by the two dotted lines .
this 2d quantum wire has been studied by many authors@xcite for the case of static transport . while this is a simple 2d system
, we found that much physical intuition about various quantities can be obtained for the dynamic transport ( see below ) .
we have computed the frequency dependent admittance for this system using the current conserving formalism of ref .
@xcite .
the correction to the dc - conductance due to a time - dependent potential is related to the local partial density of states which we compute numerically .
the current conservation is explicitly confirmed by computing the global density of states and comparing it with a quantity which is related to the electron dwell time .
our direct calculation clearly reveals the physical meaning of the partial density of states which plays an essential role in the dynamic transport theory . for the sake of presentation we briefly review the current conserving dynamic transport formalism of bttiker , thomas , prtre , gasparian and christen@xcite .
their theory proceeds in three steps .
the first step is to determine the current and charge density response of the system to the time - variation of the external potential .
this leads to an admittance matrix with elements given by , to first order in @xmath2 , @xmath3 where the first term on the right hand side is the admittance ( conductance ) when frequency @xmath4 and this term is given by the usual bttiker multi - probe conductance formula@xcite .
the second term gives a correction due to the time - dependence of the external potential , and is determined by the global partial density of states@xcite ( global pdos ) , @xmath5 .
here the indices @xmath6 denote scattering from a probe labeled by @xmath7 to that labeled by @xmath8 .
this admittance matrix @xmath9 does not conserve current , since @xmath10 . to correct this problem
, one must consider the internal potential distribution induced by the external perturbation .
this is computed in the second step of the theory@xcite , and the result is rather simple if only thomas - fermi linear screening is included . in the final step ,
one computes the currents in the probes induced by this internal potential distribution , which leads to an additional term in the admittance matrix .
thus the total admittance is given by @xmath11 , with @xmath12 and @xmath13 \left[\frac{dn({\bf r})}{de}\right]^{-1 } \left[\frac{dn({\bf r},\beta)}{de}\right]\ \ \ . \label{dab}\ ] ] in eq .
( [ dab ] ) the quantity @xmath14 is the _ injectivity _ which measures the additional local charge density brought into the sample by the oscillating chemical potential at probe @xmath8 . without a magnetic field ,
the _ emissivity _ @xmath15 equals to the injectivity .
it was shown@xcite that these quantities could be computed from the electron dwell time or the scattering green function .
finally , the total local density of states is @xmath16 . with these expressions ,
we obtain the final formula for the frequency dependent admittance to linear order in @xmath2 , @xmath17 it is now straightforward to prove that current is conserved since the admittance matrix @xmath18 satisfies @xmath19 .
this can be seen by realizing that @xmath20 is the injectance which is identical to @xmath21 .
let s apply this formalism to the t - shaped 2d conductor .
the incident electron comes from probe 1 , it scatters at the t - junction , and then reflects back to probe 1 or transmits to probe 2 . to be concrete
we have fixed the wire width to be @xmath22 , the width and height of the side stub are fixed at @xmath23 and @xmath24 with @xmath25 .
the units are fixed by @xmath26 with @xmath27 the effective mass of the electron , and lengths measured in terms of @xmath22 . for simplicity
we have focused on the first _ transport _ subband only , thus the incoming electron energy is restricted : @xmath28 . in this case elements of the scattering matrix @xmath29 are simple complex numbers . with more than one subbands
they become matrices in the subband space .
however the computation proceeds in similar fashion with one or more than one subbands .
we have solved the quantum scattering problem using a mode matching method@xcite , where the wavefunction in the scattering region was expanded using a suitable basis set and in particular @xmath30 modes were included .
we have checked the convergence that including more modes in the scattering region essentially does not change the results .
the scattering probabilities were obtained by matching the wavefunctions and their derivatives at the boundaries between the scattering region and the probes . to compute the admittance we need to know various partial density of states .
these quantities turn out to have very interesting and physically clear property once they are plotted .
the global pdos is related to the scattering matrix , and is approximately given by@xcite @xmath31 for a finite scattering volume there are corrections to this expression of the order @xmath32 where @xmath33 is the electron wavelength and @xmath34 the system size@xcite . for large system sizes or large energies , eq .
( [ gpdos ] ) is adequate .
later when we check the current conservation , the correction term will be added . since in 2d
the expression for the scattering matrix can not be written down analytically , we have decided to perform the energy derivatives of ( [ gpdos ] ) numerically .
in particular we computed @xmath35 at various values of energy @xmath36 and used a @xmath37-point numerical derivative to find @xmath38 . the behavior of the global pdos is plotted in fig .
( 2 ) together with the transmission coefficient @xmath39 . as our quantum wire system is very transmissive , @xmath40 has large values in general except at the two resonance levels where @xmath41 . in a previous work we have shown that the energies where @xmath41 corresponds to the quasi - bound states located in the t - junction@xcite .
the behavior of the global partial density of states coincides well with that of the transmission coefficients .
in particular @xmath42 , which is the global pdos for reflection , peaks at the energies where transmission coefficient is minimum or reflection maximum . on the other hand ,
@xmath43 , which is the global pdos for transmission , takes minimum value when @xmath44 is minimum .
indeed , a general discussion for an one - dimensional system with a symmetric scatterer@xcite reveals that @xmath45 while @xmath46 with @xmath47 and @xmath44 the reflection and transmission coefficients , and @xmath48 the global dos .
hence these quantities have vivid physical meaning .
the next quantity of interest is @xmath49 given by eq .
( [ dab ] ) . to compute this quantity
, we first find the injectivity which is given by the scattering wavefunction@xcite in the t - junction , at zero temperature this is given by @xmath50 where @xmath51 is the incoming particle flux .
clearly a spatial integration of this quantity gives the electron dwell time@xcite @xmath52 .
in general @xmath49 is obtained with two calculations of the wavefunction for particles coming from left and from right .
fig .
( 3 ) shows this quantity as a function of the electron energy .
both @xmath53 and @xmath54 peak at the energies where the transmission takes the minimum values .
this is understandable since these special energies correspond to the scattering states@xcite where the electron dwell time takes maximum value , and the injectivity represents essentially the electron dwelling in an area @xmath55 at position @xmath56 irrespective where it is finally scattered . with the global partial density states and the quantity @xmath49 calculated , we have thus obtained the admittance @xmath57 from eqs .
( [ ge ] ) , ( [ gi ] ) , and ( [ gfinal ] ) . fig .
( 4 ) shows the @xmath2-dependent part of @xmath18 , @xmath58 as a function of energy @xmath36 .
apart from the prefactor @xmath59 , this quantity @xmath60 is the imaginary part of the admittance , and is called _ emittance_. at the quasi - bound state levels where transmission coefficient @xmath39 takes minimum values , @xmath61 also takes extremal values .
it is interesting to observe that the transmissive part @xmath62 takes minimum value where @xmath63 , while the reflective part @xmath64 takes maximum value at the same energy .
thus in this sense , the dynamic part of the admittance @xmath65 has the same behavior as the static admittance @xmath66 as a function of energy .
furthermore , @xmath65 changes sign as energy is varied : the system responds either capacitively when @xmath67 , or inductively otherwise . hence at the resonance ( where @xmath68 for our system ) @xmath62 and @xmath64 are capacitive .
fig .
( 4 ) shows the clear crossover between the capcitive and inductive responses for this 2d quantum wire as energy is changed .
we can now explicitly confirm the current conservation by summing up the admittance matrix elements and check whether or not @xmath69 .
since the global pdos obtained using eq .
( [ gpdos ] ) is not exact , a correction should be added .
for an 1d system gasparian _ et .
al._@xcite have shown that @xmath70 where @xmath36 is the electron energy .
unfortunately for a 2d system such as ours , the green s function can not be written down analytically hence how to derive a similar correction term as that in ( [ conservation ] ) is unclear .
however since we have numerical results of all the quantities , a reasonable correction term can easily be obtained .
we found that the same form as ( [ conservation ] ) led to almost perfect current conservation , provided we use the _ transport _ energy @xmath71 as the energy @xmath36 in ( [ conservation ] ) . our data in fig .
( 5 ) clearly and unambiguously shows that the following is established @xmath72 with this result , we have thus explicitly shown@xcite the current conservation @xmath73 . since the correction term is inversely proportional to @xmath71 , it plays a role only at small @xmath74 .
this can already be seen in fig .
( 4 ) where the two curves add up to zero except at low energies . in summary
we have , for the first time , implemented the current conserving dynamic conductance formalism for 2d metallic conductors . in the simple case of applying thomas - fermi linear screening for the interacting electrons ( hence the analysis is more suitable for metallic samples ) , the current response to the internal potential can be computed from the electron dwell time using the scattering wavefunctions .
the frequency dependent admittance of a t - shaped 2d quantum wire is calculated to linear order in frequency .
we have explicitly confirmed the electric current conservation and found that a correction to the total density of states is needed at low energies , in the same fashion as that of 1d case .
it is very interesting to clarify the physical meaning of the partial dos .
these quantities provide information on the density of carriers _ and _ the transmission of carriers from one contact to another . at quantum resonances
these quantities take extremal values .
there are many further applications of this important theoretical formalism , to situations involving multi - mode , magnetic fields , and non - linear screening .
another very challenging extension is to push the theory to higher order in frequency@xcite and formulate the theory in a way that permits a computational implementation .
we hope to be able to report these results in the future . |
some time ago it was suggested to study certain systems on lattices with trivial first homotopy group @xcite .
the d - dimensional boundaries of ( d+1)-dimensional hypercubes have this property , being homotopic to @xmath4 . here
we give a brief summary of recent results obtained by various subgroups of the present authors for such studies for compact @xmath2 pure gauge theory with mixed action @xcite , for the 2d ising model @xcite and for the 2d potts model @xcite .
the original motivation was to check the possible relevance of periodic boundary conditions for the dynamics of looplike excitations ( like monopole loops in compact u(1 ) gauge theory ) . in the thermodynamic limit at the critical point
the manifold becomes flat and should lead to a universal field theory independent of the details of the boundary conditions .
other 2d studies on similar lattices include random as well as tetrahedral lattices @xcite and a recent investigation of the 7-vertex model ( the strong coupling schwinger model ) for a pillow - like topology @xcite .
we studied the following lattices : @xmath5}$ ] : : periodic b.c .
, i.e. torus topology for the @xmath6 hypercubic lattice @xmath7}$ ] : : the surface of a @xmath8-dimensional @xmath9 hypercube , or its dual lattice @xmath10 ( where points of @xmath11 are identified with @xmath12dimensional objects of @xmath10 etc . ) @xmath13}$ ] : : like @xmath11 but with weight factors correcting for a spherelike , smooth distribution of the curvature over the lattice @xmath14}$ ] : : a pillow - like shape , like the surface of a @xmath15 hypercube ; may be visualized as two copies of a @xmath6 hypercube glued together at their boundaries in the finite size scaling ( fss ) analyses we use @xmath16 as the generic lattice size parameter , where the volume @xmath17 denotes the number of sites or plaquettes on 2d or 4d lattices , respectively .
the monte carlo updating details depend on the studied system and are discussed below . in all cases
bulk quantities were obtained at various values of the coupling and combined with the multihistogram method @xcite , giving @xmath18 this allows us to obtain continuous curves for the observables and to determine the complex positions of the partition function zeroes .
also the combined data for the `` multihistogram '' provides for a reliable identification of possible 2-state signals in the distribution .
we studied the peak positions ( pseudocritical couplings @xmath19 ) and peak values of the cumulants @xmath20 and the positions @xmath21 of the fisher zeroes closest to the real axis . for 1@xmath22 order transitions
one expects the fss behaviour @xmath23 with `` surface '' corrections of @xmath24 due to the inhomogeneities of the boundary ( curvature ) ; in @xmath25 this is the leading term @xcite .
the parameter @xmath26 in ( [ pscfss ] ) depends on the considered cumulant . for 2@xmath27 order transitions we expect @xmath28 with josephson s law @xmath29 ( for gaussian systems @xmath30 ; for a discontinuous transition @xmath31 ) . again in @xmath19 one expects `` surface '' corrections of @xmath24 .
this was studied in @xcite in order to identify the fss effects of the new topology in more detail .
we used lattices of various size and of type @xmath32 ( where there exists an analytic solution ) , @xmath33 , @xmath34 and @xmath35 .
the updating was with the swendsen - wang cluster algorithm with typically @xmath36 measured configurations ( with an autocorrelation time @xmath37 ) per lattice size .
the results were typical for the other systems as well . since @xmath38 one expects a leading fss behaviour @xmath39 .
the pseudocritical points for the spherelike topologies are compared to the torus values very close to their thermodynamic values and exhibited a _ very small _
contribution @xmath39 ; the scaling behaviour is dominated by the non - leading term @xmath40 which we interpret as coming from the boundary or curvature , respectively .
this holds also for the real part of the closest fisher zero .
both , the peak value of @xmath41 as well as @xmath42 , however , show scaling in perfect agreement with @xmath38 . from fig .
[ fig1 ] we find , that for the spherelike topology the zeroes approach the real axis in the complex coupling constant plane almost perpendicularly .
thus the real part ( related to the pseudocritical points derived from the cumulants ) is always very close to the thermodynamic value with its scaling behaviour dominated by the @xmath40 term .
in @xcite the @xmath43-state potts model for @xmath44 at the ( for @xmath45 extremely weak ) 1@xmath22 order transition is studied with multicanonical updating @xcite , typically @xmath36 updates for each lattice size .
the lattices were of type @xmath46 , @xmath47 and @xmath48 .
even for the weak signal of the @xmath45 model the histogram for the energy distribution exhibits a clear double peak structure ( cf .
[ fig2 ] ) .
thus the spherical lattices do no `` wash out '' weak 2-state signals , as might be suspected .
note that there is only one interface as compared to two for the periodic boundary conditions .
this difference may be of some technical advantage in the determination of the interface tension .
the study @xcite of compact @xmath2 pure gauge action in the wilson formulation with an additional adjoint term ( coupling @xmath49 ) has been continued@xcite .
for @xmath50 , studies for torus geometry find consistently stable 2-state signals , reconfirmed by the recent results of @xcite and @xcite ( who in particular emphasized the rle of monopole clusters for the phase transition ) . already in 1985 a careful study of this action at positive @xmath49 hinted at a tricritical point ( tcp ) @xcite near @xmath51 .
however , 2-state signals have been observed even below that value @xcite . on the other hand ,
studies for spherelike topology show no 2-state signals in the energy distribution at @xmath52 @xcite for the studied lattices @xmath53 ( although they do show 2 states at e.g. @xmath54 ) . in order to respond to a possible criticism of our original investigation of the model on @xmath11-lattices where the curvature was concentrated on the edges and corners of the hypercubic lattice we have continued work on an `` almost smooth sphere '' @xmath55 .
there we project sites , links and plaquettes of @xmath56 or @xmath57 ( dual ) onto the surface of 4d spheres and introduce weight factors similar to those used by @xcite in their study of random triangulated lattices , @xmath58\;.\ ] ] here , @xmath59 and @xmath60 denote the areas of the corresponding plaquette ( and its dual , respectively ) of the projected lattice .
more technical details of the simulation have been discussed @xcite and will be discussed elsewhere @xcite . we have worked with lattices @xmath55 for @xmath61 ranging between 4 and 12 .
the couplings were chosen in the immediate neighbourhood of the critical values of @xmath62 for @xmath63 .
since we never observed 2-state signals we did not implement multi - canonical updating . for each lattice size
we typically accumulated @xmath64 updates ( 3-hit metropolis , for @xmath50 with an additional overrelaxation step ) .
close to the phase transition we found quite large autocorrelation lengths ( 300 - 2000 ) .
we find no indication of 2-state signals in the multi - histogram distributions .
the values of the cumulants at their respective extrema are compatible with a continuous phase transition . like for the ising model , also here we observe in general smaller fs corrections than for the torus geometry ( periodic b.c . ) .
the leading behaviour seems to be dominated again by the boundary or curvature term , i.e. suppressed by @xmath40 , which at first sight can not be distinguished from a leading scaling behaviour for @xmath65 .
the peak values of @xmath41 show power law behaviour and in the table we give the range of values for @xmath66 ( and the resulting @xmath67 , if we use josephson s law ) from fits to data for all lattice sizes or without the largest and/or smallest lattices . these values appear to depend on @xmath49 .
one has to keep in mind , however , that the fss law used for this fit is asymptotic and that there is a regular contribution to the value of @xmath41 which may be non - negligible for the lattice sizes entering the fit .
if we study the closest fisher zeroes ( in the complex @xmath62-plane ) , we find a behaviour reminiscent to the discussed ising results .
[ fig3 ] exhibits fss consistent with @xmath69 for all three values of @xmath49 ( cf .
the results of the fit in table [ tabfit ] ) .
once again @xmath68 gives the cleanest and most consistent result .
we should mention , that this value of @xmath67 is compatible with neither the gaussian value 0.5 nor the `` discontinuity '' value 0.25 ; it is compatible with results obtained in the mid-80s with mcrg methods .
more detailed analyses with improved statistics will be presented elsewhere @xcite . | we present results for 2d and 4d systems on lattices with topology homotopic to the surface of a ( hyper ) sphere @xmath0 or @xmath1 .
finite size scaling is studied in situations with phase transitions of first and second order type .
the ising and potts models exhibit the expected behaviour ; for the 4d pure gauge @xmath2 theory we find consistent scaling indicative of a second order phase transition with critical exponent @xmath3 . |
electron - electron interactions are important in understanding the properties of many transition metal oxides.@xcite novel ordered phases are found in this class of compounds : e.g. , high-@xmath5 superconductors , @xmath6-electron heavy fermions , and quantum magnets . the inclusion of disorder into such materials , and into the models of such systems , adds another level of complexity . in this report
we focus on metal - to - insulator transitions that are controlled by the combined effects of both interactions and disorder . in part , this work is part of an effort to understand several experimental results , including the properties of the weakly doped cuprates,@xcite for which disorder effects are known to be important .
initially , we were motivated to conduct such theoretical work to better understand this transition in the material lial@xmath7ti@xmath8o@xmath9.@xcite the ordered and undoped material liti@xmath10o@xmath9 undergoes a transition to a superconducting phase@xcite around @xmath11 , and has been proposed to be related to the family of high-@xmath5 superconductors.@xcite if such a conjecture is correct , there must be reasonably large electron - electron interactions in this system .
the al - doped system undergoes a metal - insulator transition for @xmath12 , and recent theoretical work by one of us and co - workers have suggested that ( i ) the effects of disorder alone can not lead to such a transition@xcite ; ( ii ) inclusion of electron - electron interactions in the form of an on - site hubbard energy can lead to such a transition.@xcite this occurs by the suppression of the density of states at the fermi energy to near zero .
the results presented in this report are a natural continuation of such work .
in addition , other systems display such transition .
sarma et al.@xcite have investigated the electronic structure of lani@xmath13mn@xmath14o@xmath15 .
it is known that for a critical concentration of @xmath16 a metal - to - insulator transition is found .
the point - contact tunnelling conductance spectra , which provides information on the density of states , show a suppression at the fermi energy in the form of a downward cusp for both @xmath17 and @xmath18 .
the conductance is found to be proportional to @xmath19 , where @xmath20 is the bias voltage .
this behaviour was predicted by altshuler and aronov@xcite , and is similar to the results that we encountered in studies of lial@xmath7ti@xmath8o@xmath9.@xcite kim et al.@xcite have studied the transport and optical properties of srti@xmath13ru@xmath14o@xmath15 .
when @xmath21 decreases from @xmath22 to @xmath23 , srti@xmath13ru@xmath14o@xmath15 evolves from a correlated metal , srruo@xmath15 ( @xmath24 ) , to a band insulator , srtio@xmath15 ( @xmath25 ) . depending on @xmath21
, there are six types of electronic states .
the metal - to - insulator transition takes place at @xmath26 . as @xmath21 decreases from one , the concentration of conduction electron decreases , and the effective interaction strength increases .
these systems are very complicated , having several important @xmath6 orbitals for different conducting sites , and many different interaction , disorder and hopping energy parameters .
the theoretical considered below are a minimalist s approach to such interesting behaviour .
a simplified model that describes interacting electrons moving on a spatially disordered lattice is the so - called anderson - hubbard model .
its hamiltonian is given by @xmath27 electron annihilation / creation operators for site @xmath28 and spin @xmath29 are represented by @xmath30 and @xmath31 , respectively .
the spatially inhomogeneous environment in which the electrons move is accounted for by on - site energies , @xmath32 , which are random .
often , these are selected to be chosen from a uniform distribution , @xmath33 $ ] , and therefore @xmath34 is the energy scale characterizing the strength of the disorder .
interactions between electrons are accounted for by the intra - site hubbard interaction term , characterized by the energy scale @xmath3 .
the near - neighbour hopping frequency is denoted by @xmath35 , and all other symbols have their usual meaning .
it is hoped that this model can capture some of the essential physics of the metal - insulator transitions@xcite of disordered transition - metal oxides .
the electronic properties of systems described by this model hamiltonian are indeed complicated , as can be understood from the following reasons .
when disorder is sufficiently strong it can lead to the localization of electrons , and favours large electronic occupancies on sites with low on - site energies .
therefore , disorder generates both localization effects and an inhomogeneous distribution of electronic charge . there exists a critical disorder , @xmath36 , beyond which all states are localized and the system becomes an anderson insulator.@xcite in a spatially uniform half - filled system , electron - electron interactions also lead to the effective localization of electrons ; however , in contrast to disorder , interactions favour single occupancy of electrons on all sites . in general , at a critical interaction , @xmath37 , the system undergoes a transition from a metal to a mott insulator .
however , the mott transition may not take place when the electronic concentration is away from half - filling .
due to the randomness of the on - site energies , it is usual to address this model using numerical techniques .
also , except for some small clusters that can be solved exactly , large lattices have to be solved with the help of approximate schemes .
this difficulty notwithstanding , a small number of exact " numerical results are available , coming from both exact diagonalization ( ed)@xcite and monte carlo ( mc)@xcite studies . because of their relevance to our work , we mention one aspect of the results in two of these papers .
that is , some ed calculations showed the presence of a suppression of the dos at the fermi energy , in both one@xcite and two@xcite dimensions .
these results are consistent with the hf papers mentioned below .
therefore , this provides partial verification of the results based on the hf method , the method used in the remainder of this paper .
the hartree - fock ( hereafter hf ) method is well known , and its application to disordered systems is extensive .
the accuracy of the results was recently critiqued by us and coworkers,@xcite where it was shown that provided one allowed for sufficient magnetic degrees of freedom the energies and charge densities of ( small ) exactly solvable systems agreed well with those obtained from hf .
however , since the spin correlations are essentially those of pairs of classical spins of variable directions and lengths , and therefore do not include quantum fluctuations , hf is less successful at capturing the correct spin - spin correlations . as mentioned , the hf approximation has been applied in many studies of the anderson - hubbard and related models .
some examples are ( i ) a study of a two - dimensional point - defect model@xcite which represents the acceptors and donors in the high-@xmath5 cuprate la@xmath38sr@xmath14cuo@xmath9 through determining the magnetic phase diagram , ( ii ) a proposal of a novel inhomogeneous metallic phase in two dimensions@xcite as a combined effect of disorder and electronic interactions when their strengths are comparable , and ( iii ) a detailed study of the three - dimensional anderson - hubbard model , determining both magnetic and electric phase diagrams at half - filling.@xcite in refs .
@xcite , tusch and logan have focused on @xmath39-filling and some other fillings lower than @xmath40 . at @xmath39-filling with a disorder strength of @xmath41 ,
the density of states ( hereafter denoted by dos ) shows a suppression for both @xmath42 and @xmath43 . using the inverse participation ratio ( ipr ) technique , the system is determined to be metallic at @xmath42 and insulating at @xmath44 .
the ipr is compared with a threshold mean ipr ( obtained in the noninteracting limit ) that scales with system size to determine whether the system is metallic or insulating .
however , the effect of the interactions on the threshold mean ipr is unknown , as are the effects the unusual statistical distribution of the ipr , the latter having been discussed by mirlin.@xcite in part to circumvent such problems , here we focus on the characterization of the metallicity of a system via its optical conductivity . in the hf treatment of lial@xmath7ti@xmath8o@xmath9 by one of us and coworkers,@xcite a metal - to - insulator transition , again found from an examination of the behaviour of the ipr , was found . in that work
the local magnetic moments were restricted to lie along the @xmath45-axis , geometrically frustrated corner - sharing tetrahedral lattices were studied , and a quantum site percolation model of disorder was used .
open questions concern whether the suppression of the dos at the fermi level depends on the lattice type and the disorder model . to answer these questions
, we choose to study three - dimensional simple cubic lattices , which are unfrustrated , and consider a uniform box distribution type of disorder .
we also allow local magnetic moments to develop in the @xmath46-plane , which increases the spin degree of freedom , believed to be important in some circumstances.@xcite we also mention the effective - field theory analysis of local moment formation in disordered interacting fermi liquids by milovanovic , sachdev , and bhatt.@xcite the potential importance of such moments in the metal - insulator transition was left as an outstanding question , and will be one of the main aspects addressed in this paper . in this paper , we apply the real - space self - consistent hf method to electrons moving on simple cubic lattices at @xmath47-filling , with various strengths of interaction and disorder . here
we report our results from calculations of the dos and ac conductivity . when examining systems with a disorder strength of @xmath48 , we find that the dos and the low - frequency conductivity are enhanced by a weak interaction ( @xmath49 ) , and for this range of interactions there are no magnetic moments in the system , and the ac conductivity is drude - like . with a stronger interaction ( @xmath50 ) , a suppression of the dos at the fermi level and
qualitative changes ( non - drude - like ) in the low - frequency conductivity are found .
we find that concomitant with these changes in behaviour is the appearance of local magnetic moments in the system , although no evidence of magnetic ordering is present .
for this disorder strength and electron concentration a metal - to - insulator transition is likely to take place at a critical @xmath51 ; that is , roughly 3/4 of the noninteracting bandwidth .
we have also examined the weaker disorder strength of @xmath52 , and although one finds a small suppression of the dos with increasing hubbard interaction , for no value of @xmath53 do we find a metal - to - insulator transition .
for ordered systems , one may transform the hamiltonian to wave vector space , and then expand the many - particle wave function in a complete set of bloch wave functions in the corresponding periodic potential .
however , in disordered systems there is no translational symmetry , and working in wave vector space does not simplify the problem .
therefore , we will work in real space . in the real - space formulation of hf
theory ( _ e.g. _ , see page 349 of ref .
@xcite ) the disorder is treated exactly , and therefore this approximation allows for us to include in our calculations _ the effects of localization_. in the real - space formulation of hf theory the local hubbard interaction term is replaced by @xmath54 here , the terms that are proportional to fluctuations about the mean values squared are ignored .
also , we do not consider the expectation values that arise from superconducting correlations .
substituting eq .
( [ hubbardterm ] ) into the anderson - hubbard hamiltonian in eq .
( [ andersonhubbardham ] ) , one finds the hf effective hamiltonian given by @xmath55 where @xmath56 @xmath57 here , @xmath58 is the expectation value of the number of electrons with spin @xmath59 on site @xmath60 ; @xmath61 is the spin dependent effective on - site energy at site @xmath60 , and it is the sum of the original on - site energy plus the hubbard @xmath3 times the number of electrons with the opposite spin on that site .
therefore , in the absence of the last term one includes the effects of the hubbard interaction via an effective shift of the on - site energies .
the last term in eq .
( [ effham ] ) does include the @xmath62 effective local fields , and these guarantee that the effective hamiltonian is invariant under rotations in spin space . in the self - consistent formulation of hf
theory , one must ensure that the solutions of the effective one - electron hamiltonian given in eq .
( [ effham ] ) lead to local spin - resolved charge densities and effective local fields that satisfy eqs .
( [ eq : eff1],[eq : eff2 ] ) when the expectation values are taken with respect to the hf ground state wave function .
as mentioned above , one is required to solve the effective hamiltonian self - consistently . in a system with @xmath63 sites
, one has @xmath64 variational parameters that have to be determined numerically so that the ground - state energy is minimized .
( in addition , one includes a chemical potential to fix the electronic density at whatever concentration is desired . in our results below we focus on 1/4 filling . ) to solve the effective hamiltonian self - consistently , one begins with a random initial guess for the @xmath64 parameters that satisfies the constraint of fixing the total number of electrons , and then iterates to self consistency .
one needs to repeat the above procedure with many other initial guesses of the variational parameters and below we examine the self - consistent states having the lowest energy hf energies , @xmath65 .
after self - consistency is achieved for the hf solutions , we obtain effective single - electron energies , \{@xmath66 } , and the dos is then given by @xmath67 in our results shown below , we broaden each delta function using a gaussian function given by @xmath68 where @xmath69 is where @xmath70 takes maximum value , and @xmath6 is the standard deviation ( or the broadening width ) . to calculate the ac conductivity , one first obtains the imaginary part of the current - current correlation function using the kubo formula : @xmath71 where @xmath72 then , the real part of the ac conductivity is@xcite @xmath73 to calculate the ac conductivity numerically , we broaden each delta function in eq .
( [ i m a ] ) using a gaussian function given by eq .
( [ gaussian_func ] ) . in our results
below we state the broadening for each quantity .
while the anderson - hubbard model is often used to study disordered electronic systems with strong electron - electron interactions , it is not always clear how the parameter space studied for such models relates to real materials , particularly with regards to the strength of the disorder potential .
here we focus on the appropriate range of disorder strengths . when characterizing the strength of the potential energy it is usual to compare @xmath3 to the noninteracting bandwidth for electrons moving on an ordered lattice .
for a 3d simple cubic lattice that is @xmath74 .
weak interactions correspond to @xmath75 , intermediate to @xmath76 , and strong to @xmath77 .
this leads to the question , is it permissible to use the same characterization for the disorder strength ?
we have examined the @xmath78 anderson model ( at 1/4 filling - the same as the electronic concentration studied throughout this paper ) using a box distribution disorder
all on - site energies between @xmath79 and @xmath80 are equally probable .
we have then found the ac conductivity of this system for various @xmath81 , and here we discuss the results corresponding to @xmath82 and 6 .
if one fits our conductivity data to the real conductivity of a drude model , the latter given by @xmath83 one obtains estimates of the relaxation time and the dc conductivity shown in table [ table : drude ] .
the difference between the @xmath84 extrapolations quantifying the dc conductivities are striking , in that for @xmath85 we find @xmath86 , whereas for @xmath48 we find @xmath87 , implying that the conductivity of these systems differs by almost a factor of 60 .
clearly , the noninteracting @xmath48 ( or equivalently @xmath88 ) system corresponds to a bad metal " .
( however , while the @xmath48 system is a very poor conductor , for this disorder strength we are not approaching a 1/4-filled anderson insulator , and , in fact , our calculations and existing data@xcite allow us to determine that @xmath89 , with @xmath90 being the location of the mobility edge . )
.[table : drude ] conducting properties of noninteracting electrons moving on a 3d simple cubic lattice with a box distribution of disorder of strength @xmath34 , for an electron filling of 1/4 .
the @xmath91 and @xmath92 are found from fits of the ac conductivity to eq .
( [ eq : drude - real ] ) . [ cols="^,^,^",options="header " , ] further justification for this claims follows from a second way of characterizing the strength " of the disorder , and corresponds to evaluating the elastic mean free path , @xmath93 , in the born approximation .
one finds@xcite @xmath94 where @xmath95 is the lattice constant and @xmath45 is the coordination number for a particular lattice .
therefore , for @xmath85 , 2 , and 6 one has @xmath96 , 13 , and 1.4 , respectively . while one does not expect the born approximation to be accurate quantitatively , especially when it predicts @xmath97 , from these results it follows that for noninteracting electrons moving on a three dimensional simple cubic lattice with @xmath88 one has a very short mean free path .
so , while it might be conventional to refer to a @xmath98 correlated electronic systems as being intermediate coupling , for a disorder strength of @xmath88 the system remains metallic but the effect of the disorder potential is indeed very large .
as we see in the results below , it is only for such large disorder strengths , which by itself leads to substantial localization effects , that we find a disorder+correlation - driven metal - insulator transition .
( also , note that in our previous study of lithium titanate@xcite we employed a quantum site percolation model , which formally corresponds to setting all a site energies to be zero while those on the b sites are infinite - that is , the b sites are removed from the conduction path . with such a large disorder potential ( of infinite strength for a binary alloy model of disorder ) we did obtain a metal - insulator transition.@xcite ) as mentioned in the introduction , one of the motivations for this study was to better understand the conditions necessary for the appearance of a suppression in the dos at the fermi level .
previous work@xcite was performed on a lattice appropriate for the description of lial@xmath99ti@xmath8o@xmath9 , namely on a corner - sharing tetrahedral lattice .
also , that work was completed using a restricted hf theory : in terms of the above - presented formalism , the effective local fields @xmath100 were set to be zero , meaning that the spin degrees of freedom were restricted to develop along the quantization axis ( @xmath45 ) .
all of our results here are for a 3d simple cubic lattice unlike the corner - sharing tetrahedral lattice , this lattice is bipartite and therefore unfrustrated . in terms of the magnetic degrees of freedom , here we present results from our new hf calculations for two different situations .
first , we further restrict the system to be paramagnetic , meaning that we do not allow for the formation of _ any _ local moments this means that @xmath101 is forced to be equal to @xmath102 .
second , we relax the restriction of the moments being along the @xmath45-axis , and allow them to form in the @xmath46-plane , i.e. , @xmath103 is nonzero but real . density of states of a three - dimensional simple cubic lattice of size @xmath104 for an electronic filling of @xmath47 .
the interaction strength is @xmath105 , and the disorder strength is @xmath48 .
the nonparamagnetic hf result ( dashed curve ) shows a strong suppression of the dos at the fermi level , whereas the paramagnetic hf result ( solid curve ) shows no such suppression .
the broadening used corresponds to @xmath106.,title="fig:",width=7 ] + some of our results for the dos for a three - dimensional simple cubic lattice are shown in fig .
[ dos_3dsc_l18_vdis6.0_u11.5_edensity0.5 ] .
the lattice size is @xmath104 , and the electronic filling factor is @xmath47 .
the interaction strength is @xmath105 , where the dos at the fermi level has its largest suppression , and the disorder is modelled with a box distribution for a disorder strength of @xmath48 .
the nonparamagnetic hf result shows a strong suppression of the dos at the fermi level ; however , the paramagnetic hf result shows no suppression at all . since we only found a suppression of the dos with nonparamagnetic hf solutions , both with @xmath107 ( in ref .
@xcite ) and @xmath103 being real , the magnetic moments are essential to the suppression and to the metal - insulator transition ( at least within the hf context ) .
the moments found for the real @xmath103 hf ground state are strongly noncollinear _ e.g. _ , see our discussion in ref .
therefore , the restriction of collinear moments employed in refs .
@xcite is not important to the suppression .
also , the type of lattice does not matter because the suppression is found both for the frustrated corner - sharing tetrahedral lattices ( in ref .
@xcite ) and for the unfrustrated simple cubic lattices studied in this paper . also , the type of disorder does not matter , since the suppression is found either with a quantum site percolation model ( in ref .
@xcite ) or with a uniform box distribution type of disorder . to better quantify the presence of the local moments we have calculated the average magnitude of the moment per electron .
we define an edwards - anderson - like order parameter @xmath108 where @xmath109 is the spin operator on site @xmath60 , @xmath110 are the pauli matrices , and @xmath111 ( @xmath63 ) is the number of electrons ( lattice sites ) .
the quantity @xmath112 is similar to the edwards - anderson order parameter@xcite in spin glass theory , which is used to distinguish between glass and nonglass phases,@xcite but here we use @xmath112 as a characterization of local magnetic moment formation .
we have considered many different parameter sets and lattice sites ( see discussions below ) , and our results for @xmath112 are shown in fig .
[ localmoments ] .
we see that are no local moments in the noninteracting ( @xmath113 ) or weakly interacting systems ( @xmath49 ) for @xmath114 and 6 . for the @xmath104 lattice and @xmath115 and @xmath48 ,
@xmath112 is only about @xmath116 , and for the @xmath104 lattice and @xmath115 and @xmath52 we find @xmath117 ; therefore , this quantity is expected to be zero in the thermodynamic limit . for larger @xmath53
we do find moments , and for a given interaction strength @xmath112 increases as one increases the strength of the disorder . in all of the subsequent data that we show having a suppression of the dos at the fermi level , and non - drude - like conductivity ,
the average magnitude of the local magnetic moments @xmath118 * is always nonzero . * the average magnitude of the magnetic moment per electron ( @xmath112 ) _ vs. _ interaction strength ( @xmath53 ) for a @xmath119 lattice with disorder strength @xmath48 ( squares ) , and for an @xmath104 lattice with @xmath52 ( diamonds ) and @xmath120 ( circles).,title="fig:",width=8 ] + now , we consider the variation of the dos in more detail .
we have calculated the dos for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath119 for various strengths of interaction @xmath113 , @xmath22 , @xmath121 , , @xmath122 , @xmath43 , and @xmath123 .
the on - site energies obey a uniform box distribution , and the disorder strength is @xmath48 .
results are averaged over @xmath124 realizations of disorder a limited study using more realizations of disorder found no qualitative changes when more realizations were employed .
the maximum differences for the self - consistency are between @xmath125 ( for @xmath126 and @xmath121 ) and @xmath127 ( for @xmath105 ) .
( typically , the average differences were at least one order magnitude smaller than the maximum differences . ) in fig .
[ dos_l12_w6_box ] , we show the dos of this system obtained with @xmath128 .
( the dos curves are smooth and well separated from one another around the fermi level , whereas for smaller broadening these curves are bumpy and not well separated , and we therefore choose @xmath129 as the broadening width for the dos of this system . ) density of states of the anderson - hubbard model on a @xmath47-filled three - dimensional simple cubic lattice of size @xmath119 for a disorder strength of @xmath48 .
results without interactions ( @xmath113 ) and with interaction strengths @xmath126 , @xmath121 , , @xmath120 , @xmath43 , and @xmath123 are averaged over @xmath124 realizations of spatial disorder .
the vertical line represents the fermi level .
the broadening width is @xmath129.,title="fig:",width=8 ] + as shown in fig .
[ dos_l12_w6_box ] , for @xmath113 to @xmath130 , each dos does not show any suppression .
in fact , the dos at the fermi level is enhanced as the interaction is turned on and then increased .
the suppression appears for the first time when @xmath131 , and the amount of the suppression increases as @xmath53 is increased .
we find that the maximum suppression , for 1/4 filling , occurs around @xmath105 , although there is little difference between the dos for this @xmath53 and other hubbard interactions close to this strength .
extracting the low - frequency behaviour of the conductivity is nontrivial , and we briefly outline the numerical approach taken . we have examined the ac conductivity obtained with different values of the gaussian broadening width @xmath6 in eq .
( [ gaussian_func ] ) , and have selected a value for @xmath6 based on the following : for a three - dimensional simple cubic lattice of size @xmath119 with disorder strength @xmath48 and interaction strength @xmath42 , we calculated the ac conductivity with broadening widths @xmath132 , @xmath133 , and @xmath134 ( results not shown ) . because the broadening width @xmath6 has a finite value , the contributions of frequencies between @xmath23 and @xmath6 makes the imaginary part of the current - current correlation function @xmath135 nonzero around @xmath136 . as a result ,
when calculating @xmath137 by dividing @xmath138 by @xmath139 , we obtain diverging @xmath137 around @xmath136 .
the divergence is simply an artifact of numerical procedure used ( a broadening width @xmath6 that is finite ) , and is not associated with the physics of the system being investigated .
therefore , we have to cut off the conductivity curves at low frequencies ( @xmath140 ) where they `` turn up '' . recalling that the dos data of fig .
[ dos_l12_w6_box ] used @xmath141 , we choose @xmath142 as the broadening width for the conductivity because the corresponding conductivity curves are smooth and still retain much of the low - frequency behaviour simply , we are not forced to discard as much low - frequency data as we would with @xmath141 .
( as reviewed in the discussion , this low - frequency behaviour is an important quantity to know . )
we now discuss our results , juxtaposing dos and ac conductivity data for each parameter set .
first , we discuss results for studies done on lattices of size @xmath119 with a disorder strength of @xmath143 ; an average over @xmath124 realizations of disorder is used for each data set .
we increase the interaction strength @xmath53 from zero up to roughly @xmath144 .
density of states ( a ) and ac conductivity ( b ) for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath119 with a disorder strength @xmath143 with no interactions ( @xmath113 ) and with hubbard interaction of strengths @xmath126 , @xmath121 , and @xmath130 .
the dos do not show any suppression at the fermi level , and in fact increase as the interaction strength @xmath53 increases .
the low - frequency conductivity is enhanced as @xmath53 is increased , and the ac conductivity is drude - like .
we have used a broadening corresponding to @xmath141 for the dos and @xmath145 for the conductivity.,title="fig:",width=8 ] + the results for the noninteracting case ( @xmath113 ) and for interactions @xmath146 , @xmath121 , and @xmath130 are shown in fig .
[ dos_conductivity_l12_ne864_w6_u0_1_2_3 ] .
the solid lines in ( a ) and ( b ) represent the dos and the ac conductivity for the noninteracting electrons , respectively .
the ac conductivity has a shape that is typical of a metal , but the low - frequency peak is broad .
when we turn on the interaction to @xmath146 ( dotted lines ) , we see an enhancement in the dos at the fermi level .
concomitantly , the low - frequency ac conductivity also increases from the noninteracting value .
these enhancements could be a result of the screening of the disorder by the hubbard interactions . as the interaction strength @xmath53 increases to @xmath121 (
dashed lines ) and @xmath130 ( dash - dotted lines ) , we find that the dos at the fermi level and the low - frequency ac conductivity are both further enhanced .
therefore , for a disorder of @xmath143 at 1/4 filling , a weak interaction of @xmath49 enhances the low - frequency ac conductivity due to an increase in the dos at the fermi level .
recall , from fig .
[ localmoments ] , that for this range of interaction strengths no local moments are formed .
density of states ( a ) and ac conductivity ( b ) for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath119 with a disorder strength @xmath143 and hubbard interaction strengths @xmath115 , @xmath124 , @xmath147 , and @xmath120 .
each of the dos for @xmath148 , @xmath147 , and @xmath120 shows a suppression at the fermi level , and the amount of the suppression increases as @xmath53 increases .
the low - frequency conductivity decreases as @xmath53 increases , and the ac conductivity is no longer drude - like.,title="fig:",width=8 ] + the results for interactions @xmath115 , @xmath124 , @xmath147 , and @xmath120 are shown in fig .
[ dos_conductivity_l12_ne864_w6_u3_4_5_6 ] .
a suppression of the dos at the fermi level first appears for @xmath148 , and its value is smaller than that of the noninteracting electrons and that of the @xmath115 system .
the low - frequency conductivity for @xmath131 is also smaller than that of the @xmath149 system . as the interaction strength @xmath53 increases to the values of @xmath147 and @xmath120 , the amount of the suppression of the dos at the fermi level increases , and the low - frequency conductivity decreases .
each of the ac conductivity curves still extrapolates to a nonzero value when the frequency @xmath150 .
we note that starting from @xmath151 , the low - frequency conductivity is smaller than that of the noninteracting electrons .
more importantly , the ac conductivity for @xmath148 , @xmath147 , and @xmath120 is no longer drude - like .
in fact , our conductivity curves are similar to the localization - enhanced drude theory of mott and kaveh,@xcite and such conductivities have been observed experimentally previously in the metallic phase near the metal - insulator transition in experimental work on conducting polymers.@xcite we show the results for interaction strengths @xmath42 , @xmath43 , and @xmath123 in fig .
[ dos_conductivity_l12_ne864_w6_u6_9_11.5 ] . compared to @xmath42 , the dos at the fermi level
gets further suppressed .
however , the amount of the suppression for @xmath44 and @xmath123 are not that different .
it is very clear that , on the scale of fig .
[ dos_conductivity_l12_ne864_w6_u6_9_11.5 ] , the dc conductivity for @xmath44 and @xmath123 are both zero . the ac conductivity for @xmath44 and @xmath123 both increase as @xmath152 increases up to the value of @xmath153 .
we fit the ac conductivity for @xmath42 , @xmath43 , and @xmath123 with power - law relations . for @xmath42 ,
the system appears to be metallic , and we fit its ac conductivity using the following equation : @xmath154 where @xmath155 , @xmath156 , and @xmath157 are the parameters to be determined . the range of frequency over which we chose to fit the data is @xmath158 ( recall that we used @xmath145 in producing these conductivity curves ) , and we obtain @xmath159 here , the uncertainties are one standard deviation .
we learn from this power - law fit that the dc conductivity for the system with @xmath160 is @xmath161 , which is quite close to but still above zero .
therefore , we expect that the system is a metal .
we also see that the exponent of the frequency is @xmath162 , which is the exponent ( @xmath163 ) that appears in the ac conductivity of noninteracting electrons with strong disorder , in the metallic regime close to the transition,@xcite where @xmath164 near @xmath136 goes as @xmath165 .
density of states ( a ) and ac conductivity ( b ) for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath119 with a disorder strength @xmath143 and hubbard interactions of strength @xmath42 , @xmath43 , and @xmath123 .
each dos shows a suppression at the fermi level , and the low - frequency conductivity is also suppressed both suppressions increase with increasing interactions . the ac conductivity for @xmath42 extrapolates to a nonzero dc conductivity as @xmath166
, however , the dc conductivity for @xmath44 and @xmath123 are both very close to zero.,title="fig:",width=8 ] + we fit the ac conductivity for @xmath44 and @xmath123 over the ranges of @xmath167 $ ] and @xmath168 $ ] , respectively . setting @xmath169 we obtain @xmath170 for @xmath44 and @xmath171 for @xmath105 .
these exponents are very close to @xmath121 , and , again , this is the same as the exponent for noninteracting electrons for an even stronger disorder , namely for a disordered system in the insulating phase,@xcite where @xmath172 in the low - frequency regime .
therefore , the system with @xmath48 is metallic for @xmath173 and insulating for @xmath174 . by examining other @xmath53 around 9 ( data not shown )
we can identify that the metal - to - insulator transition for this disorder and electronic filling indeed takes place at some critical @xmath175 .
it is always desirable to complete studies with larger lattices , where finite - size effects are hopefully less punitive . in our study
the benefits of having such data are ( i ) better energy resolution of the dos near the fermi level , and below we make clear the usefulness of such data ; and ( ii ) better resolution of the ac conductivity at low frequencies .
using our computer resources the largest lattice that we have managed to treat , and are able to explore various parameters sets , is an @xmath104 lattice with @xmath47-filling . for this lattice with our hf formulation employing real @xmath176
, it takes about _ one month _ to achieve self - consistency in the hf calculations for a single realization of disorder , and due to this enormous amount of time we solve the problem only for one realization of disorder . for a disorder strength of @xmath48 and interaction strengths of @xmath113 , @xmath130 , @xmath120 , and @xmath43 ,
we show the resulting dos in fig .
[ dos_l18_w6_box ] .
( due to the higher number of energy levels per unit frequency we have used 1/2 the broadening width ( @xmath106 ) as we did for the 12@xmath177 studies . )
we note that the behaviour of the dos in this larger lattice is very similar to the @xmath119 lattice that we have studied earlier .
density of states for the anderson - hubbard model on a @xmath47-filled three - dimensional simple cubic lattice of size @xmath104 . for an intermediate disorder of @xmath48 , interaction strengths of @xmath113 , @xmath130 , @xmath120 , and @xmath43 are considered .
the gaussian broadening width is @xmath178.,title="fig:",width=8 ] + the corresponding ac conductivity for the same system is shown in fig .
[ conductivity_l18_w6_box ] .
note that this data corresponds to a very small gaussian broadening of only @xmath179 , and therefore we can obtain data down to very low frequencies .
we clearly see that the data is similar ( qualitatively _ and _ quantitatively , indicating that our data does not suffer from strong finite - size effects ) to that shown earlier for a 12@xmath177 lattice .
further , the extrapolation of the conductivity in the zero frequency limit makes clear the nonconducting nature of the @xmath44 system .
ac conductivity of the anderson - hubbard model on a @xmath47-filled three - dimensional simple cubic lattice of size @xmath104 for a disorder potential of strength @xmath48 , for interaction strengths of @xmath113 , @xmath130 , @xmath120 , and @xmath43 .
the gaussian broadening width is @xmath179.,title="fig:",width=8 ] + so far , we have studied systems with a disorder strength of @xmath48 . as mentioned earlier
, this corresponds to a noninteracting disordered electronic system with a very low dc conductivity that is , this corresponds to a bad metal " .
we found a suppression of the dos at the fermi level for interaction strengths @xmath50 .
when this happens , we find that the ac conductivity is no longer drude - like , and for strong enough interactions , @xmath180 , the system no longer possesses a metallic conductivity .
a natural question then arises : for smaller strengths of disorder does one still obtain a metal - insulator transition ? in this subsection we present our results for a weaker disorder corresponding to @xmath52 .
compared to the noninteracting @xmath48 system , the noninteracting @xmath52 system has a mean free path and dc conductivity almost an order of magnitude larger . for such disorder
we are forced to use a large lattice because the energies are strongly degenerate in small lattices . in fig .
[ dos_l18_w2 ] we show the dos for an @xmath104 simple cubic lattice with interaction strengths @xmath115 , @xmath120 , @xmath43 , @xmath123 , @xmath144 , and @xmath181 .
density of states for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath104 with a box distribution type of disorder of strength @xmath182
( @xmath124 realizations of disorder ) and hubbard interaction of strengths @xmath115 , @xmath120 , @xmath43 , @xmath123 , @xmath144 , and @xmath181.,title="fig:",width=302 ] + for this weak disorder strength , the interaction strength of @xmath115 does not produce a suppression in the dos at the fermi level .
however , for @xmath42 , @xmath43 , @xmath123 and @xmath144 a suppression in the dos appears at the fermi level , and the amount of suppression is deepest at @xmath105 . for @xmath183
we find that the suppression has been eliminated .
ac conductivity for a @xmath47-filled three - dimensional simple cubic lattice of size @xmath104 with a box distribution type of disorder of strength @xmath182
( @xmath124 realizations of disorder ) and hubbard interaction of strengths @xmath115 , @xmath120 , @xmath43 , @xmath123 , and @xmath144.,title="fig:",width=302 ] + in fig .
[ conductivity_l18_w2 ] , we show the ac conductivity for the same system with interaction strengths up to @xmath184 .
the conductivity for @xmath115 is clearly metallic and drude like , and extrapolates to roughly @xmath185 in the low - frequency limit .
this value is roughly half that of noninteracting electrons with the same strength of disorder . with increasing @xmath53 the conductivity at low frequencies
is lowered , with @xmath44 and 11.5 having very similar behaviour . however , for @xmath184 the conductivity increases near @xmath186 this behaviour persists for even larger hubbard energies .
therefore , this disorder strength does not lead to a nonmetallic conductivity for any hubbard energy that we have studied . to aid in the understanding the above results , namely that for a larger disorder one does obtain a metal - insulator transition ,
whereas for smaller disorder one does not , we have examined the local charge densities for these parameter sets . in fig .
[ charge_hist_l18_w2and6 ] , we show histograms of the local charge density for the @xmath104 lattice with disorder strength @xmath48 for @xmath113 and @xmath43 ( a ) , and @xmath52 for @xmath113 and @xmath123 ( b )
. histograms of the local charge density for the @xmath104 lattice .
( a ) @xmath113 and @xmath43 for a disorder @xmath48 .
( b ) @xmath113 and @xmath123 for a weak disorder @xmath52.,title="fig:",width=302 ] + for disorder strength @xmath48 , when the system is noninteracting the charge density is spread out over a large range , with one large peak close to @xmath187 .
that is , in the absence of interactions and for this seemingly large disorder ( see previous discussion ) the ground state corresponds to a very small occupation of sites with large on - site energies , with a increasing occupation as the on - site energies decrease .
however , when the interaction strength is strong ( @xmath44 ) , the original peak at the small charge density does not change much , but a large new peak around @xmath188 forms .
that is , for this disorder the interactions lead to very different charge densities , and correspond to those of an insulating system .
the contrasting situation for weak disorder is clear from the figure the charge distribution is quite similar between the noninteracting ( @xmath113 ) and strongly interacting ( @xmath105 ) systems , and in both cases one has metallic conduction .
we have also plotted the histograms for the effective on - site energies , that is @xmath61 from eq . (
[ eq : eff1 ] ) averaged over both spins , since this quantity has been suggested to be important in the understanding of the metallicity of such ground states.@xcite however , the histograms for @xmath189 , for the same parameter sets as those used in fig . [ charge_hist_l18_w2and6 ] , show essentially no difference for these interacting hf ground states , one of which we found to be metallic and one insulating .
our numerical analysis of the real - space self - consistent hf treatment of the anderson - hubbard model away from 1/2 filling on 3d simple cubic lattices gives rise to three main conclusions , each of which we discuss in relation to previously published theoretical and experimental work . as preliminary work to our study
, we first characterized the consequences of different strengths of disorder for noninteracting electrons . for a disorder strength much less than the noninteracting bandwidth , given by @xmath74 , we find a large dc conductivity , consistent with theoretical predictions of a large mean free path . however . for a disorder strength that is 1/2 of this bandwidth , we find a greatly reduced dc conductivity , and applying eq .
( [ eq : mfp ] ) , estimate a mean free path of the same order as the lattice constant . therefore , for the latter disorder strength the effects of the localization of electronic eigenstates is considerable .
note that we have treated the disorder of the anderson - hubbard model exactly , and have solved systems having large lattices with as many as 18@xmath177 sites , thereby allowing for such localization physics to be present in our hf ground states .
at 1/4 filling and a disorder strength of @xmath48 , we find a critical interaction strength of @xmath51 leads to a metal - insulator transition , whereas a smaller disorder strength of @xmath52 does not lead to any such transition for any value of @xmath53 that we have studied .
therefore , one concludes that a disorder that is sufficiently strong and thereby leads to strong localization effects is required to obtain such a transition . of course , as always throughout our conclusions , this is what one may conclude via a hf treatment of this model .
the importance of local moment formation in the disordered metallic state was made clear in the seminal work of milovanovic , sachdev , and bhatt.@xcite the importance of such moments in obtaining the metal - insulator transition is demonstrated in our work .
the results from fig .
[ dos_3dsc_l18_vdis6.0_u11.5_edensity0.5 ] make clear the necessity of allowing for local moments ( in a hf treatment ) to form if one is to obtain a suppression of the dos .
further , the results from fig .
[ localmoments ] , along with the dos curves shown in figs .
[ dos_conductivity_l12_ne864_w6_u0_1_2_3 ] , [ dos_conductivity_l12_ne864_w6_u3_4_5_6 ] and [ dos_conductivity_l12_ne864_w6_u6_9_11.5 ] , demonstrate that to obtain such a suppression of the dos local magnetic moments must be present .
the lattice and model of disorder do not influence whether or not such moments appear .
we note that similar suppressions of the dos within hf treatments of this and related models have been found by logan and tusch,@xcite one of us and coworkers,@xcite and most recently by shinaoka and imada.@xcite in all of these studies a restricted hf formulation that required magnetic moments to form along some single chosen quantization axis was employed , again emphasizing that importance of allowing this degree of freedom .
however , whether or not twisted spins " that point in all directions possible are allowed@xcite does not seem to be required . due to the considerable interest in the dmft treatment of correlated electrons , it seems appropriate to note that theories going beyond the single - site kkr - cpa treatments of disorder within dmft , such as ref .
@xcite , also find non - fermi - liquid behaviour .
more recently , a variant of dmft that , like this paper , also tries to treat the disorder exactly , and used the hf approximation in the evaluation of the off - diagonal self energy , has found@xcite zero bias anomalies similar to those shown in this paper .
* when such moments develop , in addition to the suppression of the dos , one obtains a non - drude - like ac conductivity . with sufficiently strong disorder and interactions ,
the dc conductivity is suppressed to zero and one obtains a metal - insulator transition . at least within our hf treatment , just above and just below the interaction strengths leading to the metal - insulator transition
, one finds @xmath190 and @xmath191 power law behaviour , the same as one finds for noninteracting electrons .
whether this is a by - product of hf being an effective one - electron theory remains as an outstanding question .
we note that novel ac conductivities , including results that are beyond that which we conducted , namely that included the temperature dependencies , were also seen in the recent work of kobayashi _
et al._@xcite these authors used the same hf decomposition as that used here .
however , unlike our work these authors focussed on 1/2-filled systems in two dimensions , and , in particular , on the properties of the novel 2d metallic phase that was predicted based on earlier hf work.@xcite these authors do not propose specific frequency dependencies , unlike the results we give in eqs .
( [ metallic_powerlaw],[insulating_powerlaw_fit_u9],[insulating_powerlaw_fit_u11.5 ] ) , so the potentially important effect of differing dimensionalities remains unknown . as mentioned in the introduction ,
the suppression of the dos at the fermi energy and the associated metal - insulator transition found in our earlier work on lithium titanate,@xcite was the original motivation for this study . since that paper new exact diagonalization and monte carlo results@xcite have made clear ( albeit in two dimensions ) that such physics is not an artifact of using a hf approximation .
subsequent hf work,@xcite using averages over very large numbers of complexions of disorder , has found qualitatively similar results , but has also managed to produce many more energy eigenvalues close to the fermi energy , and thereby better probe the functional form of the suppression of the dos found in all of these papers . the origin of the suppression of the dos is not determined by the present study , although we have placed constraints on what physics ( local moments and sufficiently strong disorder ) must be included in a model that will produce such behaviour . in terms of existing theories
, we note that for weak disorder and weak interactions one may complete a derivation@xcite of the altshuler and aronov result@xcite for the anderson - hubbard model , and one finds that one gets an increase in the dos as @xmath53 increases from zero .
that is , one finds @xmath192 where @xmath193 is some positive constant and @xmath194 is the dos of the noninteracting system at the fermi energy .
this is consistent with the @xmath48 and @xmath195 and 3 dos curves that we found for the 12@xmath177 lattice , as well as the @xmath48 and @xmath113 and 3 curves for the 18@xmath177 lattice ( even though the weak disorder @xmath196 assumption of the altshuler and aronov theory is not obeyed for our @xmath48 system ) . for larger @xmath53
our hf numerics show that the sign of the shift of the dos changes , in that one has a suppression rather than enhancement , as is clear from figs .
[ dos_conductivity_l12_ne864_w6_u0_1_2_3 ] and [ dos_conductivity_l12_ne864_w6_u3_4_5_6 ] .
given that we find that one must have local moments to have any suppression , even for weak disorder ( see figs .
[ localmoments ] and [ dos_l18_w2 ] ) , it s clear that one must extend the weak disorder paramagnetic fermi altshuler and aronov theory , which remains an outstanding problem .
however , a completely different idea was made in a recent hf and exact diagonalization study,@xcite and these authors proposed that while the altshuler and aronov form of @xmath197 may be adequate away from the fermi energy , very near the fermi energy one instead finds a so - called soft hubbard gap .
they showed that such physics in three dimensions gives rise to @xmath198 for one - dimensional systems with strong localization effects , these authors provide independent justification for their theory .
however , to access the energies very close to @xmath199 in both one and three dimensions , averages over very large numbers of complexions of disorder were carried out.@xcite our numerics do not have such good resolution , but we have examined dos data for an @xmath104 lattice for @xmath48 and @xmath105 , now averaged over three different complexions of disorder .
first , we note that with a broadening roughly 1/5 of that shown in our fig .
[ dos_l18_w6_box ] we find that the dos indeed does indeed go to zero at @xmath199 ; this is consistent with refs.@xcite .
second , fitting the dos data with these smaller broadening factors , for @xmath200 we find that eq .
( [ eq : aanofw ] ) fits the data well , and we obtain @xmath201 .
therefore , we find the same exponent as in the theory of altshuler and aronov.@xcite however , while our data for @xmath202 is sparse and therefore not completely reliable , the functional form proposed in eq .
( [ eq : imadanofw ] ) indeed fits our dos curves reasonably well .
our limited numerics therefore support the soft hubbard gap proposal@xcite at energies very close to @xmath199 .
finally , we mention the relationship of our work to some experimental results on transition metal oxides that were mentioned in the introduction .
sarma et al.@xcite have investigated the electronic structure of lani@xmath13mn@xmath14o@xmath15 .
the conductance is proportional to @xmath19 , where @xmath20 is the bias voltage .
as mentioned above , such a form fits our data very well .
kim et al.@xcite have studied the transport and optical properties of srti@xmath13ru@xmath14o@xmath15 .
their low frequency optical conductivity data is very similar to that which we obtained within hf .
our data also agrees with the optical conductivity found in the disordered metallic phase of conducting polymers.@xcite the analysis of this data , based on the localization modified drude model,@xcite leads to the same 0.5 exponent as we obtained in our unbiased power law fitting , the result of which is given in eq .
( [ metallic_powerlaw_fit ] ) .
we wish to thank david johnston , farhad fazileh , dariush heidarian , nandini trivedi and bill atkinson for numerous helpful communications at the beginning of this work .
this work was supported in part by the nserc of canada . | the anderson - hubbard model is considered to be the least complicated model using lattice fermions with which one can hope to study the physics of transition metal oxides with spatial disorder .
we have completed a numerical investigation of this model for three - dimensional simple cubic lattices using a real - space self - consistent hartree - fock decoupling approximation for the hubbard interaction . in this formulation
we treat the spatial disorder exactly , and therefore we _ account for effects arising from localization physics_. we have examined the model for electronic densities well away 1/2 filling , thereby avoiding the physics of a mott insulator .
several recent studies have made clear that the combined effects of electronic interactions and spatial disorder can give rise to a suppression of the electronic density of states , and a subsequent metal - insulator transition can occur .
we supplement such studies by calculating the ac conductivity for such systems .
our numerical results show that weak interactions enhance the density of states at the fermi level and the low - frequency conductivity , there are no local magnetic moments , and the ac conductivity is drude - like .
however , with a large enough disorder strength and larger interactions the density of states at the fermi level and the low - frequency conductivity are both suppressed , the conductivity becomes non - drude - like , and these phenomena are accompanied by the presence of local magnetic moments .
the low - frequency conductivity changes from a @xmath0 behaviour in the metallic phase , to a @xmath1 behaviour in the nonmetallic regime . for intermediate disorder at 1/4 electronic filling , a metal - to - insulator transition is predicted to take place at a critical @xmath2 ( @xmath3 being the hubbard interaction strength and @xmath4 the electronic band width ) . _
our numerical results show that the formation of magnetic moments is essential to the suppression of the density of states at the fermi level , and therefore essential to the metal - insulator transition .
_ at weaker disorder a small lessening of the density of states at the fermi level occurs , but screening suppresses the spatial disorder and with increasing interactions no metal - insulator transition is found . |
the deformation theory of non - orientable surfaces deals with the problem of studying parameter spaces for the different dianalytic structures that a surface can have .
it is an extension of the classical theory of teichmller spaces of riemann surfaces , and as such , it is quite rich . in this paper
we study some basic properties of the teichmller spaces of non - orientable surfaces , whose parallels in the orientable situation are well known .
more precisely , we prove an uniformization theorem , similar to the case of riemann surfaces , which shows that a non - orientable compact surface can be represented as the quotient of a simply connected domain of the riemann sphere , by a discrete group of mbius and anti - mbius transformation ( mappings whose conjugates are mbius transformations ) .
this uniformization result allows us to give explicit examples of teichmller spaces of non - orientable surfaces , as subsets of deformation spaces of orientable surfaces .
we also prove two isomorphism theorems : in the first place , we show that the teichmller spaces of surfaces of different topological type are not , in general , equivalent .
we then show that , if the topological type is preserved , but the signature changes , then the deformations spaces are isomorphic .
these are generalizations of the patterson and bers - greenberg theorems for teichmller spaces of riemann surfaces , respectively .
a riemann surface @xmath0 is a topological surface @xmath1 with a complex structure @xmath2 , that is , a covering of @xmath1 by charts with holomorphic changes of coordinates .
since holomorphic functions have positive jacobian , it turns out that riemann surfaces are orientable .
the natural generalisation to the case of non - orientable surfaces is that of a * dianalytic structure * , where we require that the changes of coordinates are either holomorphic or anti - holomorphic ( the complex conjugate is holomorphic ) .
a pair @xmath0 , where @xmath1 is a surface and @xmath2 is a dianalytic structure , is called a * klein surface*. in particular , riemann surfaces are klein surfaces .
it is classical fact that any klein surface can be represented as @xmath3 , where @xmath4 is either the riemann sphere , the complex plane or the upper half plane , and @xmath5 is a group of dianalytic bijections of @xmath4 . except for a few ( finite number of ) cases ,
klein surfaces are covered by the upper half plane ; these are called * hyperbolic surfaces*. a compact non - orientable surface @xmath1 is the connected sum of @xmath6 ( real ) projective planes ; @xmath6 is called the * genus * of the surface .
observe that here we use the genus in the topological sense ; some authors ( in particular , @xcite ) use the so - called arithmetic genus , which is equal to @xmath7 .
a non - orientable surface is hyperbolic if and only if @xmath8 .
in the first result of this paper , we prove a uniformization theorem , by groups which are more suitable for computations that groups acting on the upper half plane .
[ thm - unif ] let @xmath1 be a compact non - orientable surface of genus bigger than @xmath9 .
then there exists a kleinian group @xmath10 , acting discontinuously on a simply connected set @xmath11 of @xmath12 , and an antiholomorphic function @xmath13 , such that : + 1 .
@xmath14 for all @xmath15 ; @xmath16 ; + 2 .
@xmath17 is isomorphic to the complex double @xmath18 of @xmath1 ; + 3 .
@xmath13 is of the form @xmath19 , with @xmath20 ; + 4 .
@xmath21 , where @xmath5 is the group generated by @xmath10 and @xmath13 ; + 5 .
@xmath5 is unique up to conjugation by mbius transformations . here by a kleinian group we mean a group of mbius transformations that acts discontnuously on a non - empty open set of the riemann sphere .
the * complex double * of @xmath1 is a riemann surface @xmath18 , together with a unramified double cover @xmath22 . if @xmath1 is hyperbolic , then @xmath18 is also hyperbolic ( see 2 below )
. let @xmath23 denote the set of dianalytic structures , on the non - orientable surface @xmath1 , that are compatible with the differential structure induced by @xmath2 .
the quotient of @xmath23 by the group of diffeomorphisms homotopic to the identity ( acting by pullback , see
@xmath24 ) , is the * teichmller space * @xmath25 of @xmath1 .
it has a natural real analytic structure given by projecting the natural structure of @xmath23 .
it is not hard to prove that @xmath25 embedds in the teichmller space of @xmath18 ( see @xmath24 ) . combining this embedding with theorem [ thm - unif ] and the results of i. kra in @xcite
, we can give presentations for the deformation spaces of some non - orientable surfaces .
as an example , we compute the teichmller space of a surface of genus @xmath24 .
[ thm - ex1 ] the space @xmath25 of a non - orientable surface of genus @xmath24 can be identified with the set of points @xmath26 of @xmath27 , such that @xmath28 we introduce the concept of puncture on a non - orientable surface as a generalisation of the corresponding idea on riemann surfaces : a puncture is a domain on @xmath1 , homeomorphic to the unit disc minus the origin , that can not be completed to be homeomorphic to the unit disc , and such that any change of coordinates in the domain is holomorphic .
the above theorems extends easily to the case of surfaces with punctures .
for example , we can identify the deformation space of a surface of genus @xmath29 with two punctures .
[ thm - ex2 ] the space @xmath25 , where @xmath1 is the ( real ) projective plane with two puntures , can be identified with the set of points of the upper half plane with imaginary part bigger then @xmath29
. one can define a * klein hyperbolic orbifold * as a non - orientable surface @xmath1 , with finitely many ( maybe zero ) punctures , such that the covering from the upper half plane to @xmath1 is ramified over a finiter number of points .
the surface @xmath18 carries an anticonformal involution @xmath30 , such that @xmath31 .
we have that @xmath25 can be identified with the set of fixed points of the anticonformal involution @xmath32 , induced by @xmath30 in @xmath25 .
we say that the teichmller spaces of two non - orientable surface @xmath33 and @xmath34 are * real isomorphic * , if there exists a biholomorphic mapping @xmath35 , such that @xmath36 .
the following result is a generalisation of the bers - greenberg isomorphism for riemann surfaces .
[ bers - greenberg theorem for non - orientable surfaces ] if @xmath37 , @xmath38 , are two non - orientable hyperbolic orbifolds , with the same genus and number of ramification points , then the spaces @xmath39 and @xmath40 are real isomorphic.[thm - bers ] this paper is organized as follows : in 2 we prove the uniformization theorem ; 3 contains the proof of theorem [ thm - bers ] and other results about isomorphisms of deformation spaces ; finally ,
4 has the examples .
* akcnowledgements * : i would like to thank r. r. simha , for useful discussions .
i would like also to thank the tata institute of fundamental research , for providing me with excellent research facilities .
classically , hyperbolic klein surfaces are uniformized as the quotient of the upper half plane by a discrete group of dianalytic self - homeomorphisms ( mbius and anti - mbius transformations with real coefficients ) , known as nec groups . in this section
, we will prove a uniformisation theorem by a different type of groups , which are more suitable for computations .
we will use these groups , in
@xmath41 , to produce some _ explicit _ examples of deformation spaces of non - orientable surfaces .
we start by recalling some facts of uniformization of riemann surfaces . a * partition
* @xmath42 on a riemann surface @xmath1 , of genus @xmath43 , is a collection of simple closed disjoint curves , such that no curve of @xmath42 is homotopically trivial , and no two curves of @xmath42 are freely homotopically equivalent .
a partition consists of at most @xmath44 curves ; if this bound is attained , we say that the partitions is * maximal*. see @xcite for the proof of the existence of partition on surfaces .
given a + riemann surface @xmath1 and a maximal partition @xmath45 , there exists a kleinian group @xmath10 , known as a terminal regular b - group , such that : + 1 .
there is a unique maximal simply connected set @xmath11 of the riemann sphere , where @xmath10 acts discontinuously , and @xmath14 for all @xmath15 ; + 2 .
@xmath46 ; + 3 . to each curve of @xmath42
corresponds a maximal conjugacy class of cyclic subgroups of @xmath10 generated by a parabolic transformation ; + 4 . besides @xmath1 ,
the group @xmath10 uniformizes the @xmath47 thrice punctured spheres obtained from squeezing each curve of @xmath42 to a puncture ; + 5 .
@xmath10 is unique up to conjugation by mbius transformations . a * symmetry * @xmath30 on a riemann surface is an anticonformal involution .
if @xmath48 denotes the set of fixed points of @xmath30 , then we have that @xmath49 consists of at most two components .
it is a well known fact that @xmath50 is orientable if and only if @xmath49 is not connected ( @xcite , @xcite ) . the classical result about
the structure of @xmath48 is the following .
if @xmath30 is a symmetry on a compact surface @xmath1 of genus @xmath6 , then @xmath48 is either empty or consists of @xmath51 simple disjoint curves @xmath52 , with @xmath53 .
this theorem can be improved as follows .
in addition to the curves ( if any ) @xmath54 , there exists closed curves @xmath55 , such that : + 1 .
@xmath56 is a collection of disjoint curves ; + 2 .
@xmath57 , for all @xmath58 ; + 3 .
@xmath59 consists of two components , @xmath33 and @xmath34 ; + 4 .
@xmath30 interchanges @xmath33 and @xmath34 .
the existence of maximal partitions invariant under symmetries is a well known fact ; but our proof is different from those in the literature ( see , for example ( * ? ? ?
117 - 120 ) ) , but we include it here for the sake of completeness .
[ lemma - part ] let @xmath1 be a riemann surface of genus @xmath60 , and let @xmath30 be a symmetry on @xmath1
. then there exists a maximal partition @xmath42 invariant under @xmath30 , that is , @xmath61 . let @xmath62 denote the set of curves given by the harnack - kra - maskit theorems .
we claim that @xmath62 is a partition on @xmath1 .
in fact , we have that if a curve of @xmath62 is homotopically trivial , then @xmath33 and @xmath34 are discs , and therefore , @xmath1 will be homeomorphic to the riemann sphere . similarly ,
if two curves of @xmath62 are freely homotopic , we get that @xmath1 is a torus .
if @xmath62 is maximal , we are done .
if not , let @xmath63 be a curve such that @xmath64 is a partition .
we claim that @xmath65 is a partition .
this can be seen in three easy steps : + 1 .
@xmath66 is not homotopically trivial , since @xmath30 is a homeomorphism , and @xmath63 is not trivial ( being a curve in a partition ) ; + 2 .
@xmath66 is not ( freely ) homotopically equivalent to any curve of @xmath62 .
if there is a curve @xmath67 in @xmath62 , freely homotopic to @xmath66 , then , applying @xmath30 , we would get that @xmath63 is freely homotopic to @xmath67 , contradicting the fact that @xmath68 is a partition ; + 3 .
@xmath66 is not freely homotopic to @xmath63 .
if these two curves are freely homotopic , then we have that @xmath63 and @xmath66 bound a cylinder in @xmath1 . since these curves lie in different components of @xmath69 , we get that there is a curve , @xmath67 in @xmath62 , in that cylinder .
but this implies that @xmath63 is homotopically equivalent to @xmath67 , which is again not possible . any non - orientable surface @xmath1 has a double unramified cover by a riemann surface @xmath18 , called the * complex double * ( ( * ? ? ?
* pgs 37 - 40 ) ) . if @xmath1 has genus @xmath6 , then @xmath18 has genus @xmath7 .
@xmath1 has a symmetry @xmath30 , such that @xmath31 .
we have now all the necessary tools to prove theorem [ thm - unif ] .
proof of theorem [ thm - unif]let @xmath18 be the complex cover of @xmath1 , and let @xmath30 be the symmetry on @xmath18 such that @xmath31 . by our hypothesis
, @xmath18 has genus greater than @xmath29 , so applying the lemma [ lemma - part ] we obtain a @xmath30-invariant maximal partition @xmath42 on @xmath18 . using the maskit uniformization theorem , we get a kleinian group @xmath10 , uniformizing @xmath18 in the invariant simply connected component @xmath11 .
we only need to show that the symmetry @xmath30 lifts to an anti - mbius transformation , in the covering determined by @xmath10 ( i.e. , it is of the form given in the statement of the theorem ) . for simplicity ,
assume first that @xmath30 is orientation preserving .
then @xmath30 induces a set of conformal mappings , @xmath70 , among the parts @xmath71 of @xmath72 .
the infinite nielsen extension , @xmath73 , of @xmath74 is a thrice punctured sphere , obtained from @xmath74 by completing the holes to punctured discs .
it is a classical fact that @xmath75 extends to a quasiconformal mapping , denoted by @xmath76 , from @xmath73 to @xmath77 , with maximal dilatation @xmath78 ( @xcite , @xcite ) . since @xmath75 is conformal , we have that its dilatation is equal to @xmath29 , and therefore @xmath79 , that is , @xmath76 is also conformal .
let @xmath80 be a component of @xmath81 , where @xmath82 is the natural quotient mapping from @xmath11 onto @xmath1 , and let @xmath83 .
we have that the @xmath84s are triangle groups with two invariant components ; let @xmath85 be the component that does not contain @xmath11 .
then , the mapping @xmath75 induces a conformal mapping between @xmath85 and @xmath86 , for a proper choice of @xmath87 . this can be done with all the components of @xmath81 , and all the @xmath88 , obtaining in this way a conformal self - mapping @xmath89 , of @xmath90 .
but this set is the region of discontinuity of @xmath10 ( that is , the set of points of the riemann sphere were @xmath10 acts discontinuously ) . since @xmath10 is finitely generated , we have that the complement of the region of discontinuity has measure zero .
therefore , the classical theory of quasiconformal mappings gives us a conformal automorphism of the riemann sphere that extends @xmath89 .
such mapping should be a mbius transformation . to complete the proof of the theorem it suffices to observe the following two facts , which are easy to prove : + 1 .
the theory of quasiconformal mappings has a natural extension to cover the orientation reversing mappings @xcite ; and + 2 .
bers results on nielsen extensions can be applied to orientation reversing mappings .
we define a * ramification point * @xmath91 on a klein surface as a point such that the universal covering looks like @xmath92 , in a neighborhood of @xmath91 , ( which corresponds to the points @xmath93 ) for some finite positive integer @xmath94 .
the number @xmath94 is called the * ramification value * of @xmath91 .
ramification points correspond to fixed points of orientation preserving transformations , of finite order .
a * puncture * is a domain @xmath95 in @xmath1 satisfying the following conditions : + 1 .
@xmath95 is homeomorphic to @xmath96 ; + 2 . for any sequence of points in @xmath97 converging to the origin ,
the corresponding sequence in the surface diverges ; + 3 .
if there are two patches on @xmath1 , whose images contain some sets of the form @xmath98 ( that is , neighborhoods of the missing point ) , then the change of coordinates is holomorpic . given a klein surface with ramification points and/or punctures , called a * klein orbifold * , we define its * signature * as a collection of numbers ( and a symbol ) of the form @xmath99 , where @xmath6 is the genus of the surface , @xmath94 is the number of special points , and @xmath100 are the ramification values , with puntures having ramification value equal to @xmath101 . if the orbifold is orientable , then we take the symbol @xmath102 , while @xmath103 is used for non - orientable surfaces .
if all the ramification values are equal to @xmath101 , then we will write the signature as @xmath104 .
it is not difficult to see that if @xmath1 has signature @xmath105 , then the signature of @xmath18 must be @xmath106 .
a klein orbifold @xmath107 is hyperbolic if and only if @xmath108 is positive , where @xmath109 if @xmath1 is not orientable , and @xmath110 in the orientable case .
since the maskit uniformization theorem and the theory of quasiconformal mappings extend to the case of surfaces with ramification points , we have that theorem [ thm - unif ] can be applied also to hyperbolic klein orbifolds .
a natural problem in deformation theory is to study which properties of a surface are determined by its teichmller space , and vice versa .
more precisely , in this section we will see that , if the teichmller spaces of two surfaces are equivalent , then the surfaces are homeomorphic
. reciprocally , if two klein orbifolds have the same genus and number of ramification points , we will prove that their deformation spaces are isomorphic .
we start by recalling the definition of the modular group , and some basic facts about hyperelliptic surfaces .
the * modular group * @xmath111 of a non - orientable surface is the quotient of the group of diffeomorphisms , by those homotopic to the identity ( in the case of riemann surfaces , one takes only the orientation preserving diffeomorphisms ) .
we have that @xmath111 acts on @xmath25 by pullback : given a mapping @xmath112 , and a real analytic structure structure @xmath2 , we define @xmath113 as the unique structure on @xmath1 that makes @xmath114 dianalytic .
the mapping @xmath115 \rightarrow [ f^*(x)]$ ] becomes dianalytic in the natural structure of @xmath25 .
we say that a non - orientable surface @xmath1 is * hyperelliptic * if it is a double cover of the ( real ) projective plane ( respectively , the riemann sphere , in case of orientable surfaces ) .
hyperelliptic surfaces carry the so - called * hyperelliptic involution * , which is a dianalytic involution ( holomorphic , in the case of orientable surfaces ) @xmath116 , such that @xmath117 is the projective plane .
it is not hard to see that if @xmath1 is hyperelliptic , so is its complex cover @xmath18 , and that @xmath116 lifts to the hyperelliptic involution @xmath58 of @xmath18 . since the hyperelliptic involution on a riemann surface is unique , we obtain the reciprocal result : if @xmath18 is the complex double of a surface @xmath1 , and @xmath18 is hyperelliptic , then @xmath1 is also hyperelliptic ; moreover , the involution @xmath58 can be pushed down to the hyperelliptic involution @xmath116 in @xmath1 . as in the case of riemann surfaces , we have that the modular group acts effectively on teichmller space , except for a finite number of cases .
@xmath111 , for a hyperbolic non - orientable surface @xmath1 , compact with finitely ( maybe zero ) punctures , acts effectively on @xmath118 , except for the following cases : the projective plane with two puntures , the klein bottle with one puncture , or the connected sum of three projective planes . as usual , let @xmath18 denote the complex double of @xmath1 , and let @xmath30 be the involution associated to such covering .
it is a well known fact ( ( * ? ? ?
149 ) or below ) that @xmath25 can be identified with the set of fixed points @xmath119 of @xmath32 in @xmath27 . if @xmath112 is a diffeomorphism of @xmath1 , there is a unique holomorphic lift , @xmath120 ( see ( * ? ? ?
20 ) and ( * ? ? ?
39 ) ) . by uniqueness
, we have that @xmath121 satisfies @xmath122 .
this proves that @xmath111 embeddes into the set @xmath123 .
the modular group of a hyperbolic riemann surface acts properly on the corresponding teichmller space , unless the signature of the surface is @xmath124 , @xmath125 , @xmath126 or @xmath127 .
since complex covers have an even number of punctures , we get that only the first three signatures can give non - orientable surfaces .
we therefore obtain that , the cases where @xmath128 may fail to act effectively correspond to the signatures @xmath129 , @xmath130 and @xmath131 .
this proves the first part of the proposition .
if , in the other hand , @xmath1 has signature in the above list , we have that the only elements of @xmath111 that do not act properly are the classes of the identity and the hyperelliptic involution . by the remarks before the theorem , we also have that these classes are the only elements that act trivially on the teichmller space of hyperbolic klein surfaces .
it would be interesting to know whether the image of @xmath111 is equal to the whole set @xmath132 described in the above proof .
the proof of the following result is straightforward from the parallel result for riemann surfaces . nevertheless , the proposition is interesting , because it shows the great similarity between the theory of deformation of orientable surfaces , and that of non - orientable ones .
[ patterson theorem for non - orientable surfaces ] let @xmath37 , @xmath38 , be two hyperbolic compact , with finitely many ( possibly zero ) punctures , non - orientable surfaces , and suppose that either @xmath33 or @xmath34 has genus not equal to @xmath24 .
if @xmath39 is real isomorphic to @xmath40 , then @xmath33 is homeomorphic to @xmath34 .
it suffices to observe that if @xmath39 and @xmath40 are real isomorphic , then the spaces @xmath133 and @xmath134 are biholomorphic , and therefore , @xmath135 and @xmath136 are homeomorphic.[thm - pat ] in order to prove theorem [ thm - bers ] , we need to review a basic concept of teichmller theory : quadratic differentials and beltrami coefficients .
let @xmath137 denote the space of * bounded quadratic differentials * on a klein surface .
these are simply quadratic differentials , regular on the surface , with at most simple poles at the punctures , and with zeros of certain order ( determined by the ramification value ) at the ramification points ( see , for example @xcite ) .
it is easy to see that the set @xmath137 , for a non - orientable surface , can be identified with the subspace of elements of @xmath138 , that are preserved by @xmath30 , that is , @xmath139 . here by @xmath140 we mean @xmath141 .
the dimension of @xmath137 over @xmath142 is equal to the dimension of @xmath138 over @xmath143 .
the embedding from @xmath137 into @xmath138 is an isometry in the norm , @xmath144 where @xmath145 is the metric obtained by pushing down the poincar metric of the upper half plane onto the corresponding surface ( and the supremum is taking over the whole surface ) .
the cotangent bundle of @xmath25 can be naturally identified with the space @xmath137 .
let @xmath1 be a klein surface , uniformized by the nec group @xmath5 .
let @xmath146 denote the space of * beltrami differentials * for @xmath5 .
this set consists of ( classes of ) measurable functions @xmath147 , with support in the upper half plane , and @xmath148-norm less than one , satisfying @xmath149 , if @xmath150 is orientation preserving , or @xmath151 , if @xmath150 reverses the orientation . for each @xmath152
, there is a unique quasiconformal homeomorphism @xmath153 , of the upper half plane , with dilatation @xmath147 , that fixes @xmath154 , @xmath29 and @xmath101 .
two beltrami coefficients , @xmath147 and @xmath155 , are * equivalent * if @xmath156 on the real line .
the space of beltrami differentials , quotiented by the above equivalence relation is the teichmller space @xmath157 of the group @xmath5 .
it can be proven that @xmath157 is naturally isomorphic to @xmath25 , where @xmath158 .
it is easy to see that , if @xmath10 is the subgroup of @xmath5 consisting of the orientation preserving mappings , then @xmath159 .
we can identify the beltrami differentials for @xmath5 with those beltrami differentials for @xmath10 , that are invariant under @xmath30 , that is , @xmath160 .
this allows us to identify the deformation space @xmath25 with the set of fixed points of @xmath32 , @xmath119 , in the deformation space of the complex double @xmath27 . in this way
we obtain a teichmller s lemma for non - orientable surfaces : on each equivalence class of beltrami differentials there is a unique mapping with minimal dilatation , which is of the form @xmath161 of @xmath29 . first proof of theorem [ thm - bers ] let @xmath1 be a hyperbolic surface of signature @xmath162 @xmath163 , where we assume that at least one of the ramification values is finite .
let @xmath164 be the surface of signature @xmath165 , obtained by removing from @xmath1 all the points with finite ramification value .
let @xmath5 and @xmath166 be nec groups uniformizing @xmath1 and @xmath164 respectively .
define @xmath167 as @xmath168 fixed points of elliptic elements of @xmath169 .
then , by our hypothesis we have that @xmath170 . since @xmath171 , we have a covering map @xmath172 , that makes diagram @xmath29 commutative . the function @xmath173 induces a group homomorphism @xmath174 , defined by the rule @xmath175 .
the mapping @xmath173 induces a mapping , @xmath176 , between beltrami coefficients , given by the expression @xmath177 ( see below for a proof of the fact that @xmath173 is a holomorphic function ) .
it is not hard to see , using the same arguments that in the orientable case , that @xmath178 induces an real analytic bijection between the spaces @xmath179 and @xmath25 .
see , for example , @xcite ( or @xcite , for a more details ) . by analytic continuation
, we can extend @xmath178 to a biholomorphic function between @xmath180 and @xmath181 , that obviously commutes with the involutions @xmath182 and @xmath30 , giving the desired isomorphism .
second proof of theorem [ thm - bers]in this case , we will use the bers - greenberg theorem for riemann surfaces . consider the same setting as in the first proof .
let @xmath10 and @xmath183 be the orientation preserving subgroups of @xmath5 and @xmath166 respectively .
then we have @xmath184 and @xmath185 .
we get that diagram @xmath9 is commutative .
the functions @xmath186 , @xmath187 and @xmath188 are the natural projections .
the mapping @xmath173 is defined as in the first proof .
the function @xmath112 is the unique holomorphic mapping that makes the lower triangle commutative .
we have that @xmath186 is holomorphic ( a covering of a riemann surface by an open set of the complex plane ) , so the function @xmath173 is holomorphic .
this implies that the group homomorphism @xmath189 , of the first proof , takes the subgroup @xmath183 onto @xmath10 . since the surfaces @xmath1 and
@xmath18 have the same genus and number of ramification points / punctures , we have that the spaces @xmath190 and @xmath27 are isomorphic , via the function @xmath178 induced by @xmath173 , as in the first proof . to prove theorem [ thm - bers ] it suffices to show that @xmath173 commutes with the antiholomorphic involutions @xmath30 and @xmath182 ( that produce the surfaces @xmath1 and @xmath164 , respectively ) . in other words , we have to show @xmath191 .
but we have that @xmath192 , so @xmath193 is equal to either @xmath194 or @xmath195 .
since this last function is holomorphic , we must have @xmath196 , as claimed . identifying @xmath179 and @xmath25 with the set of fixed points of @xmath197 and @xmath32 in @xmath198 and @xmath27 , respectively , we get the bers - greenberg theorem for non - orientable surfaces .
in this section , we will shown with two examples , how the techniques of kra of @xcite can be applied to the case of non orientable surfaces .
we will work with deformation spaces of kleinian groups , which are equivalent ( if the groups are chosen properly , for example , groups given by theorem @xmath29 ) to deformation spaces of riemann or klein surfaces ( see @xcite or @xcite for more details ) . in out first example , we consider a klein surface , @xmath1 , of signature @xmath199 .
its complex double , @xmath200 , has signature @xmath201 .
a kleinian group , @xmath202 , uniformizing @xmath18 is generated by the transformations @xmath203,~
b=\left[\begin{array}{cc}-1&0\\2&-1\end{array}\right],~ b_\alpha= \left [ \begin{array}{cc } -1 + 2\alpha&-2\alpha^2\\2 & -1 - 2\alpha\end{array}\right],\ ] ] where @xmath204 ( see above reference ) .
the coordinate of @xmath202 in the teichmller space @xmath205 ( notation should be obvious ) is given by the expression @xmath206 here @xmath207 denotes the cross ratio of four points in the riemann sphere , chosen so that @xmath208 , and @xmath209 denotes the unique fixed point of the parabolic transformation @xmath210 .
a maximal partition in @xmath18 consists of a simple closed curve , say @xmath211 .
we can assume that the punctures @xmath212 and @xmath213 lie on the same component of @xmath214 .
the mbius transformation @xmath132 corresponds to the partition curve .
let @xmath215 , @xmath216 , be a small simple loop around the puncture @xmath217 , oriented such that the puncture lies to the left of @xmath215 .
the parabolic elements @xmath218 , @xmath219 , @xmath220 and @xmath221 , correspond to these four loops . without loos of generality
, we can assume that @xmath222 corresponds to @xmath215 .
consider on @xmath200 the involution @xmath223 , where @xmath224 is a rotation by @xmath225 degrees on the axis of figure @xmath24 , and @xmath13 is an anticonformal reflection on @xmath211 .
the anticonformal mapping @xmath30 has not fixed points , and the quotient @xmath226 has signature @xmath227 . in order to identify @xmath25 in @xmath27
, we have to study the action of the mappings @xmath224 and @xmath13 on the group @xmath202 .
the transformation @xmath224 intechanges the punctures that lie on the same component of @xmath228 , that is , @xmath224 sends @xmath229 to @xmath230 and @xmath231 to @xmath232 ( up to free homotopy ) .
the function @xmath13 , not only intechanges the two components of @xmath228 , but also changes the orientation of the loops , sending @xmath229 to @xmath233 and @xmath230 to @xmath234 .
we have that @xmath224 lifts to @xmath235 , while @xmath13 lifts to @xmath236 , @xmath237 , in the covering determined by the group @xmath202 .
observe that , although a mbius transformation may have may square roots , parabolic elements have only one , and therefore , @xmath235 is well defined .
from these obervations we can compute the action of @xmath32 on @xmath238 as follows .
first of all , observe that the group @xmath202 is generated by @xmath132 , @xmath239 and @xmath240 , with the property that @xmath241 and @xmath242 are parabolic elements .
we will use the notation @xmath243 to emphasis this fact .
observe also that @xmath244 .
the mapping @xmath224 sends @xmath243 to @xmath245 . since the transformation @xmath13 is orientation reversing , we have that its action is given by conjugating the group @xmath246 into @xmath247 + @xmath248 .
the mapping @xmath32 has therefore the form @xmath249 therefore , the teichmller space @xmath25 can be identified with the set of points @xmath250 such that @xmath251 . by the work of kra
, we have that @xmath25 is precisely the set @xmath252 .
consider now the case of a surface @xmath1 of signature @xmath253 .
the complex double of @xmath1 is a riemann surface @xmath18 , of genus @xmath9 without punctures . in @xcite
we can find a group @xmath254 uniformizing @xmath18 , generated by the mbius transformations : @xmath255,~ a_2=\left[\begin{array}{cc } 1&-2\\2&-3\end{array}\right],~ c_1=i\left[\begin{array}{cc } \tau_1&1\\1&0\end{array}\right],\ ] ] @xmath256,\ ] ] @xmath257.\ ] ] the mapping @xmath258 correspond to the curves @xmath259 of the partition of @xmath18 of figure @xmath41 .
the @xmath260 are loxodromic elements with the property that @xmath261 and @xmath262 are parabolic elements .
the coordinates on @xmath263 are given by the cross ratios @xmath264 let @xmath265 be defined in a similar way as in the previous example : @xmath224 is a rotation by @xmath225 degrees on the line of figure @xmath41 , and @xmath13 is an antiholomorphic reflection on the curve @xmath266 .
the computation of the case @xmath199 applies to the coordinate @xmath267 , since the part corresponding to it ( that is , @xmath268 ) is a surface of signature @xmath201 .
so we get that the action of @xmath32 on @xmath267 is @xmath269 .
we have that the mapping @xmath224 lifts to @xmath270 , and the lift of @xmath13 is @xmath271 computing as in the previous example , and taking care of the fact that @xmath13 reverses orientation , we see that @xmath272 is given by the cross ratio of the points @xmath273 where @xmath274 .
this cross ratio gives @xmath275 .
similarly , one gets @xmath276 .
therefore , the teichmller space @xmath25 can be identified with the set of points @xmath277 such that @xmath278 which proves theorem @xmath29 of the introduction .
the above computations give us some other isomorphisms , different from those of the previous section .
observe that the transformation @xmath224 is just the hyperelliptic involution on @xmath18 .
it is not hard to see that @xmath279 acts like the identity in @xmath27 ( ( * ? ? ?
the mapping @xmath13 has a curve of the partition as the set of fixed points .
we have that @xmath280 is a sphere with one hole and two punctures , in the first example , or a torus with a hole in the second example .
let us denote this surfaces by @xmath281 and @xmath282 , respectively . | in this paper , we prove some isomorphisms theorems between teichmller spaces of non - orientable compact surfaces .
we also develop a technique , based on similar results for riemann surfaces , to give explicit examples of teichmller spaces of klein surfaces .
[ section ] [ thm]proposition [ thm]corollary [ thm]definition [ thm]lemma psfig |
the low mass x - ray binary ( lmxb ) 4u1636 - 536 is a well - known type - i x - ray burster , which has been extensively observed over the past four decades and is considered well - established as a persistent x - ray source ( see references in * ? ? ?
its optical counterpart ( v801 ara ) revealed its 3.8h orbital period @xcite through photometric variations , but the period has also been detected spectroscopically @xcite .
however , during 2001 , the _ all sky monitor _ onboard the _ rossi x - ray timing explorer _
( rxte / asm ) recorded an obvious , gradual decline of its x - ray flux level . within a year
, it began to exhibit a substantial quasi - periodic , long - term variability of @xmath146d @xcite .
more remarkably , shih et al . also reported that this variation was present at both soft ( 2 - 12 kev , rxte / asm ) and hard ( 20 - 100 kev , integral / ibis ) x - ray energies , but that these were _ anti - correlated_. such a pronounced energy - dependent phenomenon led to the suggestion of an accretion disc instability as the most likely explanation , as the result of a decrease in the mass transfer rate .
+ @xcite provided further evidence regarding the nature of the long - term variability of 4u1636 - 536 based on a @xmath11.5 yr rxte / pca monitoring campaign from march 2005 to september 2006 .
the lightcurves display long - term variability of 30 - 40d which is consistent with the temporal and energy changes that occur between two states of atoll sources : the _ island _ ( hard ) and _ banana _ ( soft ) states @xcite . from recently published swift / bat hard x - ray ( 20 - 50 kev ) light curves that began in 2005 , @xcite have confirmed the anti - correlated relationship between the soft and hard x - rays in 4u 1636 - 536 , providing further support for the suggestion of a state change / accretion instability within the source . + according to the standard model for lmxbs , their optical emission is dominated by x - ray reprocessing in the accretion disc around the compact object @xcite .
thus , the varying x - ray flux of 4u1636 - 536 should also influence the apparent brightness of its optical counterpart . with this in mind , we began a @xmath1100d optical daily monitoring program which was expected to cover at least 2 cycles of the previously observed long term x - ray variability in 4u1636 - 536 .
+ in this paper , we report the results of our optical observations , and compare them with the simultaneous x - ray data available from the rxte / asm and swift / bat archives .
these x - ray monitoring data also allowed us to investigate the long - term variability properties of 4u 1636 - 536 as a function of x - ray energy .
an x - ray spectroscopic analysis of rxte / pca and hexte data was carried out to investigate the evolution of its spectral properties as a function of the long - term variability .
we also discuss possible physical mechanisms that might drive such phenomena .
we performed a 3 months optical monitoring of 4u1636 - 536 using the noao smarts 1.3 m telescope with the andicam on cerro tololo , chile between 01 april and 30 june 2008 .
the ccd for the andicam is a fairchild 447 2048x2048 ccd with 15-micron pixels .
each night , a pair of exposures of 2 minutes each through both b and v filters was taken . in total
, we obtained 67 nights of good data during the campaign .
data reductions , such as bias subtraction and flat - field correction , were provided by the yale smarts team .
+ we utilised the aperture photometry package iraf / daophot to extract photometric information from the data .
a number of bright comparison stars ( listed in the nomad catalog ) on the same ccd frame were selected to reconstruct the variability of the source , but absolute calibration of the field was not performed in this work . from the scatter of the differential magnitudes of companion stars ,
the accuracy of the differential magnitude measurement of 4u 1636 - 536 is typically @xmath10.006 mag .
+ since the 1960s 4u1636 - 536 has been frequently observed by many x - ray missions . the all - sky monitor
( asm , 2 - 12 kev ) onboard the long - lived rossi x - ray timing explorer ( rxte ) has been providing the longest continuous coverage of all luminous x - ray sources , stretching from 1996 to the present day ( see * ? ? ?
additionally , the hard x - ray transient monitor program , using the burst alert telescope ( bat , @xmath115 - 50 kev ) onboard swift , has been operating since 2005 @xcite .
we retrieved both datasets up to 09 february 2009 from the heasarc archive and the website of the swift / bat hard x - ray transient monitor , respectively .
figure [ multiple lightcurves ] displays the three months of smarts monitoring in the optical b and v bands , together with the simultaneous soft ( rxte / asm ) and hard x - ray ( swift / bat ) data .
the first feature to note is the presence of a nearly regular variation in all wavelengths , which is even more obvious in the x - ray data . in figure [ ls periodogram ]
the lomb - scargle periodogram reveals a significant , consistent period of @xmath130 d at all wavelengths except the optical v band .
we notice that the peak profile is rather wide , so we simulated our data with a 30d sine wave ( plus poisson noise ) using a similar count rate as observed for 4u 1636 - 536 .
this simulation shows a peak with similar width to that observed , suggesting that it is a result of the short length of the simultaneous dataset ( only covering a few cycles ) .
indeed , the peak width becomes narrower as we increase the duration of data included ( rxte / asm and swift / bat only ) .
for example , within one year surrounding the campaign , the source clearly exhibited a significant long - term variability of @xmath1 30d in both energy bands .
+ confidence level ( see text).,title="fig:",width=340 ] confidence level ( see text).,title="fig:",width=340 ] the period present in the 2008 campaign was clearly shorter than that exhibited in 2004 ( @xmath146 d , * ? ? ? * ) , and was not always consistent with other reports , i.e. 30 - 40d @xcite and @xmath146d @xcite . until recently , such variability studies have been rather patchy , and the correlation between soft and hard x - rays was not fully established .
however , with the availability of both rxte / asm and swift / bat monitoring over extended periods of time , a time - dependent power spectral analysis , similar to that reported by @xcite on the superorbital modulation in smc x-1 , was performed in order to investigate the long - term variability evolution in 4u 1636 - 536 . as the timescale of interest is @xmath150d , a sliding data window of 200d
was used . in order to provide some smoothing of the resulting dynamical power spectrum , each subsequent window moved forward by 50d .
the resulting power spectrograms from both datasets are shown in figure [ swift and rxte power spectra ] .
+ 46d modulation reported by @xcite is marked by the dashed vertical line.,width=302 ] our analysis suggests that even the dominant periodicity ( the darkest region ) is transient in nature .
the most recent ( 2009 ) data show that the variation has evolved to a shorter timescale ( @xmath130d ) , but our analysis did not reveal any ultra long - term periodic trend , as was seen in e.g. smc x-1 @xcite .
near the three months of smarts observations ( hjd @xmath154600 ) , it appears that the @xmath130 d period was only significant in swift / bat data .
however , as we have shown in figure [ ls periodogram ] , the @xmath130 day period is also significant in rxte / asm data .
+ to investigate the correlation between different wavebands , we performed a cross - correlation analysis with the three datasets : ( i ) optical and soft x - rays , ( ii ) optical to hard x - rays , ( iii ) soft and hard x - rays . since the variability is more significant in optical b band , that was used in the analysis ( after conversion from magnitude to intensity ) .
+ our analysis adopted the interpolation method of calculating the cross - correlation function ( ccf , * ? ? ?
the light curves were shifted backwards and forwards up to 30 days against each other in steps of 1 day .
the results of this analysis are shown in figure [ cross correlation ] .
+ the ccf analysis reveals a clear anti - correlation between the soft and hard x - rays . on the other hand ,
the optical emission is positively correlated with the soft x - rays .
however , we also note that the optical emission seems to lag the soft x - rays by @xmath1 1 - 2 days , which is much too long to be the result of x - ray reprocessing / irradiation ( time - scale of @xmath1 minutes ) .
the optical emission , like the soft x - rays , is anti - correlated with hard x - rays , and the lag is slightly longer .
however , we caution that the lag appearing in the ccf may not be real , but is instead an artefact related to the gaps in the optical data . a more regular and longer term optical campaign will be needed to confirm this .
+ in an attempt to understand how the x - ray spectral properties of 4u 1636 - 536 evolve through the different phases of the long - term variability , a broad - band spectroscopic analysis was performed .
we used archival rxte / pca and hexte data because of their wide spectral range between 2 and 250 kev @xcite .
two observations , which represent two distinct epochs in the rxte / asm light - curve ( at maximum and minimum ) and sufficiently long to provide good signal - to - noise ratio , were selected ( see figure [ rxte_total ] ) : * obsid 70036 - 01 - 02 - 00 ( mjd 54270 ) , at the peak of an oscillation * obsid 70034 - 01 - 01 - 000 ( mjd 53612 ) , at the minimum of an oscillation the spectra were derived from pca standard-2 and hexte archive mode following standard procedures : using only pcu 2 ( the only pcu that operates in all our datasets ) in pca and cluster b ( which provides the necessary background information ) in hexte and the latest calibration data @xcite .
the combined pca and hexte data were then analysed with xspec v.12.x . throughout the analysis , the galactic column density ( @xmath2 ) toward the source
was fixed at 3@xmath310@xmath4 @xmath5 @xcite .
we note that this value is well within the range determined by the rosat position sensitive proportional counter observations ( pspc , * ? ? ?
* ) . to estimate the luminosity and radii ( see context below ) ,
the distance of 4u 1636 - 535 is referred to @xmath1 5.9 kpc @xcite .
+ .spectral model fits to two distinct states of 4u1636 - 536 with the rxte pca / hexte data .
the x - ray luminosity is determined from the unabsorbed flux . [
cols="<,^,^ " , ] figure [ spectral fitting result ] shows the data and model fits , with table [ spectral data ] summarising the best fit model details .
the spectral properties of 70036 - 01 - 02 - 00 ( rxte / asm high epoch ) can be simply described by a model consisting of a multi - temperature accretion disc ( diskbb ) component plus a thermal comptonisation ( comptt , a spherical geometry was assumed ) component .
+ on the other hand , the spectral properties of 70034 - 01 - 01 - 000 ( rxte / asm low epoch ) is more complicated .
the same two - component model fit to the data is poor ( @xmath6 2 ) .
instead , it is more compatible with a single - temperature blackbody component ( bbody ) plus a thermal comptonised one ( using the same input soft photon temperature t@xmath7 as in the high - state epoch ) .
the blackbody component likely represents the emission from the surface of the neutron star . additionally , a power - law component for the hard x - rays @xmath8 50 kev and a lorentz profile at @xmath1 6.3 kev ( which may represent a relativistically broadened fe k@xmath9 emission line ) are required to improve the fitting result .
we note that these additional components are quite similar to those observed in the hard state of black hole x - ray binaries , such as gro j1655 - 40 ( see * ? ? ?
from the 3 years of simultaneous rxte / asm and swift / bat archival data on 4u1636 - 536 , it is clear that the variability , on both short and long time - scales , is indeed strongly energy - dependent .
this was first noted in simultaneous rxte / asm and integral / ibis data @xcite , and has been a persistent feature of this source since 2002 , when rxte / asm observed the beginning of a dramatic decline in flux .
additionally , we find that the long - term variability of 4u1636 - 536 has been constantly evolving over the last decade , although not in a periodic manner . + a tilted / warped , precessing accretion disc was suggested as an explanation for such long term variations @xcite , and the system parameters ( in particular the low mass - ratio ) of 4u 1636 - 536 do allow for a precessing accretion disc @xcite
. however , the presence of an evolving trend in the long - term period , the anti - correlation between soft and hard x - rays , and the distinct x - ray spectral properties at different epochs are all difficult to understand in terms of a precessing disc .
+ the x - ray spectroscopic analysis clearly reveals two different spectral states in the high and low epochs .
it suggests that , in the high epoch , 4u1636 - 536 is like a typical lmxb : the accretion disc is fully developed and extends into the inner boundary region ( r@xmath10 25 km , assuming face - on orientation , @xmath11=0@xmath12 ) , which is likely surrounded by a comptonised region , or corona @xcite .
the energy state is softer as the spectrum is dominated by soft x - rays ( 2 - 10 kev ) .
+ in the low epoch , on the other hand , no reliable multi - temperature disc component could be derived suggesting that the accretion disc probably retreated away from the inner region , or was _ truncated_. instead , we have determined a soft x - ray emitting region of radius @xmath1 5 km , which most likely corresponds to the neutron stellar surface .
we also note that the spectral data in the low epoch contains a high energy component ( @xmath850 kev ) which is very similar to that in a `` peculiar '' integral observation of 4u1636 - 536 @xcite , as well as to those of black hole binaries in hard x - ray states .
clearly , 4u 1636 - 536 became significantly harder during the low epoch , as the hard x - ray luminosity ( l@xmath13 erg s@xmath14 ) composed roughly half of the bolometric x - ray flux ( 2.0 - 80 kev , see table [ spectral data ] ) .
+ the observed x - ray variability phenomena from 4u1636 - 536 resemble the main x - ray spectral states of black hole x - ray binaries , namely the `` hard '' and `` thermal '' states @xcite .
currently , various flavours of accretion flow models have been developed , mainly for black hole x - ray binaries , to not only explain these main spectral states , but also the more subtle states observed ( e.g. see the review by * ? ? ?
the most commonly used explanation assumes that the optically thick inner disc is replaced by a hot , optically thin , flow when black hole x - ray binaries change to the `` hard '' state .
although a similar scenario is used to explain the state transition in atoll sources , there is a fundamental difference with their black hole cousins : the presence of an observable solid surface at the neutron star . with its frequent state changes and abundant archival data from rxte , swift and integral
, 4u1636 - 536 could be an ideal target to study the behaviour of the accretion flow in neutron star binaries , and thereby adapt the models that are currently used for the bhxrbs .
+ our three - month optical monitoring campaign on 4u1636 - 536 reveals a clear modulation which was obviously correlated with the x - ray states .
not surprisingly , this is consistent with the currently accepted mechanism for lmxb optical counterparts that they are dominated by x - ray reprocessing on the accretion disc and heated face of the donor star by the central x - ray source ( e.g. * ? ? ?
* ; * ? ? ?
thus the observed optical component largely follows variations in the intrinsic x - ray flux .
+ our campaign also reveals that the increasing hard x - rays ( @xmath8 10 kev ) does not seem to influence the optical modulation .
this is in contrast to lmc x-2 where its optical / x - ray correlation was much stronger with the hard x - rays than the soft @xcite , suggesting that different accretion mechanisms are working in the inner regions of these systems .
+ one possible scenario that could explain our observation of 4u 1636 - 536 is compton reflection ( see * ? ? ?
* ; * ? ? ?
* ) . in this case ,
the softer part of x - ray continuum ( power - law component with @xmath15 2.12 ) are photoelectrically absorbed by material in the disc / stellar surface , while a large fraction of the hard x - rays are scattered out of the system with little change in photon energy .
since this would not explain the observations of lmc x-2 , we think that further research into this topic is warranted , especially given the improvement in hard x - ray spectroscopic sensitivity over the last decade .
we have demonstrated that 4u1636 - 536 exhibits a coherent long - term variation across a wide range of wavelengths .
the soft and hard x - rays are clearly anti - correlated on the @xmath130 - 50d variability timescale , and this appears to be a persistent phenomenon .
the evolution of the broad - band spectral characteristics of 4u 1636 - 536 resembles the _ hard _ and _ soft / thermal _ state change usually seen in black hole x - ray binaries . +
our work suggests that 4u 1636 - 536 is a unique and valuable target to understand the physics of x - rays binaries , including the size and location of the hard component in neutron star lmxbs , the interaction of hard x - rays with their surrounding environments such as the accretion disc / companion star surface , and the similarity or dissimilarity of accretion disc evolution between neutron star and black hole x - ray binaries .
we are planning to explore these topics through a comprehensive optical spectroscopic and broad - band x - ray spectral analysis covering different variation epochs in the future .
we acknowledge the quick - look results provided by the asm / rxte team , and the swift / bat transient monitor results provided by the swift / bat team .
ics would like to thank the department of physics and astronomy , michigan state university for providing the support and accessing the 1.3 m telescope managed by smarts consortium .
rc acknowledges a ramon y cajal fellowship ( ryc-2007 - 01046 ) and a marie curie european reintegration grant ( perg04-ga-2008 - 239142 ) .
we would also like to thank for the anonymous reviewer s valuable comments .
+ barthelmy , s. d. , et al .
2005 , space science reviews , 120 , 143 belloni , t. , homan , j. , motta , s. , ratti , e. , & mndez , m. 2007 , mnras , 379 , 247 casares , j. , cornelisse , r. , steeghs , d. , charles , p. a. , hynes , r. i. , obrien , k. & strohmayer , t. e. 2006 , mnras , 373 , 1235 charles , p. , clarkson , w. , cornelisse , r. & shih , i .- c . 2008 , new astron.rev .
51 , 768 clarkson , w. i. , charles , p. a. , coe , m. j. , laycock , s. , tout , m. d. , & wilson , c. a. 2003 , mnras , 339 , 447 cornelisse , r. , et al .
2003 , a&a , 405 , 1033 dickey , j. m. , & lockman , f. j. 1990 , ara&a , 28 , 215 done , c. , gierliski , m. , & kubota , a. 2007 , aapr , 15 , 1 done , c. , & gierliski , m. 2003 , mnras , 342 , 1041 farrell , s. a. , barret , d. , & skinner , g. k. 2009 , mnras , 393 , 139 fiocchi , m. , bazzano , a. , ubertini , p. , & jean , p. 2006 ,
apj , 651 , 416 hasinger , g. & van der klis , m. 1989 , a&a , 225 , 79 jahoda , k. , markwardt , c. b. , radeva , y. , rots , a. h. , stark , m. j. , swank , j. h. , strohmayer , t. e. , & zhang , w. 2006 , apjs , 163 , 401 & yamaoka , k. 1998 , pasj , 50 , 667 levine , a.m. , bradt , h. , cui , w. , jernigan , j.g . ,
morgan , e.h . , remillard r. , shirey , r.e . & smith , d.a .
1996 , apj , 469 , 33 liu , q. z. , van paradijs , j. , & van den heuvel , e. p. j. 2007 , aap , 469 , 807 magdziarz , p. , & zdziarski , a. a. 1995 , mnras , 273 , 837 mcgowan , k. e. , charles , p. a. , odonoghue , d. , & smale , a. p. 2003 , mnras , 345 , 1039 mitsuda , k. , inoue , h. , nakamura , n. , & tanaka , y. 1989 , pasj , 41 , 97 reynolds , c. s. , & nowak , m. a. 2003 , physics reports , 377 , 389 remillard , r. a. , & mcclintock , j. e. 2006 , ara&a , 44 , 49 rothschild , r. e. , et al .
1998 , apj , 496 , 538 schulz , n. s. 1999 , apj , 511 , 304 shahbaz , t. , & kuulkers , e. 1998 , mnraas , 295 , l1 shaposhnikov , n. , jahoda , k. , & markwardt , c. b. , 2009 , http://www.universe.nasa.gov/xrays/programs/rxte/pca/ doc / rmf / pcarmf-11.7/ shih , i. c. , bird , a. j. , charles , p. a. , cornelisse , r. , & tiramani , d. 2005 , mnras , 361 , 602 smale , a.p . &
mukai , k. , 1988 , mnras , 231 , 663 van paradijs , j. , & mcclintock , j. e. 1994 , aap , 290 , 133 white , r. j. , & peterson , b. m. 1994 , pasp , 106 , 879 | we have conducted a 3-month program of simultaneous optical , soft and hard x - ray monitoring of the lmxb 4u1636 - 536/v801 ara using the smarts 1.3 m telescope and archival rxte / asm and swift / xrt data .
4u1636 - 536 has been exhibiting a large amplitude , quasi - periodic variability since 2002 when its x - ray flux dramatically declined by roughly an order of magnitude .
we confirmed that the anti - correlation between soft ( 2 - 12 kev ) and hard ( @xmath0 20 kev ) x - rays , first investigated by @xcite , is not an isolated event but a fundamental characteristic of this source s variability properties .
however , the variability itself is neither strictly stable nor changing on an even longer characteristic timescale .
we also demonstrate that the optical counterpart varies on the same timescale , and is correlated with the soft , and _ not _ the hard , x - rays .
this clearly shows that x - ray reprocessing in lmxb discs is mainly driven by soft x - rays .
the x - ray spectra in different epochs of the variability revealed a change of spectral characteristics which resemble the state change of black hole x - ray binaries .
all the evidence suggests that 4u1636 - 536 is frequently ( @xmath1monthly ) undergoing x - ray state transitions , a characteristic feature of x - ray novae with their wide range of luminosities associated with outburst events . in its current behavioural mode , this makes 4u1636 - 536 an ideal target for investigating the details of state changes in luminous x - ray binaries .
[ firstpage ] accretion , accretion disc binaries : close stars : neutron x - rays : binaries |
the expansion of the universe is now being studied by several methods , including observations of type ia sne ( riess et al .
1998 ; perlmutter et al . 1999 ) , weak lensing ( refregier 2003 ) , baryon acoustic oscillations ( seo & eisenstein 2003 ; eisenstein et al . 2005 ; pritchard et al . 2007 ;
percival et al .
2007 ) , and cluster counts ( haiman et al .
2000 ) , among several others .
each of these methods presents its own set of difficulties , among them a dependence on integrated quantities , such as the luminosity distance which , however , is not independent of the assumed cosmology .
it is therefore quite difficult to use the data for unbiased , comparative studies to test different expansion histories .
this problem is particularly acute in the case of type ia sne , where at least four ` nuisance ' parameters characterizing the standard candle must be optimized simultaneously with the model s free parameters , rendering the data compliant to the underlying cosmology ( see , e.g. , melia 2012a ) .
even so , some progress has been made recently with attempts at comparing predictions of the @xmath2 universe ( melia 2007 ; melia & shevchuk 2012 ) with the data , and with @xmath1cdm .
the evidence thus far seems to suggest that the @xmath2 cosmology is a better match to the observations at high redshifts , particularly when it comes to the large - scale fluctuations of the cosmic microwave background ( cmb ) , expressed through its angular correlation function and the apparent alignment of its quadrupole and octopole moments ( for a summary of these comparisons , see melia 2012a ) . in the local universe ,
these two models are virtually indistinguishable , e.g. , in predicting a very similar luminosity distance all the way out to a redshift of 6 and beyond .
thus , given the problem of identifying model - independent data through type ia supernova observations , it is not easy to evaluate one model against the other on the basis of these measurements alone . however , some clarification begins to emerge beyond a redshift of 6 , where the high-@xmath14 quasars are now known to be accreting at , or near , their eddington limit ( see , e.g. , willott et al .
2010a , b ; de rosa et al .
we showed recently that a hubble diagram ( hd ) constructed from these sources reveals a cosmic expansion fully consistent with the @xmath2 universe , assuming a current value of @xmath15 km s@xmath5 mpc@xmath5 for the hubble constant @xmath6 ( melia 2012b ) .
interestingly , @xmath1cdm can also fit the high-@xmath14 quasar hd , but only for a very specific set of parameters , including a matter energy density @xmath16 , scaled to its current critical value .
but whereas the @xmath2 universe has only one free parameter the hubble constant the standard model has as many as six ( depending on how one parametrizes the dark - energy equation of state @xmath17 ) including @xmath6 , @xmath18 , and @xmath19 .
the implication of this is that the optimization of @xmath1cdm simply forces it to relax to the @xmath2 expansion profile , which is more robust .
moreover , though the distance
redshift relationship is essentially the same in these two models ( even out to @xmath20 ) , the age
redshift relationship is not .
in fact , these same high-@xmath14 quasars present a seemingly insurmountable problem for @xmath1cdm because they suggest that @xmath21 supermassive black holes appeared only 700900 myr after the big bang .
instead , in @xmath2 , their emergence at redshift @xmath22 corresponds to a cosmic age of @xmath23 gyr .
this was enough time for them to begin growing from @xmath24 seeds ( presumably the remnants of pop ii and iii supernovae ) at @xmath25 ( i.e. , _
after _ the onset of re - ionization ) and still reach a billion solar masses by @xmath26 via standard , eddington - limited accretion ( melia 2013 ) .
this kind of tangible result suggests that the @xmath2 universe relieves the growing tension between @xmath1cdm and the observations , but it would be highly beneficial for us to find a way of testing this cosmology and quantifying its superiority over @xmath1cdm by exploiting model - independent data in the nearby universe .
the purpose of this paper is to demonstrate that the use of luminous red galaxies as cosmic chronometers ( jimenez & loeb 2002 ) can do just that .
we shall show that over the redshift range @xmath27 , the measured hubble constant @xmath0 is fitted better by the @xmath2 model than by @xmath1cdm ; and especially so , if one takes account of the reduction in the number of free parameters . unlike other indicators that rely on the expansion history of the universe , the cosmic chronometers may therefore offer us the best evidence yet that @xmath2 is to be preferred over @xmath1cdm .
we introduce the cosmic chronometers in 2 , and in 3 discuss the aic , kic and bic tools we use to test @xmath1cdm and the @xmath2 universe against these data .
the results of our comparison between @xmath1cdm and @xmath2 are presented in 4 and discussed in 5 .
we conclude in 6 with a discussion of future prospects .
cosmic chronometers offer us the possibility of measuring the differential age of the universe , circumventing the limitations associated with the use of integrated histories , by directly measuring the derivative @xmath28 , which represents the change in cosmic time as a function of redshift . and since @xmath29 , in terms of the expansion factor @xmath30 , a measurement of @xmath28 directly yields the expansion rate , because @xmath31 for various reasons , the best cosmic chronometers appear to be galaxies that are evolving passively on a time - scale much longer than their age difference .
observations indicate that the most massive galaxies contain the oldest stellar populations up to redshifts @xmath32 ( dunlop et al .
1996 ; spinrad et al . 1997 ; cowie et al . 1999
; heavens et al . 2004 ; thomas et al .
less than @xmath33 per cent of the stellar mass in these massive galaxies formed at @xmath34 ( heavens et al . 2004 ; panter et al .
2007 ) . in high - density regions ( i.e. , galaxy clusters ) , star formation ceased by redshift @xmath35 ( thomas et al . 2005 ) , and other massive systems those with stellar masses @xmath36 finished their star formation activity by @xmath37 ( treu et al .
2005 ) .
the empirical evidence therefore suggests that galaxies in the highest density regions of clusters formed their stellar content at @xmath38 , and have been evolving passively since that time , without any additional episodes of star formation .
one can therefore view these galaxies as tracing the ` red envelope , ' hosting the oldest stars in the universe at every redshift .
thus , given their viability as cosmic chronometers , a great deal of effort is being expended to calculate @xmath28 and therefore @xmath0 using their measured properties ( see , e.g. , stern et al .
2010 ; stern et al .
2012 ; moresco et al .
2012a , b ) . for example ,
one of the most direct ways of determining the age of the galaxy is to use the 4000 break in its spectrum , which depends linearly on age for old stellar populations ( moresco et al .
this break is a discontinuity of the spectral continuum due to metal absorption lines whose amplitude correlates linearly with the age and metal abundance . if the metallicity is known , then the difference in age between two galaxies is proportional to the difference in their 4000 amplitudes .
however , one must also be aware of the fact that many systematic sources of uncertainty can bias this kind of analysis ( see , e.g. , moresco et al .
these include : ( 1 ) the degeneracy between the effect of a change in age and an effect due to a change in stellar metallicity or the star formation history ; ( 2 ) the possible biasing of the estimate of @xmath0 by the choice of stellar population synthesis model , used to estimate the age or calibrate the 4000 versus age relation ; and ( 3 ) the possible existence of a progenitor bias ( van dokkum & franx 1996 ) , in which high - redshift samples of early - type galaxies might not be statistically equivalent to those at low redshifts . these caveats notwithstanding ,
one is none the less encouraged by the agreement seen between the results of several different approaches .
the data set shown in figure 1 , including both @xmath0 measurements and error bars , was assembled from the compilations of simon et al .
( 2005 ) , stern et al .
( 2010 ) , and moresco et al .
( 2012a ) , and spans the redshift range @xmath39 .
together , these compilations paint a fairly consistent picture of the universal expansion , particularly when viewed in terms of the theoretical expectations , which we shall consider shortly , following our discussion of model selection statistics .
to compare the evidence for and against competing models , such as models of the distance
redshift relationship , the use of the akaike information criterion ( aic ) is now common in cosmology ( see , e.g. , takeuchi 2000 ; liddle 2004 , 2007 ; tan & biswas 2012 ) .
the aic can be viewed as an enhanced ` goodness of fit ' criterion , which extends the usual @xmath40 criterion by taking account of the number of parameters in each model .
it prefers models with few parameters to those with many , unless the latter provide a substantially better fit to the data .
this reduces the possibility of overfitting : the fact that by optimizing over a greater number of parameters , one may simply be fitting the noise . as developed ( akaike 1973
; see also burnham & anderson 2002 , 2004 ) , the aic provides the relative ranks of two or more competing models , and also a numerical measure of confidence that each model is the best .
these confidences are analogous to likelihoods or posterior probabilities in traditional statistical inference . but unlike traditional inference methods , the aic can be applied to models that are not ` nested . ' comparing a pair of models that are nested , in the sense that one is a specialization of the other , is straightforward : after fitting each model to the data , one computes the @xmath40 per degree of freedom for each , and decides which is a better fit .
one can also calculate ( say , by applying an f - test ) a likelihood that the simpler model should be rejected , or the likelihood of the null hypothesis that the simpler model is a better approximation to the ` true ' one . by exploiting the aic
one can generalize this procedure : one can compare a pair of models , neither of which is a specialization of the other ; such as @xmath1cdm and an alternative model .
the aic can be applied after regression of the following kind is performed .
suppose that for values @xmath41 of an independent variable there are measured values @xmath42 of a dependent one , with ( known ) error bars @xmath43 ; and suppose the errors are normally distributed .
suppose that a model @xmath44 predicts values @xmath45 , computed from a formula @xmath46 that involves a parameter vector @xmath47 comprising @xmath48 unknown parameters , i.e. , @xmath49 .
that is , the data model @xmath44 is really a statistical one , of the form @xmath50 where @xmath51 are independent standard normal random variables .
( in the case of linear regression , @xmath52 would be @xmath53 for known coefficients @xmath54 ; typically , @xmath55 for known functions @xmath56 of @xmath14 . ) for model @xmath44 , the @xmath40 goodness of its fit to the data is given by @xmath57 ^ 2/\sigma_i^2\;,\ ] ] i.e. , a ( weighted ) sum of squared errors , and the reduced @xmath40 ( i.e. , the @xmath40 per degree of freedom ) by @xmath58 ( it is assumed that @xmath59 . ) the parameters @xmath60 are chosen to minimize the @xmath40 , yielding the best fit to the data .
the aic for the resulting fitted model is then given by @xmath61 if there are two or more competing models for the data , @xmath62 , and they have been separately fitted , the one with the least resulting aic is assessed as the one most likely to be nearest to the ` truth , ' i.e. , to the unknown model @xmath63 that generated the data .
a more quantitative ranking of models can be computed as follows .
if @xmath64 comes from model @xmath65 , the unnormalized likelihood that @xmath65 is closest to the truth is the ` akaike weight ' @xmath66 .
informally , @xmath65 has likelihood @xmath67 of being the best choice .
( the @xmath68 s here could of course be omitted by redefining @xmath69 , but the normalization implicit in equation ( 5 ) is traditional . ) in the case of a pair of models @xmath70 , the difference @xmath71 determines the extent to which @xmath72 is favored over @xmath73 .
it is clear that the @xmath74 term in equation ( 5 ) , proportional to the parameter count @xmath48 , exponentially disfavors models with too many parameters , though such models can be favored if they do a much better job of fitting the data .
the choice of proportionality constant ( i.e. , @xmath68 ) is not entirely arbitrary , being based on an argument from information theory that has close ties to statistical mechanics .
the following is a brief summary , with many more details to be found in the statistics literature .
( the reader should note that most of the literature focuses on the case when the error variances @xmath75 are both unknown and equal to some common variance @xmath76 , a nuisance parameter that must be estimated as part of the fitting process ; the setup given in equations 23 is actually simpler . )
any two statistical models of the data set @xmath77 , such as a ` true ' model @xmath78 and another model @xmath44 , can be viewed as probability density functions ( pdf s ) on @xmath79 , say @xmath80 and @xmath81 , respectively . in information theory one says that the discrepancy of the pdf @xmath82 from the pdf @xmath83 , which is a measure of distance , is given by the kullback
leibler formula @xmath84 ( where in an obvious notation , the argument @xmath85 stands for the entire data set @xmath86 $ ] ) . in a thermodynamic interpretation
this is a relative entropy in the sense of boltzmann and hasenhrl . to select the best model @xmath44 from a set of candidate models
, one would choose the one with the minimum @xmath87 .
of course @xmath78 is not known , so this can not be done literally .
but the case when @xmath44 is a parametrized model , and its parameters are chosen ( by minimizing @xmath40 ) to fit a data set generated by @xmath78 , is special .
it can be shown that the aic of the fitted model @xmath44 is a good approximation to @xmath88 , up to an unimportant additive constant .
this is especially the case when @xmath78 is a model of the same type , with unknown parameters @xmath89 .
specifically , @xmath90 is an unbiased estimator of the distance @xmath87 : exactly so for linear regression , and to leading order for non - linear regression .
the phrase ` unbiased estimator ' means that _ on average _ the two are the same , where the averaging is over data sets generated by @xmath78 , with pdf @xmath83 .
of course the fitted model @xmath44 depends on the data set , so in the context of @xmath78 , both @xmath87 and @xmath91 are random variables . in probabilistic language
, the lack of bias means that they have the same expectation .
the extent to which the fitted aic is an _ accurate _ estimate of @xmath88 , data set by data set , as well as being the same on average , has been investigated theoretically ( yanagihara & ohmoto 2005 ) .
its variability has also been studied empirically ; for example , by repeatedly comparing @xmath1cdm to other cosmological models on the basis of data sets generated by a bootstrap method ( tan & biswas 2012 ) . it is known that the aic is increasingly accurate when @xmath92 is large , but it is felt that for all @xmath92 , the magnitude of the difference @xmath93 should provide a numerical assessment of the evidence that model 1 is to be preferred over model 2 .
a rule of thumb that has been used in the literature is that if @xmath94 , the evidence is weak ; if @xmath95 or @xmath96 , it is mildly strong ; and if @xmath97 , it is quite strong . besides using fixed thresholds , one can weight each candidate model in a boltzmann - like way by its akaike weight , i.e. , according to equation ( 6 ) . for each model @xmath65 ,
the likelihood @xmath98 , which is determined by the differences between @xmath64 and the aic s of the other model(s ) , is loosely analogous to a posterior probability in statistical inference , despite its not being computed by a bayesian procedure ( no bayesian prior is involved ) .
but in the absence of a general theory of aic variability , deciding between models 1 and 2 can not be viewed as a hypothesis test , at any fixed level of significance such as 0.05 .
several alternatives to the aic have been considered in the literature .
a lesser - known one arises as follows .
the discrepancy @xmath87 is not symmetric in the pdf s @xmath99 , and it has been suggested that it should be replaced by a symmetrized version , which is arguably a better tool for distinguishing between data models ( cavanaugh 1999 ) .
the unbiased estimator for the symmetrized version has been given the name kic ( kullback information criterion ) , and is given by @xmath100 the kic , with @xmath48 multiplied by the coefficient 3 rather than 2 , disfavors overfitting more than does the aic , and has been shown to perform favorably against the aic as a tool for model selection ( cavanaugh 2004 ) .
it has long been felt ( since , e.g. , bhansali & downham 1977 ) that the problem of overfitting may be best dealt with by choosing a coefficient that is larger than 2 , and perhaps even than 3 .
but the aic and kic are the only such schemes that follow readily from information theory .
a better - known alternative to the aic is the bic ( bayes information criterion ) , which is a misnomer in that it is not based on information theory , but rather on an asymptotic ( @xmath101 ) approximation to the outcome of a conventional bayesian inference procedure for deciding between models ( schwarz 1978 ) .
it is defined by @xmath102 and suppresses overfitting very strongly if @xmath92 is large .
liddle et al .
( 2006 ) and liddle ( 2007 ) make the case for using bic in cosmological model selection , and it has now been used to compare several popular models against @xmath1cdm ( see , e.g. , shi et al .
however , it should be noted that the monograph of burnham & anderson ( 2002 ) , which popularized equation ( 6 ) for assigning aic - based likelihoods to models , strongly prefers aic to bic as a tool for model selection .
they elsewhere note that the aic can in fact be interpreted in bayesian terms , as being the consequence of imposing a nonuniform but reasonable choice of prior distribution on the set of candidate models ( burnham & anderson 2004 ) . kuha ( 2004 ) draws further analogies between aic and bic , and argues that they are both valuable tools . in the comparison below
, we employ the aic , kic , and bic .
we do not employ the so - called corrected aic , denoted @xmath103 , which includes a correction term intended to remove bias when @xmath92 is small ( burnham & anderson 2002 ) .
the correction term is small ( cf .
tan & biswas 2012 ) .
more importantly , the form of this term is appropriate only for data sets without explicit error bars , with the common error variance @xmath76 estimated as part of the fitting process ( maier 2013 , in preparation ) .
the @xmath2 universe is a flat friedmann
robertson walker ( frw ) cosmology that strictly adheres to the constraints imposed by the simultaneous application of the cosmological principle and weyl s postulate ( melia , 2007 ; melia & shevchuk 2012 ; melia 2012a ) .
when these ingredients are applied to the cosmological expansion , the gravitational horizon @xmath104 must always be equal to @xmath105 .
this cosmology is therefore very simple , because @xmath106 , which also means that @xmath107 , with the ( standard ) normalization that @xmath108 .
therefore in the @xmath2 universe , we have the straightforward scaling @xmath109 notice , in particular , that the expansion rate @xmath0 in this model has only one free parameter . by comparison
, @xmath1cdm has as many as six parameters ( depending on the application ) , including @xmath6 , the scaled matter energy density @xmath18 ( @xmath110 , in terms of the matter energy density @xmath111 and the critical density @xmath112h_0 ^ 2 $ ] ) , and the dark - energy equation of state @xmath113 . in this paper
, we shall take the minimalist approach and optimize @xmath1cdm using only these three free parameters .
( using additional parameters would weaken the statistical significance of the fit even further , so by selecting this minimal set , we present @xmath1cdm in its best possible light . )
the hubble constant in this cosmology is therefore given by @xmath114^{1/2}\;,\ ] ] where @xmath115 and @xmath116 for radiation and dark energy , respectively , are defined analogously to @xmath117 .
in addition , we shall assume a flat @xmath1cdm cosmology , for which @xmath118 , thus avoiding the introduction of @xmath116 as an additional free parameter . of course , @xmath115 ( @xmath119 ) is known from the current temperature ( @xmath120 k@xmath121 ) of the cosmic microwave background radiation . for each model @xmath65 ( with @xmath122 specifying the @xmath2 universe and @xmath1cdm , respectively ) , we optimize the fit by finding the model parameter vector @xmath123 that minimizes the @xmath40 .
equivalently , we choose @xmath124 to maximize the joint likelihood function @xmath125 ^ 2/2\sigma_i } \over \sqrt{2\pi}\sigma_i}\;,\ ] ] where the @xmath126 are the measured values of the hubble constant at redshift @xmath127 , and the @xmath128 are the corresponding theoretical values computed from the parameter vector @xmath129 . for @xmath122 ,
the number of parameters ( i.e. , the length @xmath48 of the vector @xmath124 ) is respectively 1 and 3 , as stated .
the fitting is a linear regression in the case of the @xmath2 universe and a non - linear one for @xmath1cdm , as is evident from equations ( 10 ) and ( 11 ) .
the data set @xmath130 to which each model is fitted was assembled from the @xmath0 compilations of moresco et al . ( 2012a ) , stern et al .
( 2010 ) , and simon et al .
( 2005 ) , and consists of @xmath131 measured values , each with an error bar .
the results for the @xmath2 universe ( for which the best fit has @xmath132 km s@xmath5 mpc@xmath5 ) and @xmath1cdm ( for which it has @xmath133 km s@xmath5 mpc@xmath5 , @xmath134 , and @xmath135 ) are shown in figure 1 .
( these @xmath6 values are quoted with one - sigma standard errors , calculated from the corresponding @xmath40-distribution for each model . ) with @xmath136 degrees of freedom , the reduced @xmath11 for the @xmath2 universe is 0.745 . by comparison ,
the optimal @xmath1cdm fit has @xmath137 degrees of freedom , and a corresponding reduced @xmath10 . even by eye
, one can see that @xmath138 is a better fit to the data at @xmath139 .
the reduced @xmath40 overall suggests that @xmath2 is at least as good as @xmath1cdm ; especially , when its having only one parameter is taken into account .
we shall see shortly that on statistical grounds , the @xmath140 distance redshift predictions are in fact more likely than those of @xmath1cdm to be closer to the correct cosmology .
it is worth pointing out that the @xmath1cdm model optimized for the cosmic chronometer data alone is quite different from the concordance model , characterized by the parameter values @xmath141 km s@xmath5 mpc@xmath5 , @xmath142 and @xmath135 . using the concordance @xmath1cdm parameter values to fit the cosmic chronometer data yields @xmath143 , which is acceptable ( since @xmath144 , but which indicates a noticeably less good fit than both the fits shown in figure 1 . by comparing the @xmath140 universe against the optimized @xmath1cdm , rather than against the concordance model
, we are once again presenting @xmath1cdm in its best possible light . with @xmath131 data points and @xmath145 parameter ,
the aic for the optimized @xmath2 universe is @xmath146 . for the optimized @xmath1cdm , with @xmath147 ,
the corresponding value is @xmath148 .
the magnitude of the difference @xmath149 , namely @xmath150 , indicates that @xmath72 is to be preferred over @xmath73 . according to equation ( 6 ) , the likelihood of @xmath2 ( i.e. , @xmath72 ) being the correct choice is @xmath151 per cent . for @xmath1cdm ( i.e. , @xmath73 ) , the corresponding value is @xmath152 per cent .
if one uses the kic and bic statistics instead of the aic , but continues to weight the models as in equation ( 6 ) , the difference is greater , since @xmath48 is multiplied by @xmath153 in the former , and by @xmath154 in the latter .
one finds that @xmath155 and @xmath156 , yielding @xmath157 per cent and @xmath158 per cent for @xmath2 and @xmath1cdm , respectively . and
for bic , the results are @xmath159 and @xmath160 , yielding @xmath161 per cent and @xmath162 per cent . according to all three statistics ,
the predictions of @xmath2 are more likely than those of @xmath1cdm to be closer to the correct cosmology .
this is notably the case for bic , for which there is an accepted interpretation of the magnitude of @xmath163 in terms of the strength of the evidence against model 2 ( kass & raftery 1995 ; tan & biswas 2012 ) . if , as here , @xmath164 , the evidence against model 2 ( i.e. , @xmath1cdm ) would be judged ` positive ' ( the positive range for @xmath165 extends from 2 to 6 , at which point the ` strong ' range begins ) .
though a measurement of the cosmic expansion rate using early type galaxies is subject to several possible systematic errors , the fact that the inferred values of @xmath0 are model - independent makes this a highly desirable and meaningful approach for testing different cosmological models . in this paper , we have compared the fits to a data sample drawn from several sources , and have demonstrated that the @xmath2 universe is more likely than @xmath1cdm to account for the observed @xmath166 versus @xmath14 profile . in addition , the inferred value of the hubble constant @xmath6 is consistent with the rate ( @xmath167 km s@xmath5 mpc@xmath5 ) emerging from a fit to the high-@xmath14 quasar hubble diagram ( melia 2012b ) .
this is rather impressive , given that the former corresponds to a probe of the local universe ( at @xmath168 ) , whereas the latter concerns the cosmic expansion at high redshift ( @xmath169 ) .
in addition , one should not underestimate the fact that in @xmath2 , there is only one free parameter . by comparison
, the standard model of cosmology , with as many as six , depending on how one parametrizes the dark - energy equation of state , fails to account for the appearance of high-@xmath14 quasars at redshift @xmath170 ( melia 2013 ) .
this type of comparative analysis therefore supports the suggestion that @xmath2 is closer to the correct cosmology than is @xmath1cdm .
the growing tension between the predictions of @xmath1cdm and the ever improving cosmological data also suggests that the current standard model may be a useful approximation , but will probably not endure in the long run .
recently , however , some criticism has been leveled at the @xmath2 cosmology on the basis of several claimed inconsistencies , some theoretical , others observational ( bilicki and seikel 2012 ) .
one of the observational arguments was based on the same cosmic chronometer data we have addressed in this paper , from which a different conclusion was arrived at from the one obtained above . however , these earlier results are incorrect : simply , because they were not based on a proper statistical analysis .
those conclusions appear to have been based on a qualitative inspection by eye .
but clearly , the results presented here show that such an approach does not stand up well to a quantitative assessment based on comparisons of likelihoods .
and since the cosmic chronometer data favor the @xmath2 cosmology over @xmath1cdm when using a simple , direct statistical comparison , any higher - order metric employed with the @xmath0 data , particularly those designed to test the parametrization of @xmath1cdm , e.g. , the decomposition of density into the three specific components , matter , radiation , and dark energy , can not be used to meaningfully constrain the @xmath2 universe . on the contrary , as we have shown here , the cosmic chronometer data when interpreted quantitatively suggest that the @xmath2 universe is at least as good as the standard model .
the second observational argument for the criticism was based on the analysis of type ia sne .
however , here too the data were used incorrectly to arrive at an invalid result .
the use of type ia supernova data ignored the fact that these were optimized for a pre - assumed @xmath1cdm cosmology .
therefore , these can not be used for a comparative test between different expansion scenarios . a complete discussion of this problem has already been published in melia ( 2012a ) , so we shall not reproduce it here .
the danger of using data optimized for @xmath1cdm to test other cosmologies has also been highlighted recently by an examination of the gamma ray burst ( grb ) hubble diagram ( wei et al .
2013 ) . in this work
, the data were recalibrated separately for each model and , though the results are quite similar , a comparison of the reduced @xmath11 s for @xmath2 and @xmath1cdm shows that the data clearly favor the former over the latter .
this result would not have been evident without a recalibration of the data using the @xmath2 expansion history . given the preponderance of evidence , it seems likely that when the type ia sn data are calibrated correctly for each cosmology , @xmath2 will emerge as the more likely of the two to be correct .
finally , it is worth pointing out that the ` theoretical difficulties ' invoked to argue against @xmath2 are based on a failure to comprehend fully birkhoff s theorem and its corollary , and the consequence of weyl s postulate on friedmann robertson walker ( frw ) metrics .
it is not the purpose of this paper to correct this misunderstanding , but since it appears to be an issue relevant to the interpretation of cosmic chronometer data , we shall address it here as well .
birkhoff s theorem and its corollary ( birkhoff 1923 ) place no limit on scale , so bilicki and seikel s ( 2012 ) assertion that the definition of a schwarzschild ( i.e. , a gravitational ) radius makes no sense on cosmic dimensions is without foundation .
moreover , one does not ` define ' a schwarzschild radius , as was claimed ; this scale emerges automatically when one re - writes the metric in terms of observer - dependent coordinates versus the more commonly used co - moving coordinates ( see , e.g. , melia & abdelqader 2009 ) .
many are perhaps not aware of the fact that exactly the same phenomenon occurs when writing the spacetime metric for compact objects .
the distinction arises between a free - falling observer and the observer at a fixed radius ( and hence accelerated ) relative to the central mass .
the former is not aware of the gravitational radius that emerges only when the metric is written using rulers and clocks fixed to the latter . in the cosmological context , we are free - falling observers when we write the frw metric using co - moving coordinates .
however , a gravitational radius emerges when we re - write this metric in terms of an observer s fixed coordinates . the irony , of course
, is that the gravitational radius in cosmology actually first appeared as far back as 1917 , though its meaning was not then fully appreciated .
de sitter s ( 1917 ) paper on his now famous metric was originally written in terms of the observer s fixed coordinates , which included the gravitational radius , since the co - moving coordinates would be introduced by friedmann only several years later .
the argument against the validity of a gravitational horizon in cosmology would therefore imply that de sitter space is meaningless on large scales .
this is simply not true . and
since the meaning and validity of the gravitational radius in cosmology ( which , by the way , coincides with the better known hubble radius ) were not appreciated , the consequence of weyl s postulate on its permitted rate of expansion was ignored .
since the hubble radius is a ` proper ' radius , it has no choice but to expand at a constant rate , as demonstrated by melia & shevchuk ( 2012 ) and ( in the more pedagogical treatment ) by melia ( 2012c ) .
_ euclid _ ( laureijs et al
. 2011 ) and _ boss _ ( eisenstein et al . 2011 ) should provide thousands of passive galaxies at @xmath171 , which will significantly improve the accuracy of @xmath0 at these higher redshifts . in concert with this improved statistic
, it will be essential to understand better if the systematic effects , e.g. , the error due to the metallicity and star formation uncertainties , may be controlled and minimized .
it is crucial to carry out this arduous work , because these cosmic chronometers are among the few sources that provide us with model - independent data . and
as we have seen , only such model - independent data can truly distinguish between competing cosmologies .
we end with a word of caution
. it should be evident from the contents of this paper how important it is to use only model - independent data in any comparative analysis between competing cosmologies . in some cases
, it is simply not possible to avoid the ` circularity problem , ' in which a model must be pre - assumed in order to extract the data .
this is certainly the case in the type ia sn work , but also when dealing with any observations requiring the use of integrated quantities , such as the luminosity distance .
an entirely different approach sometimes used to determine @xmath0 is based on the identification of baryon acoustic oscillations ( bao ) and the alcock paczynski distortion from galaxy clustering .
that is , instead of using information on how cosmic time changes with @xmath14 ( as is the case for the data we have used here ) , this alternative approach measures how ` standard rulers ' evolve with redshift .
the results of these two different methods are sometimes combined to produce an overall @xmath166 versus @xmath14 diagram , but there is a good reason to be wary of this procedure : whereas the cosmic chronometers produce model - independent data , the second approach must necessarily assume a particular cosmology and is therefore model - dependent ( blake et al .
2012 ) . with the bao method ,
the cosmic expansion is measured from the growth of structure as a function of redshift .
redshift - space distortions arise because the recession velocities of galaxies , from which distances are inferred , include contributions from both the hubble flow and from the peculiar velocities driven by the clustering of matter ( see , e.g. , hamilton 1998 for a review ) .
the oscillations are modeled via the non - linear evolution of both the matter density and velocity fields , which are quite different between , say , @xmath1cdm and @xmath2 ( melia & shevchuk 2012 ) . in addition , to compute redshift space separations for each pair of galaxies given their angular coordinates and redshifts , one must adopt a cosmological model for the expansion to relate these quantities to each other .
unfortunately , this gives rise to a situation not unlike that currently existing with type ia sne ( melia 2012a ) , in which one must simultaneously optimize at least four nuisance parameters incorporated into the description of the measurements , along with the free parameters of the assumed model .
one must therefore avoid the use of such model - dependent data in any attempts to directly compare fits using @xmath1cdm with those of other models , such as @xmath2 .
only the cosmic chronometer data are truly model - independent and therefore appropriate for this purpose .
we are grateful to the many workers who spent an extraordinary amount of effort and time accumulating the data summarized in figure 1 .
fm is also grateful to amherst college for its support through a john woodruff simpson lectureship .
part of this work was carried out at purple mountain observatory , nanjing china .
akaike , h. 1973 , in second international symposium on information theory , eds .
b. n. petrov and f. cski ( budapest : akadmiai kiad ) , 267 bilicki , m. & seikel , m. 2012 , mnras , 425 , 1664 bhansali , r. j. & downham , d. y. 1977 , biometrika , 64 , 547 .
birkhoff , g. 1923 , relativity and modern physics ( cambridge : harvard university press ) blake , c. , et al .
2012 , mnras , 425 , 405 burnham , k. p. & anderson , d. r. 2002 , model selection and multimodel inference , 2nd edn .
( new york : springer - verlag ) burnham , k. p. & anderson , d. r. 2004 , sociol .
methods res . , 33 , 261 cavanaugh , j. e. 1999 , statist .
lett . , 42 , 333 cavanaugh , j. e. 2004 , aust
. n. z. j. stat . , 46 , 257 cowie , l. l. , songaila , a. , & barger , a. j. 1999 , aj , 118 , 603 de rosa , g. , decarli , r. , walter , f. , fan , x. , jiang , l. , kurk , j. , pasquali , a. , & rix , h .- w .
2011 , apj , 739 , article i d 56 de sitter , w. 1917 , proc .
amsterdam , 19 , 1217 dunlop , j. , et al .
1996 , nature , 381 , 581 eisenstein , d. j. , et al .
2005 , apj , 633 , 560 eisenstein , d. j. , et al . 2011 , aj , 142 , 72 haiman , z. , mohr , j. j. , & holder , g. p. 2000 , apj , 553 , 545 hamilton , a.j.s .
1998 , in the evolving universe , ed .
d. hamilton ( dordrecht : kluwer academic ) , 185 heavens , a. , panter , b. , jimenez , r. , & dunlop , j. 2004 , nature , 428 , 625 jimenez , r. & loeb , a. 2002 , apj , 573 , 37 kass , r. e. & raftery , a. e. 1995 , j. amer . statist .
assoc . , 90 , 773 kuha , j. 2004 , sociol .
methods res .
, 33 , 188 laureijs , r. , et al .
2011 , eprint ( arxiv:1110.3193 ) liddle , a. , mukherjee , p. , & parkinson , d. 2006 , astron . & geophys . , 47 , 4.30 liddle , a. r. 2004 , mnras , 351 , l49 liddle , a. r. 2007 , mnras , 377 , l74 melia , f. 2007 , mnras , 382 , 1917 melia , f. 2012a , aj , 144 , 110 melia , f. 2012b , aj , submitted melia , f. 2012c , australian physics , 49 , 83 melia , f. 2013 , apj , 764 , 72 melia , f. & abdelqader , m. 2009 , int . j. mod . phys .
d , 18 , 1889 melia , f. & shevchuk , a. 2012 , mnras , 419 , 2579 moresco , m. , et al .
2012a , jcap , 8 , article i d 006 moresco , m. , jimenez , r. , cimatti , a. , & pozzetti , l. 2011 , jcap , 03 , article i d 45 moresco , m. , verde , l. , pozzetti , l. , jimenez , r. , & cimatti , a. 2012b , jcap , in press ( arxiv:1201.6658 ) panter , b. , jimenez , r. , heavens , a. f. , & charlot , s. 2007 , mnras , 378 , 1550 percival , w. j. , et al .
2007 , mnras , 381 , 1053 perlmutter , s. , et al . 1999 ,
apj , 517 , 565 pritchard , j. r. , furlanetto , s. r. , & kamionkowski , m. 2007 , mnras , 374 , 159 refregier , a. 2003 , araa , 41 , 645 riess , a. g. , et al .
1998 , aj , 116 , 1009 schwarz , g. 1978 , ann .
, 6 , 461 seo , h .- j . & eisenstein , d. j. 2003 , apj , 598 , 720 shi , k. , huang , y. f. , & lu , t. 2012 , mnras , 426 , 2452 simon , j. , verde , l. , & jimenez , r. 2005 , phys .
d , 71 , 123001 spinrad , h. , et al .
1997 , apj , 484 , 581 stern , d. , jimenez , r. , verde , l. , stanford , s. a. , & kamionkowski , m. 2010 , apjs , 188 , 280 stern , d. , jimenez , r. , verde , l. , kamionkowski , m. , & stanford , s. a. 2012 , jcap , in press ( arxiv:0907.3149 ) takeuchi , t. t. 2000 , ap&ss , 271 , 213 tan , m. y. j. & biswas , r. 2012 , mnras , 419 , 3292 thomas , d. , maraston , c. , bender , r. , & de oliveira , c. m. 2005 , aj , 621 , 673 treu , t. , et al .
2005 , apj , 633 , 174 van dokkum , p. & franx , m. 1996 , mnras , 281 , 985 wei , j .- j . ,
wu , x. , & melia , f. 2013 , apj , submitted ( arxiv:1301.0894 ) willott , c. j. , et al . 2010a , aj , 140 , 546 willott , c. j. , et al .
2010b , aj , 139 , 906 yanagihara , h. & ohmoto , c. 2005 , j. statist .
inference , 133 , 417 | the use of luminous red galaxies as cosmic chronometers provides us with an indispensable method of measuring the universal expansion rate @xmath0 in a model - independent way . unlike many probes of the cosmological history ,
this approach does not rely on integrated quantities , such as the luminosity distance , and therefore does not require the pre - assumption of any particular model , which may bias subsequent interpretations of the data .
we employ three statistical tools the akaike , kullback , and bayes information criteria ( aic , kic and bic ) to compare the @xmath1cdm model and the @xmath2 universe with the currently available measurements of @xmath0 , and show that the @xmath2 universe is favored by these model selection criteria .
the parameters in each model are individually optimized by maximum likelihood estimation .
the @xmath2 universe fits the data with a reduced @xmath3 for a hubble constant @xmath4 km s@xmath5 mpc@xmath5 , and @xmath6 is the sole parameter in this model . by comparison , the optimal @xmath1cdm model , which has three free parameters ( including @xmath7 km s@xmath5 mpc@xmath5 , @xmath8 , and a dark - energy equation of state @xmath9 ) , fits the @xmath0 data with a reduced @xmath10 . with these @xmath11 values ,
the aic yields a likelihood of @xmath12 per cent that the distance
redshift relation of the @xmath2 universe is closer to the correct cosmology , than is the case for @xmath1cdm .
if the alternative bic criterion is used , the respective bayesian posterior probabilities are 91.2 per cent ( @xmath2 ) versus @xmath13 per cent ( @xmath1cdm ) .
using the concordance @xmath1cdm parameter values , rather than those obtained by fitting @xmath1cdm to the cosmic chronometer data , would further disfavor @xmath1cdm .
[ firstpage ] cosmological parameters , cosmology : observations , cosmology : redshift , cosmology : theory , distance scale , galaxies |
given a closed riemann surface @xmath4 of genus @xmath2 with compatible riemannian metric with khler form @xmath5 , we study the space @xmath6 , or the space of degree @xmath1 holomorphic maps from @xmath4 to @xmath3 .
this space has generated great interest in topological field theory , in which @xmath6 is the moduli space of charge @xmath1 @xmath3 lumps on @xmath4 , or the static solitons to field equations of @xmath3 model on @xmath4 within the topological sector @xmath1 . from the variational point of view
, holomorphic maps are precisely the minimizers to the @xmath0 energy functional @xmath7 where the norm above will be defined in section 2 .
the volume of @xmath6 with respect to the @xmath0 metric defined by the energy functional is a vital quantity in studying the dynamics of solitons . with a clear understanding of the volume growth
, the equations of classical gas of lumps may be deduced and motions of solitons at low energy may be approximated .
see @xcite , @xcite , or @xcite for further details .
a plausible formula for the @xmath0 volume of @xmath6 has been conjectured in @xcite and rigorously verified in @xcite for the case @xmath8 ( ie .
degree 1 maps on @xmath9 ) . in this paper
, we confirm the validity of the formula in general .
the @xmath0 volume of @xmath6 is @xmath10 where @xmath11 and @xmath2 is the genus of @xmath4 .
[ main theorem intro ] throughout this paper , we make the universal assumption @xmath12 that ensures @xmath6 to be a nonempty complex manifold .
in this section we aim to provide a concrete description of the space @xmath6 as a complex manifold of finite dimension , equipped with the standard @xmath0 metrics on maps .
theorem [ description of hol ] , the main result of this section , is an easy consequence of results from @xcite , @xcite , and @xcite . to ensure that the @xmath0 metric is meaningful , we assume that the images of the holomorphic maps discussed here are non - degenerate ,
that is , they do not lie in any hyperplane . for clarity
, we first list several basic and well known definitions of the objects used in this article and their basic properties .
readers familiar with these notions may skip to theorem [ description of hol ] and consequences following it .
the elementary building block for various moduli spaces is the @xmath1 symmetric product of @xmath4 . for riemann surface @xmath4
, it is equivalently the space of unordered @xmath1-tuple of points on @xmath4 , or the space of effective divisors of degree @xmath1 .
given the @xmath1-fold direct product of @xmath4 , we define @xmath13 where the equivalent relation is given by @xmath14 with @xmath15 , the symmetric group . [ definition of symmetric product ] elements of @xmath16 are denoted by @xmath17 , where @xmath18 s need not be distinct .
it is a standard fact that @xmath16 is a complex manifold of complex dimension @xmath1 .
the local holomorphic coordinates are given by elementary symmetric functions .
( see , for example , @xcite ) .
the space @xmath16 is closely related to the space of nonzero meromorphic functions on @xmath4 of degree @xmath1 , or @xmath19 .
since @xmath4 is compact , any @xmath20 has @xmath1 poles and @xmath1 zeros , counting multiplicities .
each meromorphic function is then associated with a pair of disjoint effective divisors of degree @xmath1 : @xmath21 where @xmath22 .
the converse statement is the classical abel s theorem , that a pair of disjoint effective divisors of the same degree gives rise to a meromorphic map as above if they have the same value under the abel - jacobi map .
precisely ( cf .
chapter v of @xcite ) , given @xmath23 , the @xmath24-basis of the space of holomorphic one forms on @xmath4 , we define @xmath25 as follows . for @xmath26 , we have @xmath27 here , @xmath28 is any point outside the support of @xmath29 , and @xmath30 is the jacobian torus of @xmath4 , or the quotient of @xmath31 by the @xmath32-span of row vectors of the @xmath33 period matrix of @xmath4 , @xmath34 , where @xmath35 is the basis for @xmath36 .
[ definition of abel jacobi ] one notes that the quotient space eliminates the ambiguities of the definition of @xmath37 in .
a few well - known properties of @xmath37 are relevant for the discussions of this section ( see @xcite for details ) : * if @xmath38 , that is , @xmath39 for some @xmath40 .
* @xmath37 is holomorphic on @xmath16 .
* @xmath41 .
the third fact in particular implies that the fiber of @xmath37 at each @xmath42 is the projectivization of the vector space @xmath43 for @xmath44 , riemann - roch theorem implies that for all @xmath45 , the fiber of @xmath37 at @xmath46 is a projective vector space of dimension @xmath47 .
therefore , @xmath16 is a vector bundle of rank @xmath47 over the jacobian torus @xmath30 with the projection map @xmath37 .
we can now state the sufficient condition for the difference of a pair of divisors to be principal .
given a pair @xmath48 with @xmath49 , there exists @xmath50 so that @xmath51 .
[ abel s theorem ] the map @xmath52 is unique up to multiplication of a non - vanishing holomorphic map . for compact riemann surface , it is then unique up to a nonzero multiple
. we are now equipped with sufficient language to describe the space @xmath6 . in @xcite ,
a complete description was provided , without explicit proof , for the space of based holomorphic maps from @xmath4 to @xmath3 . given @xmath53 and @xmath54
, we define the space of based holomorphic maps , @xmath55 , to be the subset of @xmath6 of all maps sending @xmath56 to @xmath57 .
[ definition of hol based ] the space @xmath55 can be identified with an open set subset of @xmath58 direct product of @xmath16 given by @xmath59 [ description of based holomorphic maps ] given @xmath60 , since @xmath61 is nondegenerate , it is a curve of degree @xmath62 in @xmath3 .
let @xmath63 be the hyperplane of @xmath3 defined by the canonical section @xmath64 of @xmath65 .
in local coordinate , points of @xmath63 is of the form @xmath66 $ ] .
then , counting multiplicities , we have @xmath67 since @xmath52 is nondegenerate , @xmath68 are in general position in @xmath69 , and we may define coordinate patches of @xmath3 so that each @xmath70 has homogeneous coordinate @xmath71_{i=0}^k$ ] , @xmath72 . for each @xmath73 , let @xmath74 .
since @xmath52 is of degree @xmath1 , it follows that @xmath75 . for @xmath76
, we let @xmath77)\in{sym^r \sigma}$ ] . we must show that the tuple @xmath78 satisfies the condition of the open set given above .
it is clear that @xmath79 since @xmath80 for all @xmath81 and therefore @xmath82 . for the level set conditions
, it suffices to observe that all divisors given above are linear equivalent .
indeed , define @xmath83 by @xmath84):= \frac{z_j}{z_i}.\ ] ] each @xmath85 is globally defined on @xmath3 and therefore @xmath86 .
moreover , with the choice of coordinates associated to @xmath70 s above , it is clear that for all @xmath87 , @xmath88 it then follows that all divisors chosen above are linearly equivalent .
conversely , given @xmath78 satisfying the conditions of the open subset .
abel s theorem ensures that for each @xmath89 , there exists @xmath90 such that @xmath91 , unique up to a nonzero multiple @xmath92 . since @xmath52 is a based map , the constant multiples @xmath93 are completely determined .
we define the corresponding map @xmath52 coordinate - wise by @xmath94.\ ] ] this map is holomorphic away from @xmath95 .
however , near each point of @xmath95 , we may holomorphically extend @xmath52 across the singularities by multiplying the poles of minimum order of vanishing ( see @xcite for details ) , obtaining a holomorphic map from @xmath4 to @xmath3 .
@xmath52 is clearly of degree @xmath1 , since @xmath96)=e_0 \in { sym^r \sigma}$ ] .
the two associations constitute a one - to - one correspondence between @xmath55 and @xmath97 .
lifting the restriction on based maps corresponds to the freedom of choosing @xmath98 and we have established @xmath6 is diffeomorphic to @xmath99 , where @xmath55 is defined in proposition [ description of based holomorphic maps ] .
[ description of hol ] in the proof of proposition [ description of based holomorphic maps ] , the function @xmath52 defined by @xmath100 is unique up to a choice of @xmath58 nonzero constants ( including 1 ) .
moreover , the choice of these @xmath58 constants define the same @xmath52 if they lie on the same line in @xmath101 . the complex dimension of @xmath6 can be readily computed .
@xmath6 is a complex manifold of complex dimension @xmath102 [ dimension of hol ] it suffices to show that the space of based maps @xmath55 is of dimension @xmath103 , since the remaining summand in the conclusion of theorem [ description of hol ] contributes @xmath62 to the dimension .
note that @xmath55 is the submanifold of a @xmath104-dimensional manifold defined by the linear map @xmath105 given as , @xmath106 it suffices to show that @xmath107 is of rank @xmath108 at every point on the inverse image of @xmath109 .
indeed , consider the linear map @xmath110 defined by @xmath111 from the third property of the abel - jacobi map listed immediately after definition [ definition of abel jacobi ] , we see that the inverse image of 0 is precisely @xmath112 which is of dimension @xmath113 by riemann - roch theorem ( note again @xmath12 ) .
the rank of @xmath114 is then @xmath115 and therefore each component of @xmath107 contributes rank @xmath2 to the entire map .
this completes the proof .
the space @xmath6 is not compact and is potentially problematic when discussing convergence of integrals over it . as such
, we introduce the well known uhlenbeck compactification of @xmath6 and will later discuss the convergence of integrals over this compact set .
the following description is summarized from @xcite : the _ uhlenbeck compactification _ of @xmath6 is defined by @xmath116,\ ] ] where @xmath117 is the space of degree @xmath118 holomorphic maps from @xmath4 to @xmath3 .
[ definition of uhlenbeck compactification ] the topology of @xmath119 is given by weak * topology on @xmath120 and @xmath121 topology on @xmath117 .
precisely , for each @xmath122 , we associate the distribution @xmath123 , where @xmath124 the energy density and @xmath125 is the dirac distribution supported in @xmath126 . around @xmath127 , there are two local bases of neighbourhoods of topologies : one basis @xmath128 of @xmath123 in the weak * topology and one basis @xmath129 of @xmath52 in @xmath130 . collecting them together ,
we form a basis of neighbourhoods @xmath131 in @xcite , it was shown that @xmath119 with this topology is compact . since @xmath6 is open and dense subset of @xmath119 , it is indeed a smooth compactification of @xmath6 . in our main result , we will derive the convergence statement of integrals on @xmath119 and ensure that the complement does not contribute to the integral .
one notes that for @xmath132 , this topology degenerates into @xmath133 topology of functions , which coincide with the subspace topology inherited from the direct sums of symmetric products .
indeed , the locations of zeros determines the smooth topology as well as the energy densities .
moreover , the smooth structure of symmetric spaces provides a natural smooth structure for @xmath119 , namely , the @xmath134 topology controlled by the locations of the zeros
. we may describe @xmath119 with explicit coordinates .
we observe that @xmath119 is a closed submanifold of @xmath135 .
the component from symmetric products determines the zeros from @xmath58 meromorphic functions on @xmath4 , a total of @xmath58 unordered @xmath1-tuples of points from @xmath4 . with the @xmath62 constraint equations by abel - jacobi maps of rank @xmath2
, it follows that there are @xmath136 points from @xmath137 that determine the correct zeros of each meromorphic function up to a free choice of point in @xmath3 .
@xmath119 therefore possesses local coordinate description @xmath138 where the coordinates @xmath139 is the local coordinate of the submanifold of @xmath137 defined by abel - jacobi equations and @xmath140 is the local coordinate for @xmath3 . here , @xmath141 as in lemma [ dimension of hol ] .
is also a local coordinate for the open subset @xmath6 , with the additional open condition that @xmath139 does not yield meromorphic functions with common zeros or poles .
lastly , we define the @xmath0 metrics on @xmath6 with respect to the khler metric @xmath5 on @xmath4 and fubini - study metric on @xmath3 . given @xmath142 , the tangent space of @xmath6 at @xmath52 can be identified with the space of sections of the pullback bundle of @xmath143 via @xmath52 : @xmath144 given @xmath145 , they can be viewed as a pullbacked sections on @xmath4 , which can be pushed forward by @xmath52 to be tangent vectors on @xmath69 , on which fubini - study metric can be applied .
we define [ @xmath0 metrics on @xmath6 ] the @xmath0 metric on @xmath6 is given by @xmath147 here , the @xmath148 denotes the pushforward of @xmath52 . with respect to this metric
, we aim to verify the conjectured formula for the volume of @xmath6 .
we hereby provide a brief summary of the @xmath107-vortex equations and its moduli spaces .
standard texts and papers for this subject include @xcite , @xcite , @xcite , @xcite ( of more physical interest ) , @xcite , @xcite , @xcite , and @xcite ( of more mathematical aspects ) .
the conclusion below is drawn directly from @xcite , which contains a comprehensive summary the standard texts .
on the same riemann surface defined above , we consider a hermitian line bundle @xmath149 with degree @xmath12 possessing smooth global sections .
for each @xmath150 and @xmath151 , the @xmath107-vortex equations are defined on @xmath152 , the product of the space of @xmath153-unitary connections and @xmath58 tuples of smooth sections , we have the @xmath154vortex equations @xmath155 here , @xmath156 is the curvature form of the connection @xmath29 .
we use @xmath157 to denote a @xmath158-tuple of sections @xmath159 and the second equation is an abbreviation that @xmath160 . via bogomolny - type arguments , the set of equations are precisely the minimizing equations the @xmath107-yang - mills - higgs energy functional : @xmath161 the solution space to is invariant under unitary gauge action , and we let @xmath162 be the @xmath163 gauge classes of solutions to .
elements of this space are called vortices . for a line bundle @xmath164
, it is well known that the stability condition , or the non - emptyness of @xmath162 , is simply @xmath165 note that we intentionally define @xmath58 sections for vortex equations to accommodate the main theme of this article , the space of holomorphic maps to @xmath3 .
it has been noted in @xcite and @xcite that @xmath6 is diffeomorphic to the open subset of @xmath162 , denoted by @xmath166 , consisting of vortices with @xmath58 having no common zeros : @xmath167\in \nu_{k+1}(s)\;|\;\cap_i \phi_i^{-1}(0)= \emptyset\}.\ ] ] we have , for all @xmath107 such that @xmath168 , there is a diffeomorphism @xmath169 [ moduli space correspondence ] ( _ sketch _ ) we provide only the definition of @xmath170 as a reference for later construction . for complete proofs and justifications ,
see @xcite for the case @xmath171 and @xcite for generalization to all @xmath107 . the inverse direction of @xmath170 is more obvious . for @xmath172\in \nu_{k+1,0}(s)$ ] , we define @xmath173)(z):=\tilde{\phi}(z)=[\phi_0(z),\ldots,\phi_k(z)].\ ] ] the map is well defined as @xmath174 s are never all zero . moreover , on a @xmath163 bundle , the transition map multiplies each section by a uniform nonzero scalar , having no effect on the definition of @xmath175 . since each component is holomorphic near every @xmath176 and has order of vanishing @xmath1 , @xmath177 $ ] is indeed an element of @xmath6 .
the forward direction is also a standard construction in algebraic geometry .
we start with a holomorphic map @xmath178 .
consider @xmath179 , the anti - tautological line bundle over @xmath3 with hyperplane sections @xmath180 .
each @xmath181 vanishes precisely on the hyperplane defined by @xmath182 .
let @xmath183 be the pullbacked line bundle on @xmath4 endowed with sections @xmath184 by pulling back @xmath185 via @xmath175 .
the map @xmath175 also endows a holomorphic structure @xmath186 and a background metric @xmath153 on @xmath164 by pulling back the standard complex structure and fubini - study metric @xmath187 on @xmath188 , determining a corresponding background unitary connection @xmath29 .
we then find a suitable complex gauge transformation turning @xmath189 into a pair that solves the vortex equations . on a line bundle ,
such a transformation is given by a nonzero real smooth function @xmath190 , acting on @xmath189 in the standard ways : @xmath191 and @xmath192 let @xmath193 . plugging @xmath194 into ,
we see that the function @xmath195 must satisfies the following kazdan - warner type equation : @xmath196 here , @xmath197 is the positive definite laplacian determined by @xmath5 .
it has been shown in @xcite ( for @xmath171 ) and @xcite ( for general @xmath107 ) that equation has a unique smooth solution provided that @xmath107 is large enough .
the analytic tool used to solve @xmath195 is developed in @xcite using the method of super and sub solutions .
we will mention more details in the next section . that the two correspondences are smooth inverses to each other is verified in @xcite and @xcite via straightforward arguments . with this correspondence
, the entire vortex moduli space can be identified without difficulty . for a general vortex @xmath172 \in \nu_{k+1}(s)$ ] , we first locate the common zeros of the @xmath198 s , an element @xmath199 for some @xmath200 .
the @xmath58 sections define a holomorphic map from @xmath126 to @xmath3 in identical way as @xmath201 .
this is map is bounded away from @xmath126 , and can therefore be uniquely extended to the entire @xmath4 by locally dividing out the zeros of minimum order of vanishing near each common zero .
the holomorphic map defined is then of degree @xmath118 .
see @xcite and @xcite for details .
we have the vortex moduli space @xmath202 is diffeomorphic to the uhlenbeck compactification of @xmath6 : @xmath203 [ moduli space of entire vortex ] the @xmath0 metrics for @xmath162 is also defined naturally . at @xmath204 , we define , @xmath205 here , @xmath206 denotes a tangent vector in @xmath207 , and @xmath208 .
the identification is justified by the fact that @xmath209 is a vector space and @xmath210 is an affine space modeled on the vector space @xmath211 , the space of complex valued one forms on @xmath4 .
chapter v of @xcite ) .
this identification also justifies the applications of hodge star @xmath212 and @xmath213 in the integrand of , since @xmath214 lies in essentially isomorphic spaces as @xmath194 does . by choosing tangent vectors orthogonal to @xmath215-gauge transformations ,
descends to a well defined metric on the quotient space @xmath216 and restricts to the open subset @xmath166 ( cf .
@xcite ) . with respect to this metric
, baptista has explicitly computed the formula for the volume of @xmath162 in @xcite , generalized directly from @xcite : for a degree @xmath1 hermitian line bundle @xmath164 over a closed khler riemann surface @xmath4 of genus @xmath2 , where @xmath44 , the @xmath0 volume of the vortex moduli space @xmath162 is given by @xmath217 where @xmath218
[ volume for nu k ] the method for the construction of this formula is to realize @xmath162 as a fiber bundle over the moduli space of holomorphic structures , which are identified with the jacobian torus .
the khler form of the @xmath0 metric can be decomposed into parallel and horizontal components in terms of characteristic classes and fiberwise integration is performed to obtain the formula .
the asymptotic behaviours of @xmath162 , , and as @xmath219 are our main interest .
one observes that in the @xmath154vortex equations , the only @xmath107 dependence lies in the third equation . as @xmath107 increase ,
the curvature term in the third equation becomes negligible compared to the section terms and we may formally define the vortex moduli space at infinity , @xmath220 , to be the unitary gauge class of solutions to @xmath221 it is then immediately clear from the third equation in that no @xmath58 tuples of sections in @xmath220 vanish simultaneously .
we have @xmath222 .
moreover , indicates that solutions to these equations consist of precisely pairs of unitary connection @xmath29 and @xmath58 @xmath29-holomorphic sections with image in the unit sphere @xmath223 .
modulo gauge transformation , each vortex at @xmath224 then defines a holomorphic map from @xmath4 to @xmath3 in exactly identical way provided in the proof of theorem [ moduli space correspondence ] .
we have a diffeomorphism @xmath225 defined with precisely the same way as @xmath170 . a naturally plausible attempt to compute the @xmath0 volume of @xmath6 , as conjecturized in @xcite , is then to let @xmath226 in . with the established result in @xcite that the @xmath0 metric converges to the @xmath0 metric in the sense of cheeger - gromov , we aim to further establish the fact that formula is true at @xmath224 .
we provide a brief conclusion of solving @xmath195 using techniques developed in @xcite , which are important tools for the main constructions of this article .
we recall equation , the pde characterizing solutions to vortex equations : @xmath227 solving this equation is equivalent to solving the kazdan - warner equations in @xcite in the following ways .
normalizing the khler metric @xmath5 so that @xmath228 , consider @xmath229 where @xmath230 is independent of @xmath107 and @xmath153 .
consider @xmath231 , the solution to : @xmath232 which is clearly independent of @xmath107 .
here , @xmath197 is the laplacian operator with respect to the khler form @xmath5 . setting @xmath233
, one can readily see that @xmath195 satisfies if and only if @xmath234 satisfies : @xmath235 this is of the form : @xmath236 with the strictly negative norm function @xmath237 and @xmath238 for @xmath107 in appropriate range .
lemma 9.3 in @xcite guarantees the unique solution to exist using the method of super and sub solutions , a consequence of the maximum principle applied to the operators @xmath239 , where @xmath240 is a constant determined by @xmath241 . in @xcite , sufficient conditions to solve are merely @xmath242 .
however , to obtain further uniformity and convergent behaviors of @xmath195 over @xmath107 requires @xmath243 , which depends crucially on the non - simultaneous vanishing of the @xmath58 sections . in @xcite ,
the author has proved on a compact riemannian manifold @xmath244 without boundary , let @xmath245 be any constant , @xmath246 any positive constant , and @xmath241 any negative smooth function .
let @xmath247 , for each @xmath107 large enough , the unique solutions @xmath248 for the equations @xmath249 are uniformly bounded in all sobolev spaces .
moreover , in the limit @xmath226 , @xmath250 converges smoothly to @xmath251 the unique solution to @xmath252 [ main theorem on convergence of gauge ] the theorem is an independent analytic result with the additional benefit that it holds on general compact riemannian manifolds of any dimension and complex structure is not necessary .
in particular , the uniformity of kazdan - warner equations can be deduced from this theorem with @xmath253 and @xmath254 .
theorem [ main theorem on convergence of gauge ] is proved by explicitly constructing the family of approximated solutions @xmath255 that approach @xmath256 smoothly and moreover , the difference of @xmath255 and @xmath234 approaches zero in any sobolev norm .
the family of approximated solutions is given by @xmath257 with the family of error terms @xmath258 so that @xmath259 the two families @xmath255 and @xmath234 satisfy the following lemma : for all @xmath260 , the family of actual and approximated solutions to the kazdan - warner equations , @xmath234 and @xmath255 , we have @xmath261 where @xmath262 is the sobolev space of order @xmath263 and exponent @xmath264 .
[ convergence of approximated and actual solutions ] the lemma will again be a crucial part of the main result of this paper .
the complete proof is summarized in the next section .
the convergence result and its proof may be directly applied to prove a conjecture on convergence of @xmath0 metrics of vortex moduli spaces to the ordinary @xmath0 metric on maps , posed in @xcite and proved in @xcite .
the convergence holds in the sense of cheeger - gromov : [ cheeger - gromov convergence ] for all @xmath260 and @xmath265 , a family of @xmath266-dimensional riemannian manifolds @xmath267 is said to converge to a fixed riemannian manifold @xmath268 in @xmath269 , in the sense of cheeger - gromov , if there is a covering chart @xmath270 on @xmath244 and a family of diffeomorphisms @xmath271 , such that @xmath272 as @xmath226 , for all @xmath62 and @xmath273 .
we have equipping @xmath3 with the fubini - study metric , the sequence of metrics @xmath274 on @xmath166 given by converges in all sobolev spaces , in the sense of cheeger - gromov , to the @xmath0 metric @xmath275 on @xmath6 given by .
the family of diffeomorphisms are precisely @xmath170 , as constructed in theorem [ moduli space correspondence ] .
[ precise baptista s conjecture ] ( _ sketch _ ) we provide a brief outline of the proof in @xcite with some adjustments to the notations .
fix @xmath276 .
pick canonical hyperplane sections @xmath180 with constant fubini - study norm 1 . on the pullback line bundle @xmath277 with pullback
metric @xmath153 locally given by @xmath278 the pullback sections @xmath279 are of constant @xmath153-norm 1 . as in section 3
these initial data give rise to the norm function @xmath241 and the special gauge function @xmath195 that lead to the solution of vortex equations .
infinitesimally , given a tangent vector @xmath280 , there are corresponding infinitesimal variations of metric and gauge function , denoted by @xmath281 and @xmath282 .
moreover , there are @xmath58 sections @xmath283 , identified with @xmath284 via the pullback of the euler sequence on @xmath3 via @xmath175 . with these
, we identify the pushforward of @xmath284 via @xmath170 : @xmath285}\nu_{k+1,0}(s ) , \nonumber \\
\label{pushforward definition}\end{aligned}\ ] ] where @xmath286 is the initial tangent vector of the connection form induced by @xmath284 . given @xmath287 , the pullback metric @xmath288 at @xmath175
are then determined by an @xmath289 matrix of smooth functions given by @xmath290 a fact implicitly used in this formula is that the sections @xmath157 and the metric @xmath153 are chosen so that the sum of the @xmath153 norms of @xmath157 s are constant , and therefore @xmath291 .
recall that @xmath292 , where @xmath231 is the solution to @xmath293 and the background sections are chosen so that @xmath294 , can be rewritten as @xmath295 the pointwise definition of @xmath288 extends naturally to a smooth function near @xmath175 .
take an open neighborhood @xmath296 near @xmath175 with local coordinate @xmath297 centered at @xmath175 as described in .
let @xmath298 be the corresponding frame of @xmath299 over @xmath296 .
each holomorphic function @xmath300 yields a unique set of background metric , sections , and gauge function depending smoothly on variation of holomorphic functions . the functions @xmath301 and @xmath195 in the integrand of are therefore understood to be smooth functions defined on @xmath302 , and we amend these notations to @xmath303 and @xmath304 to reflect their dependencies on @xmath305 .
the formula then defines @xmath289 smooth functions on @xmath296 : @xmath306\overline{\left[\tilde{a}^\beta+\frac{{\partial}\tilde{\varphi}_s^\beta}{{\partial}z}+2\frac{{\partial}\tilde{\psi}^\beta}{{\partial}z}\right]}}{2s^2 } dvol_\sigma \nonumber \\ & & + \int_\sigma \left[\left<\phi^\alpha,\phi^\beta\right>_h \left(-he^{\tilde{\varphi}_s}\right)-\left(he^{\tilde{\varphi}_s}\right)^\alpha \tilde{u}_s^\beta \right]dvol_\sigma .
\label{l 2 metric pull back coordinate}\end{aligned}\ ] ] the proof of theorem [ main theorem on convergence of gauge ] is then mimicked to show that the first and third terms in the integrand of converge to zero and @xmath307 smoothly as @xmath219 , at all points on @xmath296 . to this end , we have established that @xmath288 is locally represented by a finite collection of smooth functions , each converging smoothly , as functions on @xmath296 , to the finite collection of smooth functions representing the ordinary @xmath0 metrics . to establish the convergence of volumes
, we must show that the volume forms induced by each @xmath288 converses to the volume form induced by the @xmath0 metric in @xmath308 as @xmath219 .
an obvious challenge comes from the fact that singularities for @xmath234 may develop near the boundary of the non - compact set @xmath296 corresponding to vortices with common zeros . in such situations , pointwise convergence of volume forms
does not necessarily imply that integrals , namely the volume , converge to the integral of the limiting volume form .
however , the next section assures that the convergence in integral continues to hold .
in this section we establish the main theorem of this article .
the theorem follows from several technical computations .
since the spaces discussed in the main theorem are open , where boundedness conditions required for the convergence of integrals are more difficult to achieve , we wish to establish the result on the compactifications of these spaces , noting that compactifications do not affect the @xmath0 volumes .
let @xmath311 be the class of vortices with common zeros a divisor of degree @xmath263 , which corresponds to the @xmath312 stratification above .
lemma [ dimension of hol ] then implies that @xmath311 is of dimension and therefore @xmath311 is of strictly lower dimension than @xmath166 for all @xmath314 .
the arguments are valid for all @xmath107 large enough , including @xmath264 , and the proposition is therefore established . . therefore , we may compute the volume of the compactified spaces , where @xmath317 convergence is more feasible . to discuss convergence on the compactified domain ,
we first construct smooth extensions of the basic functions and forms that give rise to metric functions @xmath318 . throughout the entire constructions below ,
we reserve @xmath319 as indices for coordinates of the finite dimensional manifold @xmath119 @xmath320 described in .
moreover , we adopt the notation `` @xmath321 '' to denote the derivative , or the infinitesimal variation of an object @xmath52 , induced from the @xmath322 component of the local frame of @xmath323 . evidently , @xmath332 for some @xmath263 . for each @xmath87 , let @xmath333 .
the tuple @xmath334 then have no point in common and still satisfies the defining equations for @xmath119 ( and @xmath6 . )
it then defines a holomorphic map @xmath335 of degree @xmath118 as in proposition [ description of based holomorphic maps ] .
we then consider the pullback of fubini - study metric via @xmath336 with the locally defined function : @xmath343 is clearly smooth and we claim that it is a smooth extension of @xmath324 . indeed , for @xmath344 , pick a neighborhood @xmath345 of @xmath4 so that @xmath346 . on @xmath347 , we consider tuples of divisors the third equality is true because @xmath351 and therefore @xmath352 .
the third line in is defined on the entire @xmath345 , which also agree with the curvature given by the extended metric function @xmath339 when @xmath353 points of @xmath354 converge to @xmath355 , and the proof is complete .
one notes that in the second line of the definition of @xmath360 , the fraction is in fact a well defined smooth function on @xmath119 , as accumulations of common zeros appear simultaneously on the numerator and denominator with the same order . for the exponential term
, we observe that since the curvature form can be smoothly extended to @xmath119 , so can @xmath358 . indeed
, the solution to the laplacian equation propositions [ extension of curvature ] and [ extension of norm function ] allow us to establish the convergence of metric functions of @xmath288 to @xmath362 in all @xmath363 .
for each @xmath364 we re - examine the formula of @xmath365 : to ensure the convergence in @xmath363 , we need appropriate bounds to apply dominated convergence theorem .
the bound is obtained from uniform estimates of @xmath378 with the approximated solutions to kazdan - warner equations , constructed in @xcite . by proposition [ extension of norm function ] ,
@xmath383 and @xmath384 are smooth on @xmath327 .
@xmath385 is constant on each connected component of @xmath119 and may be treated as constant when differentiating with respect to any variable .
these functions provide natural bounds for @xmath386 .
indeed , one observes that where @xmath389 is a uniform constant over @xmath119 .
note that we have normalized the khler form of @xmath4 so that @xmath390
. the difference between @xmath378 and @xmath391 is estimated uniformly using the maximum principal of @xmath197 . for each @xmath107 and @xmath392 , since @xmath393 is smooth on the compact set @xmath4 , we may pick @xmath394 so that where @xmath407 is again a uniform constant .
this is possible since @xmath408 and for @xmath107 large enough , @xmath391 and @xmath384 are smooth and uniformly bounded on @xmath409 .
we now possess sufficient estimates for the @xmath317 bound
. for all @xmath410 , depends smoothly on variation of holomorphic maps and locations of zeros , and is uniformly bounded on @xmath119 .
it is now clear that @xmath412 are uniformly bounded on @xmath119 , and therefore an @xmath413 function on @xmath119 .
dominated convergence theorem then applies to yield the conclusion of the proposition .
we begin by differentiating kazdan - warner equation as well as the approximated equation with respect to the @xmath415 coordinate of @xmath119 .
the @xmath416-differentiation clearly commutes with @xmath197 since they are defined on different spaces .
we obtain the following linearizations : one can readily verify that for @xmath107 large enough , the fact that @xmath360 is a smooth negative function on @xmath327 implies that both @xmath421 and @xmath422 are @xmath423 times functions that are uniformly bounded on @xmath327 , and therefore uniformly converge to 0 on @xmath327 as @xmath219 .
we then repeat the arguments of maximum principle as in the proof of proposition [ l p convergence of z ] to the difference of equations and to obtain uniformly decaying bounds for @xmath424 and @xmath421 .
we find since @xmath426 is smooth , @xmath427 is uniformly bounded on @xmath327 . by , we conclude that the fraction in the right hand side of is uniformly bounded .
since @xmath422 decays to 0 uniformly on @xmath327 as @xmath373 , so does @xmath428 , which implies that @xmath429 uniformly as @xmath373 since @xmath421 does .
since @xmath324 is smooth on @xmath327 , elliptic regularity ensures that @xmath432 is smooth and therefore uniformly bounded , and so is @xmath434 . combining this fact with equation ( 4.28 ) in @xcite , namely that @xmath435 pointwise as @xmath219
we have thus shown that @xmath436 as @xmath219 pointwise on @xmath119 .
it remains to construct a uniform bound to apply the dominated convergence theorem to establish convergence in @xmath363 , which follows clearly from straightforward computations : uniformly on @xmath327 as @xmath219 . since every term in the limit is uniformly bounded over @xmath107 ,
so is @xmath438 .
it follows that @xmath439 are uniformly bounded on @xmath119 and dominated convergence theorem is applicable . by lemma 4.4 in @xcite and the smooth extension of @xmath360 , @xmath442 and @xmath443
are uniformly bounded . @xmath444 and @xmath445 are smooth on @xmath327 , independent of @xmath107 , and therefore uniformly bounded .
we need to , however , ensure that the integrals @xmath454^\alpha \wedge \left[d''\left(\log \tilde{h}\right)\right]^\beta \nonumber \\ & = & \int_\sigma d'\left(\log \tilde{h}\right)^\alpha \wedge d''\left(\log \tilde{h}\right)^\beta \nonumber \\ & = & -\int_\sigma \left(\log \tilde{h}\right)^\alpha d'd''\left(\log \tilde{h}\right)^\beta \nonumber \\ & = & -\int_\sigma \left(\log \tilde{h}\right)^\alpha \tilde{f}^\beta , \nonumber \\
\label{l 2 norm of initial connection forms}\end{aligned}\ ] ] where @xmath455 is the @xmath416-derivative of the curvature ( 1,1)-form of @xmath456 .
the indices @xmath370 commute with integration and exterior derivatives since they are operators defined independently on different spaces .
the third equality follows from stoke s theorem .
we observe that @xmath455 extends smoothly to a two form on @xmath4 depending smoothly on @xmath119 , and therefore @xmath457 is a uniformly bounded function on @xmath119 .
@xmath458 , however , might blow up near the boundary of @xmath6 .
nevertheless , we claim that the function are precisely the common zeros of the @xmath461 s that develop near the boundary of @xmath119 . in the coordinate description of @xmath119 given in , it corresponds to condition that for some @xmath462 , every @xmath461 vanishes at @xmath463 with order @xmath464 .
let @xmath465 be the minimum of these orders .
fix a open neighborhood @xmath466 around @xmath463 such that for all @xmath467 , @xmath468 , where @xmath469 is a nonvanishing smooth function on @xmath345 .
we then rewrite which is integrable over @xmath345 , using polar coordinate , and the integral depends smoothly on @xmath463 , the location of common zeros for @xmath461 s .
we conclude that there is a constant @xmath474 such that one might notices that the extension of the metrics and curvature forms over the boundary of @xmath6 lowers the topological degree of the pullback line bundle @xmath489 .
in fact , we have experimented the bubbling phenomenon in the elementary settings .
the author is eager to explore generalization of this result to more general cases , such as the one posed in @xcite . | we prove the conjectural formula for the @xmath0 volume of the space of degree @xmath1 holomorphic maps from a compact khler riemann surface of genus @xmath2 to @xmath3 .
this formula was posed in @xcite and rigorously verified in @xcite for a special case using independent techniques . |
the cdf experiment has recently discovered an anomaly in the production rate of lepton - photon-@xmath0 in @xmath4 collisions at @xmath5 , using 86.34 pb@xmath6 of tevatron 1994 - 95 data @xcite . while the number of events expected from the standard model ( sm ) were @xmath7 , the experimentally measured number corresponded to 16 .
moreover , 11 of these events involved muons ( with 4.2 @xmath8 0.5 expected ) and 5 electrons ( with 3.4 @xmath8 0.3 expected ) . in earlier papers
@xcite we suggested that the excess can be simply understood in terms of the minimal supersymmetric standard model ( mssm ) which has the following ingredients : ( 1 ) the model is @xmath9-parity violating with an @xmath10-violating @xmath11 coupling , and ( 2 ) the supersymmetric spectrum includes an ultra - light gravitino of mass @xmath12 ev .
we have demonstrated in our earlier papers @xcite that such a model provides a natural explanation of the cdf anomaly and explains not only the excess in the number of @xmath13 events but also explains the main features of the kinematic distributions of the excess events .
the excess can be explained using a small value of the @xmath10-violating coupling @xmath14 because of the resonant production of smuons in the annihilation of an initial - state @xmath15 pair .
the smuon thus produced decays predominantly into a bino - dominated neutralino and a muon , with the neutralino further decaying into a photon and a gravitino .
the production and decay has been shown in the feynman diagram in fig .
the fact that the excess is seen in final states involving photons emerges very neatly in our model because the decay @xmath16 dominates overwhelmingly over other decay modes .
( 400,170 ) ( 60,10)(120,50 ) ( 120,50)(60,90 ) ( 120,50)(180,50)5 ( 180,50)(240,10 ) ( 180,50)(240,90)5 ( 240,90)(300,70)5 ( 240,90)(300,130)54 ( 65,5)@xmath17 ( 65,90)@xmath18 ( 245,10)@xmath19 ( 210,80)@xmath20 ( 150,65)@xmath21 ( 305,65)@xmath22 ( 305,125)@xmath23 in the current article , we extend the previous studies by including two additional pieces of independent empirical information .
we include the d0 run i measurement @xcite of the @xmath24 missing @xmath25 process , as well as @xmath23 missing @xmath25 data coming from cdf in run i @xcite .
the empirical background event - rate in the @xmath26 channel is quite different in the cdf and d0 cases due to the different cuts employed .
our scenario predicts excesses in each of these channels , and we determine to what extent it is in accord with their measured rates . by performing a combined fit to all three event rates ,
we constrain the masses of the relevant sparticles in the event , as well as @xmath11 .
in order that the cross - section for the production of the smuon resonance is substantial enough to account for the anomalous events , we need the left - handed smuon to be light i.e. @xmath27 gev .
further , one needs to couple the smuon to valence quarks in the initial state implying that the @xmath10-violating operator that we need is @xmath28 corresponding to the coupling @xmath11 , which generates the interactions @xmath29 and @xmath30 ( and charge conjugates ) , along with other supersymmetrised copies involving squarks .
therefore , the operator we invoke in our model predicts supersymmetric signals in other channels which manifest themselves through the production of either sneutrinos or squarks . in our model , we take the squarks to be heavy and so their effects on experimental observables will be negligible .
on the other hand , the spontaneously broken su(2)@xmath31 symmetry in the mssm implies that muon sneutrinos have a tree - level mass squared @xcite @xmath32 where @xmath33 is the mass of the muon , @xmath34 the weinberg angle , and @xmath35 the ratio of the mssm higgs vacuum expectation values ( @xmath36 ) . since @xmath37
if we assumed that smuons were produced at the tevatron run i energy , we should expect that muon sneutrinos could also be produced .
the dominant production mechanism is resonant sneutrino production and subsequent decay , as shown in fig .
[ feyn2 ] .
it results in a @xmath3 experimental signature .
( 400,170 ) ( 60,10)(120,50 ) ( 120,50)(60,90 ) ( 120,50)(180,50)5 ( 180,50)(240,10 ) ( 180,50)(240,90)5 ( 240,90)(300,70)5 ( 240,90)(300,130)54 ( 65,5)@xmath17 ( 65,90)@xmath18 ( 245,10)@xmath38 ( 210,80)@xmath20 ( 150,65)@xmath39 ( 305,65)@xmath22 ( 305,125)@xmath23 the pattern of masses of the super - particles suggested by the cdf @xmath40 anomaly is the following : the smuon is around 150 gev , and the only other light sparticles are the neutralino ( which is typically about 45 gev lighter than the smuon ) and the ultra - light gravitino ( which is as light as @xmath41 ev ) .
we enforce degeneracy between the first two generations in order to avoid flavour changing neutral currents .
other sparticles do not play a role in this analysis , and are set to be arbitrarily heavy .
such a light gravitino materialises naturally in models of gauge - mediated supersymmetry breaking ( gmsb)@xcite . however , in the minimal version of gmsb models the chargino and the second - lightest neutralino are not much heavier than the neutralino and this feature of the minimal model is certainly not desirable for our considerations because it leads to large jets@xmath42 rates which are not seen by experiments cdf event @xcite . ] .
if the minimal versions of gmsb models do not yield the pattern of super - particle masses that we need , then the obvious question to ask is what is the high - energy model that yields this mass spectrum at low energies .
it is interesting to note that such a mass spectrum can arise in an alternate model of gmsb which is obtained from compactifying 11-dimensional m - theory on a 7-manifold of g2 holonomy @xcite . for the purposes of this paper , however ,
we simply work with the low - energy model with the mass spectrum described above and do not worry about the high energy completion of this model .
in our model , we have essentially four free parameters that are relevant to the data we fit : the gravitino mass , @xmath43 , the neutralino mass @xmath44 , the smuon mass , @xmath45 and the @xmath9-violating coupling @xmath11 .
however , instead of simultaneously fitting the four parameters using the experimental data , we choose to work with fixed values of two of these parameters close to their best - fit values while performing fits in the two other parameters . for our analysis
, we take other parameters like @xmath46 to be constant .
the coupling , @xmath11 , is constrained from @xmath47 @xcite to be @xmath48 @xcite .
but since the constraint involves a squark mass which is large in our model , it is not very relevant .
while the production of the smuon resonance is through the @xmath9-violating mode , to produce the @xmath49 excess we require that its decay goes through the @xmath9-conserving channel to a neutralino and muon final state .
the @xmath9-violating decay of the slepton is possible but constrained , in principle , by the tevatron di - jet data @xcite which exclude a @xmath50 pb at 95% c.l . for a resonance mass of 200 gev .
however , in practice this does not provide a restrictive bound upon our scenario as long as the @xmath9-violating coupling is sufficiently small , @xmath51 .
we also add that the di - jet bound is not restrictive because it suffers from a huge qcd background . by restricting @xmath11 to be small
, we also avoid significant rates for the possible @xmath9-violating decays of @xmath52 or @xmath53 .
we use the isasusy part of the isajet7.58 package @xcite to generate the spectrum , branching ratios and decays of the sparticles . for an example of parameters , we choose ( in the notation used by ref .
@xcite ) @xmath54 and @xmath55 .
@xmath19 together with other flavour diagonal soft supersymmetry breaking parameters are set to be very large .
we emphasise that this is a representative point in the supersymmetric parameter space and not a special choice .
as stated earlier , we present analyses for three sets of data in this paper : the cdf run i data on @xmath56 @xcite , the d0 run i data on @xmath13 @xcite and the cdf run i data on @xmath3 @xcite .
we now present our fiducial efficiencies and cuts , which mimic those of the relevant experiments .
the cdf experiment detects photons with the constraints that the following pseudo - rapidity regions are excluded : @xmath57 , @xmath58 and the region @xmath59 . for the @xmath56 data ,
photon detection efficiency within these cuts is @xmath60 .
muons have a 60@xmath61 detection efficiency if @xmath62 or 45@xmath61 if @xmath63 .
to improve signal to background ratio efficiency , the following cuts are used : @xmath64 gev , @xmath65 gev and @xmath66 gev .
.photon efficiency in the cdf @xmath3 analysis [ cols="^,^,^,^,^",options="header " , ] the observed number of events in the different analyses , and their standard model backgrounds ( taken from the experimental papers @xcite ) are shown in table [ expnums ] .
we now simulate the signal events for the processes in figs .
[ feyn],[feyn2 ] .
we use herwig6.4 @xcite including parton showering ( but not including jet isolation cuts ) to calculate cross - sections for single slepton production .
for each of the three data listed in table [ expnums ] , we can define a log likelihood defined by the poissonian log - likelihood for @xmath67 observed events , @xmath68 expected signal events ( for fixed values of susy breaking parameters ) , convoluted with a gaussian probability distribution of the number of expected standard model background events @xmath69 and its uncertainty @xmath70 : @xmath71 this is a good approximation provided @xmath69 is several times @xmath70 above zero , as is the case here .
@xmath72 is then the ( signal+background ) to background likelihood ratio for a given analysis .
we can form the total fit likelihood ratio @xmath73 by adding the likelihood ratio from each of the three analyses in table [ expnums ] .
we will always consider two relevant parameters to be fixed and fit the model to the other two .
the best - fit point in parameter space corresponds to the maximum value of the likelihood ratio and is found by using minuit @xcite .
@xmath74 corresponds to the equivalent number of @xmath75(standard model ) - @xmath75(best - fit)@xmath76 .
we have one degree of freedom , and therefore the 90@xmath61 and 95@xmath61 confidence level ( @xmath77 ) limits on parameters then lead to @xcite @xmath78 respectively in the fit .
we will constrain @xmath79 gev , as implied by @xmath80 lep2 data @xcite .
we now discuss the results of combined fits for various hyper planes of parameter space .
( 5.4,5 ) ( 1.35,0 ) ( 0,2.5 ) ( 2.7,2.5 ) ( 1.35,2.3)(c ) ( 0,4.8)(a ) ( 2.7,4.8)(b ) fig .
[ lamm1 ] displays the @xmath81 and @xmath82 fit regions as the area between the solid lines in the @xmath83 plane .
@xmath84 has been kept fixed at 45 gev in order to keep the decay mode @xmath85 open and @xmath86 ev .
a significant amount of parameter space fits the combined data , with ranges @xmath87 .
increasing @xmath88 produces a lower cross - section because of kinematical suppression , but this effect can be compensated by raising @xmath11 , thus increasing the production rate . the best - fit point is @xmath89 , @xmath90 gev , with @xmath91 .
[ lamm1]a shows that we expect between 3 - 8 signal events in the cdf @xmath56 channel .
this data dominates the fit because backgrounds ( and their uncertainties ) are larger in the other analyses .
1.5 - 6 events are expected at d0 , and between 1.5 and 8 signal events were predicted for the cdf @xmath3 signature , depending on the parameters .
( 5.4,5 ) ( 1.35,0 ) ( 0,2.5 ) ( 2.7,2.5 ) ( 1.35,2.3)(c ) ( 0,4.8)(a ) ( 2.7,4.8)(b ) fig .
[ mgm1 ] displays the @xmath81 and @xmath82 fit regions as the area to the left - hand side of the solid lines in the @xmath92 plane .
here , we use the default values @xmath93 gev .
it is clear from the fit regions that the data prefer lower values of the neutralino mass .
the best - fit point is : @xmath94 gev , @xmath95 ev , @xmath96 .
the figure illustrates that if @xmath97 ev , the branching ratio of @xmath98 becomes tiny , decreasing the cross - section for the cdf @xmath56 signal , which dominates the fit .
when @xmath43 is below @xmath99 ev , @xmath100 decays dominate , again decreasing the cdf @xmath13 signal .
2 - 6 cdf @xmath56 excess events , 2 - 4 d0 excess events and 0 - 3 @xmath3 cdf excess events are expected , shown by the dashed contours within the 90@xmath101 regions of figs .
[ mgm1](a - c ) .
( 5.4,5 ) ( 1.35,0 ) ( 0,2.5 ) ( 2.7,2.5 ) ( 1.35,2.3)(c ) ( 0,4.8)(a ) ( 2.7,4.8)(b ) fig .
[ lamsmu ] displays the @xmath81 and @xmath82 fit regions as the area between the solid lines in the @xmath102 plane .
here we have chosen default values of @xmath86 ev and @xmath90 gev .
a significant amount of parameter space fits the combined data .
the best - fit point is @xmath103 , @xmath104 gev , with @xmath91 .
[ lamsmu]a shows that we expect between 1 - 12 signal events in the cdf @xmath56 channel .
1 - 8 events are expected at d0 , and up to 4 signal events were predicted for the cdf @xmath3 signature , depending on the parameters .
( 5.4,5 ) ( 1.35,0 ) ( 0,2.5 ) ( 2.7,2.5 ) ( 1.35,2.3)(c ) ( 0,4.8)(a ) ( 2.7,4.8)(b ) fig .
[ mgsmu ] displays the @xmath81 and @xmath82 fit regions as the area enclosed by the solid lines in the @xmath105 plane .
here we have chosen default values of @xmath106 and @xmath90 gev .
the ranges @xmath107 gev , @xmath108 ev provide a reasonable combined fit .
the best - fit point is @xmath109 ev , @xmath110 gev , with @xmath111 .
[ mgsmu]a shows that we expect between 1 - 6 signal events in the cdf @xmath56 channel .
0 - 4 events are expected at d0 , and up to 0.2 signal events were predicted for the cdf @xmath3 signature , depending on the parameters .
we have provided combined fits for a supersymmetric model that explains the @xmath56 cdf run i excess in events , which was at the 2.7@xmath1 level @xcite .
we have used the run i @xmath112 data recently presented by cdf , as well as anomalous trilinear gauge boson coupling data from d0
. constraints upon various hyper planes in the gravitino , smuon , neutralino and r - parity violating coupling space have been displayed . in totality , the signal rates predicted by our model for the three analyses fit the data well , best fit points corresponding to a @xmath113 fit compared to the standard model , for one degree of freedom . unfortunately , background rates in the d0 anomalous trilinear gauge boson coupling data are too high for it to be very sensitive to the predicted signal rate .
we note , however , that another analysis on existing d0 run i data with cuts optimised to test a @xmath13 excess would provide a good test of our scenario .
the cdf@xmath3 channel suffers from a high @xmath114 cut because of cosmic backgrounds , which unfortunately also cuts most of the signal .
it was shown in ref .
@xcite , that if the cut could be reduced to 25 gev , a signal rate of several times that in the cdf @xmath56 channel is possible .
this is an important observation for run ii , where cosmic backgrounds could be cut by additional timing information in the detector .
we look forward to the analysis of run ii data , which will be the final arbiter on @xmath56 excess , as well as on our scenario .
we would like to thank j. berryhill , h.j .
frisch , s. hagopian , d. toback and d. wood for help and advice regarding experimental analyses and data .
we would also like to thank cern , where much of this work was carried out .
s. ambrosanio , g. l. kane , g. d. kribs , s. p. martin , and s. mrenna , phys .
76 ( 1996 ) 3498 ; g. l. kane and s. mrenna , phys .
77 ( 1996 ) 3502 ; s. ambrosanio , g. l. kane , g. d. kribs , s. p. martin , and s. mrenna , phys .
d55 ( 1997 ) 1372 .
g. corcella et al , hep - ph/0201201 ; g. marchesini , b. r. webber , g. abbiendi , i. g. knowles , m. h. seymour and l. stanco , jhep 01 ( 2001 ) 010 hep - ph/0011363 ; _ ibid . _
hep - ph/0107071 .
`` herwig : a monte carlo event generator for simulating hadron emission reactions with interfering gluons .
version 5.1 - april 1991 '' , comput . phys.commun .
67 ( 1992 ) 465 . | the cdf experiment reported a lepton photon missing transverse energy ( @xmath0 ) signal 3@xmath1 in excess of the standard model prediction in tevatron run i data .
the excess can be explained by the resonant production of a smuon , which subsequently decays to a muon , a photon and a gravitino .
here , we perform combined fits of this model to the cdf @xmath2 excess , the d0 measurement of the same channel and the cdf @xmath3 channel . although the rates of the latter two analyses are in agreement with the standard model prediction , our model is in good agreement with these data because their signal to background efficiency is low at the best - fit point .
however , they help to constrain the model away from the best fit point . , |
when two equilibrium systems in contact with reservoirs at different temperatures @xmath0 and @xmath1 are put in contact , heat flows from the hotter to the colder and the usual equilibrium concept of temperature can be used to establish the average direction of heat fluxes .
deviations from the average are described by a fluctuation principle known as the gallavotti - cohen relation @xcite , namely = q_1,2 ( - ) , [ fluc ] where @xmath2 is the probability that the first system exchanges the heat @xmath3 in a certain ( sufficiently long ) time with the other one and we have set to unity the boltzmann constant . in an attempt to generalize ideas of equilibrium thermodynamics to non - equilibrium systems , an _
effective temperature _ @xmath4 has been introduced in @xcite in the context of non - equilibrium stationary systems and in @xcite for aging systems . in this formulation
, @xmath4 can be evinced from the relation r(t , t_w)=t_eff(t , t_w)^-1 [ efftemp ] between the auto - correlation function @xmath5 of the observable @xmath6 at times @xmath7 and @xmath8 and the associated linear response function , @xmath9 , where @xmath10 is a small perturbation introducing an extra term @xmath11 in the hamiltonian . in equilibrium states , due to the fluctuation - dissipation theorem
, @xmath12 is the usual temperature which is independent of the two times and on the chosen observable @xmath6 . out of equilibrium , in principle
@xmath4 may depend on @xmath13 and @xmath6 and is not necessarily related to the temperature of the reservoir . in @xcite
it was argued that @xmath4 exhibits the properties of a temperature , in the sense that it regulates the direction of heat fluxes favouring thermalization .
specifically , if different degrees of freedom _ effectively _ interact on a given time scale , heat flows from those with higher @xmath4 to those with a lower one .
the concept of effective temperature is possibly relevant when thermal flows are small ; in the context of aging systems this usually amounts to consider the large time regime . generally , in this asymptotic domain one can identify fast and slow degrees of freedom whose typical timescales @xmath14 and @xmath15 become widely separated . indeed , while @xmath16 is generally age - independent , @xmath17 is usually an increasing function of the age @xmath18 of the sample .
when this happens , one can probe fast or slow degrees increasing the times @xmath13 along particular directions in the @xmath13 plane . specifically , by letting @xmath19 keeping @xmath20 finite , since @xmath21 the slow degrees act adiabatically and can be considered as static . in this time sector , denoted as the short time difference ( or quasi - equilibrium ) regime , one probes the dynamical behavior of fast modes .
conversely , letting again @xmath19 but keeping @xmath22 ( or some different combination of @xmath7 and @xmath18 in specific systems ) finite one has @xmath23 , focusing the analysis on the timescale of the slow ( or aging ) components . in this time sector ,
usually denoted as aging regime , the fast degrees act as a stationary background and the time evolution of the slow modes is probed . in a class of aging systems , among which mean field p - spin models and coarsening systems @xcite , the effective temperature ( here and in the following we simply quote the name for the quantity @xmath4 defined through eq . ( [ efftemp ] ) , without necessarily conforming to the debated interpretation of it as a thermodynamic temperature ) takes two different well defined values in the short time and in the aging regime . in the former sector
one has @xmath12 , the bath temperature , while in the aging regime @xmath24 .
a major difference between mean field glassy models and phase - ordering systems is that @xmath25 is finite in the former , while it is infinite in the latter .
this simple two - temperature scenario , as opposed to that of a continuously varying @xmath4 displayed by models with full replica symmetry breaking , such as the sherrington - kirkpatrik spin glass , or some systems without local detailed balance @xcite , is amenable of a simple interpretation in terms of fast and slow components : the former which are already at the equilibrium temperature and the latter retaining a self - generated higher temperature @xmath25 .
generally , the existence of different effective temperatures in the asymptotic domain implies that thermalization among different degrees of freedom is suppressed .
while on the whole the mechanism whereby thermalization is avoided is uncertain , in the case of coarsening systems , at least in the solvable large-@xmath26 model ( @xmath26 being the number of components of the vector order parameter ) fast and slow degrees can be explicitly exhibited and their statistical independence can be proved @xcite .
this is consistent with two temperatures being sustained and coexist in the system .
however , the issue of heat exchanges occurring between fast and slow modes has never carefully been analyzed so far . in this paper
we tackle this question numerically in the physical case of coarsening systems with @xmath27 .
we check to some extent the picture with slow and fast degrees , studying the heat exchanges between them , and suggesting a mechanism whereby statistical independence is preserved and effective thermalization does not occur . in particular
, we will consider the two - dimensional ising model quenched below @xmath28 .
this is perhaps the simplest case where the two - degrees / two - temperature scenario can be investigated .
we mention that the ising model in @xmath29 exhibits an anomalous coarsening behavior with a continuously varying @xmath4 @xcite which , moreover , is found to depend on the particular observables @xmath6 entering @xmath30 and @xmath31 in the definition .
these features , which are presumably related to the fact that the model is at the lower critical dimensionality @xmath32 @xcite , make the interpretation of @xmath4 as a genuine temperature doubtful . however , for @xmath33 these features disappear and the two - temperature scenario is more sound .
the paper is organized in five sections . in sec .
[ single ] we review the kinetics of binary systems in contact with a single thermal bath . in sec .
[ two ] we consider a system in contact with two reservoirs and introduce the method to compute heat exchanges . in sec .
[ numsetup ] we describe our numerical experiments to study heat flows in phase - ordering systems . in sec .
[ conclusions ] we draw the conclusions and discuss the perspective of this work .
the dynamics of a non - disordered binary system ( i.e. one that can be described by the ferromagnetic ising model ) in contact with a thermal bath , has been extensively studied since a long time .
although exact results are scarce , the overall phenomenology is nowadays quite well understood @xcite .
hereafter we recall some basic facts that will be needed in the following and specify our dynamical model , that will be later ( in sec .
[ two ] ) generalized to the case of a system in contact with two reservoirs . in the high temperature disordered phase , for @xmath34 ,
the system is ergodic and equilibrium is achieved on microscopic timescales ( if not too close to @xmath28 ) starting from any configuration . on the other hand ,
the evolution below @xmath28 depends strongly on the initial state .
in particular , if the sample is prepared in a broken symmetry configuration ( for instance , with all spins up ) the dynamics converges on a microscopic time to the broken symmetry equilibrium state .
this is characterized by finite coherence length @xmath35 and relaxation time @xmath36 , where @xmath37 is the usual equilibrium dynamical exponent . @xmath38 and
@xmath39 can be interpreted as the spatial extent and the persistence of thermal islands of reversed spins in the sea of opposite magnetization .
these fluctuations will be denoted as _
fast _ , because @xmath39 is finite ( and small if @xmath40 is sufficiently far from @xmath28 ) .
conversely , starting from a typical high temperature equilibrium configuration , namely a disordered one , and quenching to a sub - critical temperature , a coarsening dynamics sets in where domains of the two magnetized phases with a typical size @xmath41 grow in time . for the case of a dynamics with non - conserved order parameter considered in this paper one
has @xmath42 , with @xmath43 , for sufficiently long times .
if an interface is present at time @xmath18 in some part of the system , it must move a distance of order @xmath44 in order to produce an effective decorrelation . since this takes a time of order @xmath45 , the characteristic decorrelation time due to interface motion is @xmath46 .
this is the essence of the aging phenomenon , where the evolution gets slower as time elapses and @xmath15 diverges .
the interfacial degrees of freedom responsible for the aging process will be denoted as _
slow_. however , these are not the only degrees contributing to the dynamics . indeed ,
the interior of domains is basically in equilibrium in one of the broken symmetry phases described above .
it is then characterized by the fast thermal fluctuations with characteristic timescale @xmath47 . in conclusion , in the late stage evolution of a coarsening system there are two dynamical components , fast and slow , whose timescales are widely separated in the large @xmath18 limit .
in this paper we will consider the two - dimensional ising model , where @xmath26 discrete spin variables @xmath48 defined on the sites @xmath49 of a square lattice are governed by the hamiltonian @xmath50 . here
@xmath51 denotes the whole spin configuration and the sum is running over all nearest neighbours couples @xmath52 .
the critical temperature of the model is @xmath53^{-1}\simeq 2.269 \ , j$ ] . in the following
we will set @xmath54 .
we introduce a dynamics where a single spin update is attempted in each timestep ; time will be measured in montecarlo steps ( or sweeps ) , corresponding to @xmath26 timesteps .
metropolis single spin transition rates w()=min , [ metrop ] where @xmath55 is the energy change due to the flip , will be used . in order to study the evolution ,
let us consider the energy ( per spin ) @xmath56 whose equilibrium value is @xmath57 . here
@xmath58 means an average over thermal noise and initial conditions .
since the excess energy @xmath59 is associated to the domain boundaries , in the coarsening stage one has e(t)l(t)^-1t^-1/2 .
[ eq : e ] this behavior lasts until @xmath41 becomes comparable with the size of the system and the system either equilibrates or remains trapped ( particularly at low temperatures ) in long lived metastable states with one ( or few ) spanning interfaces . if the thermodynamic limit is taken from the beginning the system is free from finite size effects and coarsening lasts forever : equilibration is then never observed
. snapshots of the configurations of the system after a quench from a disordered state to a subcritical temperature @xmath60 are shown in fig .
( [ figsnap ] ) , where the growing domains are visible .
one can also see small regions of reversed spins inside the growing domains .
these are the fast degrees of freedom , with the equilibrium behavior discussed above . in fig .
( [ figene ] ) the time behavior of the excess energy @xmath61 is shown .
in the case of a quench to @xmath62 , after an initial transient , it starts decreasing with the expected power - law behavior up to a certain time where finite - size effects set in and saturation occurs . increasing @xmath26 the finite - size effect is delayed , as expected .
the same behaviour is observed in a quench to @xmath63 , the main difference being a value of @xmath64 slightly larger than the expected one @xmath43 due to preasymptotic corrections , as discussed in @xcite .
we consider now the case of two coupled binary systems ( denoted simply as 1 and 2 ) interacting with thermostats @xmath0 and @xmath1 at different temperatures @xmath65 ( for simplicity we use the same symbols for the temperature of the reservoirs and the reservoirs themselves ) .
we denote with @xmath66 ( @xmath67 ) the heat flowing in a time interval @xmath68 $ ] from the system 1 ( 2 ) to the thermostat @xmath0 ( @xmath1 ) , and with @xmath3 that passing from 1 to 2 . when @xmath69 , starting from any initial condition , the system attains a non - equilibrium stationary state after a certain ( microscopic ) time .
the same behavior is observed if one ( or both ) the temperatures are below @xmath28 , provided that the system(s ) coupled to a sub - critical temperature is prepared in an initial state with broken symmetry , in order to avoid the coarsening dynamics .
the statistics of the heats exchanged between the systems and the thermostats has been investigated analytically by a mean field approach in @xcite and numerically in @xcite .
the case where one or both the systems are interested by phase - ordering kinetics has not been studied yet , and will be considered in sec .
[ numsetup ] .
this case poses the problem of a direct measurement of @xmath3 , as it is discussed below . in general the heat exchanged between two parts of a system or between a part of a system and a thermostat in the time interval @xmath70
can be written as q = _ s = t_w^t_w+_i=1^n(s ) q_i(s ) [ quantoq ] where @xmath71 is the heat exchanged by a single degree ( a spin , in the ising model ) in an unit time around time @xmath72 and @xmath73 is the number of degrees of freedom which can effectively exchange the heat .
( [ quantoq ] ) can be straightforwardly used for the computation of the heats @xmath66 , @xmath67 . in a numerical simulation
the @xmath71 are the energies @xmath55 released by the thermostats in the flip of the @xmath74 spin ( according to eq .
( [ metrop ] ) ) in the montecarlo step occurring at time @xmath72 .
when the system is stationary , a direct computation of @xmath3 can be avoided because for @xmath75 large enough its value can be related to @xmath66 and @xmath67 . indeed , denoting with @xmath76 the energy stored in the system 1 ( and analogously for 2 ) in the time interval @xmath70 ( @xmath77 )
, one has @xmath78 and @xmath79 .
combining these equations one has q_1,2=+ .
[ qdifff ] in a stationary state the quantities @xmath80 and @xmath81 of eq .
( [ quantoq ] ) fluctuate around a @xmath7-independent average value . hence @xmath82 and @xmath83 grow proportionally to @xmath84 . on the other hand
the boundary term @xmath85 is finite ( for a finite system ) .
hence , in the large-@xmath84 limit one can compute @xmath3 as q_1,2 .
[ qdiff ] therefore , in this case , it is sufficient to collect the statistics of the heats @xmath66 and @xmath67 exchanged with the thermostats and , from this , the distribution of @xmath3 can be determined through eq .
( [ qdiff ] ) , and eq .
( [ fluc ] ) can be tested .
this was made in @xcite with a numerical setup where a two - dimensional ising model on a @xmath86 x @xmath86 square lattice was divided into two interacting halves of size @xmath87 x @xmath86 ( system 1 and 2 ) , in contact with the two heat baths . in sec .
[ numsetup ] we will prepare the two systems as containing fast modes ( system 1 ) and a mixture of fast and slow ones ( system 2 ) .
heat fluxes between different degrees will then be inferred from the knowledge of @xmath3 .
however , in this case , since 2 is aging , it is no longer true that @xmath80 and @xmath81 fluctuate around time - independent values and , hence , eq . ( [ qdiff ] ) can not be used . for this reason we need another operative definition of the quantity @xmath3 .
this is the subject of the remaining of this section . in order to do that
, it is first useful to understand the basic mechanisms whereby heat is transferred at stationarity .
this can be better illustrated by considering first the case of two systems made by one single spin each .
we also assume , for the moment , that the two spins are updated alternately with metropolis transition rates . denoting with @xmath88 and @xmath89 the two possible states of system 1 ( and similarly for 2 ) ,
let us consider the case of an initial state @xmath90 which evolves in four steps as follows 1 _ , 2 _ 1 _ , 2 _ 1 _ , 2 _ 1 _ , 2 _ 1 _ , 2 _ [ scheme ] eventually returning to the initial state . in the initial state
the system is in the lowest energy state . since 1
is in contact with the higher temperature bath , it has a larger probability to be flipped first .
if this happens , as in the first step in the scheme ( [ scheme ] ) , the heat @xmath91 ( equal to the energy increase @xmath55 of the system ) is transferred from the reservoir @xmath0 to 1 .
now it is the turn of 2 to be updated . with metropolis rates it flips with probability one , because in doing so the energy decreases .
then the heat @xmath92 is released by 2 to @xmath1 ( step two ) . at this stage a heat @xmath93 has been transferred between @xmath0 and @xmath1 : hence we conclude that the amount @xmath94 has flown between the two spins . now
the system is again in the lowest energy state and , as before , the most probable evolution is the reversal of 1 , where a heat @xmath91 is again taken from the reservoir ( step three ) .
this move surely induces the flip of 2 ( step four ) with a heat @xmath95 flowing to @xmath1 and another amount @xmath94 exchanged between the two systems . in conclusion , if this process happens , a heat @xmath96 flows from 1 to 2 .
clearly , one can analogously imagine a process where @xmath97 , but this process implies that 2 flips first , which is less probable .
this explains why , on average , @xmath3 is positive , even if negative fluctuations ( which in stationary states are regulated by eq .
( [ fluc ] ) ) are possible .
the physical mechanism described above is such that heat is transferred from one system to the other whenever the flip of one spin triggers the flip of the other .
it is then quite natural to introduce a direct measure of the heat flow as @xmath98 , where @xmath99 is the energy variation of the system in the @xmath72-th step belonging to the interval @xmath100 , @xmath101 is a subset of steps where a flip has been triggered by a previous reversal of the other spin , and @xmath102 depending on the direction of the heat flow .
we state that the flip of ( say ) 2 in the @xmath103-th step has been triggered if 1 and 2 were not aligned before reversing 2 , and 1 was the last spin flipped before @xmath103 . in this case heat
is flowing from 1 to 2 and hence @xmath104 . in the opposite situation
in which 1 is flipping at time @xmath103 and 2 was flipped previously one has @xmath105 .
the same definition is meaningful if a different sequence of updates ( i.e. random ) is assumed and/or different transition rates are considered ( i.e. glauber ) .
the main difference in this case is that after the flip of 1 , spin 2 is not forced to flip and it might happen that 1 recovers again its original value .
however in this case there is no heat transfer between 1 and 2 , and this process does not alter the computation of @xmath3 .
let us now consider the case of systems with many degrees of freedom .
the mechanism described in ( [ scheme ] ) is at work between any couple of spins on the interface between 1 and 2 .
clearly , many other processes , involving also other spins away from the interface , may occur .
however , they do not overrule our algorithm , since there is no heat transfer between 1 and 2 until the process arrives on the boundary between them .
therefore , we can adopt the same operative definition for the heat transferred between any couple of spins on the interface , and @xmath3 is then obtained by summing all these contributions . in order to test this method , we have prepared a system with the geometry discussed below eq .
( [ qdiff ] ) in a stationary state , and we have computed @xmath3 with this technique and , independently , @xmath66 and @xmath67 . if the method is reliable , the determination of @xmath3 should verify eq .
( [ qdiff ] ) for large @xmath84 , so the probability distributions @xmath106 and @xmath107=q)$ ] should obey @xmath108 .
these distributions are compared in fig .
[ comparison ] . for small @xmath84
they are quite different .
this is expected due to the presence of the boundary term in eq .
( [ qdifff ] ) .
increasing @xmath84 , however , the boundary term becomes comparably smaller and the two curves tend to merge . to be more quantitative we plot , in the inset of fig .
[ comparison ] , the _ distance _ d_as= [ distance ] between the two probability densities , showing that it is a steadily decreasing function .
this shows that the algorithm for the direct measurement of @xmath3 is reliable .
when one ( or both ) of the systems is aging , eq .
( [ qdiff ] ) is spoiled by the extra heat released to the baths because the system is lowering its energy due to relaxation . in these cases
the direct computation of @xmath3 according to the technique described above is mandatory . in the following
, it will always be used to compute @xmath3 .
in this section we study the heat exchanges in phase - ordering systems .
we do this with the help of some numerical experiments designed to collect the statistics of the heat flowing between fast and slow degrees and among them and the reservoir .
this will allow us to test the scenario with two statistically independent degrees coexisting at different ( effective ) temperatures and to discuss why such independence may be retained in the evolution . in order to do
that , the straightforward procedure would be to consider an aging system , to recognize and discriminate ( at any time ) fast and slow components , and hence to detect the heat they are exchanging .
however , this is not easily realizable . indeed , although in a coarsening system we have an idea of what slow and fast modes are , a precise definition and separation of degrees is to a large extent arbitrary .
moreover , if even one could find a technique suited to do that , this would be tuned to phase - ordering and the method would be restricted to coarsening systems .
instead , we propose a different approach where two systems are coupled . preparing such systems so that the first is in equilibrium while the second coarsens , as will be detailed below , one may assume that slow degrees are confined in the second . as we shall discuss
, this will allow us to proceed to the discussion of heat exchanges . in doing that
, we adopt the same geometry as for the stationary case described in the previous section except that , in order to guarantee that the heat exchanged is small enough we have activated only two links , at distance @xmath87 , between the two systems . in all the cases we have chosen the parameters in such a way that the system is free from finite - size effects .
system 1 is prepared in equilibrium at the temperature @xmath0 while system 2 is in an infinite temperature disordered configuration . at time @xmath109
, sample 2 is brought in contact with the thermostat at the temperature @xmath110 while 1 interacts with a bath at the same temperature @xmath0 of its initial equilibrium state .
then , from time @xmath7 onwards 2 ages via phase - ordering kinetics , while 1 remains in a stationary configuration .
precisely , 1 is not stationary in principle , due to the coupling with the non - stationary system 2 , but this effect is negligible for large system sizes and few active links between the two sub - systems .
moreover the effect is further suppressed in the asymptotic time domain , due to the progressive decrease of the number of slow degrees in sample 2 .
indeed our simulations did not show any deviation from stationarity in 1 .
the statistics of the heats is collected from some time @xmath18 to @xmath111 in a sample of size @xmath112 .
notice that in the range of times considered in this simulation , represented by the dashed box in the right panel of fig .
[ figene ] , finite - size effects can be neglected . indicating with @xmath113 , @xmath114 the fast degrees of freedom of the two coupled systems , with @xmath115 the slow ones , and with @xmath116 the heat exchanged between @xmath113 and @xmath114 ( and similarly for the heat exchanged between fast and slow ones ) , the general scheme of all the possible heat exchanges is summarized in figure [ schema ] .
we start with the case in which the two baths are at the same temperature @xmath117 .
this will allow us to draw some conclusion on the possibility that , in a single coarsening system , heat is released directly from the slow modes to the fast ones .
the probability distributions of @xmath66 , @xmath67 and of @xmath3 have been computed letting @xmath118 fixed while @xmath18 is increased ( we used values of @xmath18 up to @xmath119 , for different choices of @xmath120 ) .
this will be referred to as the aging limit . for every choice of @xmath18 and @xmath120 we found a pattern of behaviors analogous to the one shown in fig .
[ a1 ] , where @xmath121 and @xmath122 . here
one sees that @xmath123 , because 2 is transferring the excess energy associated to the interfaces to the bath @xmath1 . on the other hand
the heat @xmath124 is very small .
this happens because with the present geometry @xmath3 can be exchanged only through two links and because we are working in the aging limit .
indeed , while @xmath113 and @xmath114 are equilibrated at the same temperature , and hence @xmath125 , @xmath126 is the only contribution to @xmath127 .
however , in the aging limit since @xmath128 and the number of slow modes which contribute to @xmath124 is @xmath129 , there is a number of order @xmath130 of terms in the double sum in eq .
( [ quantoq ] ) . on the other hand ,
for the heats exchanged involving only fast degrees , such as @xmath131 , @xmath132 , and @xmath133 , @xmath134 is constant and hence there is a number of order @xmath18 of terms contributing in eq .
( [ quantoq ] ) .
therefore , the heat @xmath124 is negligible with respect to the others .
this shows that our experiments are performed in the limit of small average heat exchanged between the two systems .
however , although small , @xmath124 is not strictly zero and its behavior parallels that of @xmath67 . this can be seen in fig .
[ figq12vsq2 ] where @xmath127 , @xmath83 and their ratio is plotted for different choices of @xmath120 . in the inset of this figure
one sees that , after an initial transient , @xmath83 decays as @xmath135 .
a similar behavior is found for @xmath136 , the main difference being an oscillating behavior superimposed on the @xmath135 decay .
the origin of such oscillations is not completely clear and is probably related to the recurrent passage of interfaces across the boundary links .
their amplitude can be suppressed by increasing @xmath120 , because collecting the heats on larger time windows effectively averages out the oscillating behavior . for @xmath137
the periodic behavior is basically washed away .
notice also that the curves tend to superimpose as @xmath120 increases .
the ratio @xmath138 is plotted in the main part of the figure , showing that for sufficiently large values of @xmath18 and @xmath120 it converges to a constant value , namely @xmath127 and @xmath83 are proportional . this behavior supports the hypothesis that slow degrees transfer heat to the fast ones and to the reservoir basically in the same way .
this is so because , since fast degrees are thermalized with the reservoirs they affect the slow ones similarly to what the thermal baths do , namely absorbing heat .
[ a1 ] let us see how these findings fit into the scenario with two degrees coexisting at different ( effective ) temperatures .
this framework implies that statistical independence is retained and thermalization does not occur between them .
the question is why .
the most obvious explanation would be that slow and fast degrees do not effectively interact ; they are _ insulated _ , in some sense .
however , the above experiment has shown that this is not the case , since heat is transferred from slow degrees to the fast ones and to the reservoir basically in the same way .
hence , statistical independence must have a different origin .
a possible explanation is the following : let us consider two separated domains at time @xmath139 ( left part of fig .
( [ due_interf ] ) ) .
they successively merge at the later time @xmath140 , releasing some heat @xmath141 to the fast degrees surrounding the coalescence event .
this excess heat will outflow to the bath on the microscopic time @xmath39 .
after the event the slow degrees which have originally released the heat are annihilated .
the surviving ones are left unmodified by this process , and hence retain their original properties . in summary , if energy release is always accompanied by annihilation , the heat transferred between fast and slow components up to a certain time was provided by interfaces that do not exist any more , while those still surviving have never been involved . notice that , writing @xmath142 , where @xmath143 is the heat released to the reservoir , from eq .
( [ eq : e ] ) one has @xmath144 , where @xmath145 is the number of slow degrees .
this shows that @xmath146 is proportional to the decrease of @xmath145 , and this is consistent with the hypotheses that heat release and annihilation are deeply related processes .
having shown that fast degrees absorb the excess heat released by the interfaces it remains to be shown that this process does not spoil the thermal properties of the fast degrees , preserving in this way the statistical independence .
this is expected because bulk degrees are able to thermalize among themselves and with the reservoirs in a microscopic time @xmath147 , making their thermal properties irrespective of the presence of interfaces .
if this is true , one might predict that when two coarsening systems quenched to different temperatures @xmath0 , @xmath148 are put in contact , heat flows from the fast components of the hotter system to those of the colder .
moreover , fluctuations of heat exchanged between fast modes should be similar to those observed in stationary systems and , in particular , governed by the fluctuation principle ( [ fluc ] ) .
strictly speaking , we should verify eq .
( [ fluc ] ) but with @xmath116 in place of @xmath3 . however , we recall that eq .
( [ fluc ] ) is supposed to hold in the large-@xmath84 domain where boundary terms can be neglected , similarly to what happens in eq .
( [ qdifff ] ) . in this regime
the number of bulk spins is much larger than that of the interfacial ones and so @xmath149 is negligible with respect to @xmath116 .
then , if the thermal properties of fast degrees is unaffected by the presence of the slow ones one should be able to verify the fr ( [ fluc ] ) for the _ whole _ heat transferred @xmath3 . in order to study this ,
we have designed an experiment similar to the one discussed above . however
now we quench system 1 to @xmath150 , while 2 is in equilibrium at @xmath151 . in fig .
[ a3 ] we compare the probability distribution @xmath152 of the heat @xmath3 in this experiment , with the one @xmath153 measured in the experiment described in sec .
( [ two ] ) where the samples are again in contact with the same baths at @xmath154 , but they are both stationary .
the two distributions have a zig - zag shape .
this is not due to a lack of statistics but is a genuine feature that we will discuss later .
notice that , at any finite time , the positive tail of @xmath152 is fatter than that of @xmath153 .
this is due to the extra heat flowing from the coarsening system 1 to the stationary sample 2 , which is produced by the reduction of the interface density , similarly to the case with @xmath155 discussed above ( see fig .
[ a1 ] ) . besides this
, one observes that the two distributions tend to merge in the large time limit when all boundary terms become negligible and slow degrees are sufficiently few .
we made this observation more quantitative by plotting , in one inset of fig .
[ a1 ] , the _ distance _ @xmath156 between the two probability densities , defined in eq .
( [ distance ] ) , which decreases in a power - law way .
the merging of the two distributions indicates that the thermal properties of the fast degrees are left unchanged by the presence of the slow ones . as a further check we consider the validity of eq .
( [ fluc ] ) : after having verified that the l.h.s . of eq .
( [ fluc ] ) is well fitted by @xmath157=k(\tau)(1/t_2 - 1/t_1)q_{1,2}$ ] for all the values of @xmath84 , we have plotted the _ slope _ @xmath158 in the inset of the bottom - right panel of fig .
one sees that @xmath158 converges to @xmath159 when @xmath75 is increased , as expected .
let us now come back to the zig - zag feature of the distributions of fig .
[ a3 ] . which can be explained as follows .
let us consider the 4-steps microscopic heat transfer process ( [ scheme ] ) . according to our operative definition of @xmath3 , one measures @xmath160 or @xmath96 after the first or after the fourth step , respectively .
this implies that @xmath3 is always a multiple of 2 , in this model . in the situations considered in this section spin 2
is part of an extended system in contact with the reservoir at @xmath110 .
this system is in a coarsening stage , where the up - down symmetry is not globally broken .
however locally , the neighborhood of spin 2 has almost surely a broken symmetry character , if time is large enough .
indeed , this spin will almost surely be located in the bulk of a magnetized domain .
suppose that symmetry is broken around the @xmath161 direction .
in this situation the probability of finding the first or the last configuration of the process ( [ scheme ] ) is much higher than that of finding the other . since the final time @xmath162 over which the heats are collected is not correlated to the spin configurations , one has a much higher probability that @xmath3 is a multiple of 4 than that it is not
this explains the zig - zag behavior where values of @xmath3 which are multiples of 4 are enhanced . computed in a stationary state with @xmath163 , @xmath151 and in the setup where 2 is ( quasi)-stationary at @xmath151 while 1 is aging after a quench to @xmath163 .
the system size is @xmath164 .
different values of @xmath84 are shown , with @xmath165 .
the statistics is collected over @xmath166 realizations . in the inset of the left - bottom panel
the _ distance _
@xmath156 is plotted against @xmath18 . in the inset of the right - bottom panel
the values of the slope @xmath167 obtained from the fluctuation - relation ( see text ) are plotted against @xmath18.,title="fig:",scaledwidth=43.0% ] computed in a stationary state with @xmath163 , @xmath151 and in the setup where 2 is ( quasi)-stationary at @xmath151 while 1 is aging after a quench to @xmath163 .
the system size is @xmath164 .
different values of @xmath84 are shown , with @xmath165 .
the statistics is collected over @xmath166 realizations . in the inset of the left - bottom panel
the _ distance _
@xmath156 is plotted against @xmath18 . in the inset of the right - bottom panel
the values of the slope @xmath167 obtained from the fluctuation - relation ( see text ) are plotted against @xmath18.,title="fig:",scaledwidth=43.0% ] + computed in a stationary state with @xmath163 , @xmath151 and in the setup where 2 is ( quasi)-stationary at @xmath151 while 1 is aging after a quench to @xmath163 .
the system size is @xmath164 .
different values of @xmath84 are shown , with @xmath165 .
the statistics is collected over @xmath166 realizations . in the inset of the left - bottom panel
the _ distance _
@xmath156 is plotted against @xmath18 . in the inset of the right - bottom panel
the values of the slope @xmath167 obtained from the fluctuation - relation ( see text ) are plotted against @xmath18.,title="fig:",scaledwidth=43.0% ] computed in a stationary state with @xmath163 , @xmath151 and in the setup where 2 is ( quasi)-stationary at @xmath151 while 1 is aging after a quench to @xmath163 . the system size is @xmath164 .
different values of @xmath84 are shown , with @xmath165 .
the statistics is collected over @xmath166 realizations . in the inset of the left - bottom panel
the _ distance _
@xmath156 is plotted against @xmath18 . in the inset of the right - bottom panel
the values of the slope @xmath167 obtained from the fluctuation - relation ( see text ) are plotted against @xmath18.,title="fig:",scaledwidth=43.0% ]
coarsening systems are prototypical models which exhibit a non trivial aging behavior where two classes of degrees of freedom with different properties can be identified and studied to some extent .
they are then particularly suited to understand the issue of thermal exchanges in aging states and the related concept of effective temperatures . in this perspective ,
in this paper we have studied numerically the heat flows occurring in an ising model quenched below the critical temperature . in order to do that we have first developed an algorithm to quantify the heat transferred between two parts of a system , and we have subsequently used this tool in a series of numerical experiments designed as to detect the heat fluxes occurring between the two kinds of degrees existing in the coarsening system .
our results fit into a scenario where fast components thermalize with the reservoirs and act themselves as baths where interfaces can temporary store their excess energy when annihilation events occur .
the mechanism is such that , after a microscopic time of order @xmath39 , the statistical properties of both fast and slow modes are left unchanged .
this guarantees that the two kind of degrees may remain statistically independent , as shown analytically in a soluble limit in @xcite .
once a thermodynamic interpretation of the effective temperature ( in the spirit of ref .
@xcite ) is agreed upon ( let us mention however that our results do not speak about the consistent interpretation of @xmath4 as a genuine thermodynamic temperature ) our results help to explain why a couple of different effective temperature can be sustained indefinitely . at variance to what is naively expected , this does not occur because the two classes of degrees are thermally insulated , or not effectively interacting . on the contrary , heat is exchanged between them much in the same way as it is exchanged with the baths . the mechanism
whereby this occurs , nevertheless , is such to preserve the statistical properties of the two kind of modes .
we have given support to this allegation by proving that the gallavotti - cohen fr ( [ fluc ] ) is obeyed in the asymptotic domain also when aging occurs in systems at different temperatures .
let us stress here that @xmath0 and @xmath1 entering eq .
( [ fluc ] ) are in this case the temperatures of the thermal baths .
the presence of @xmath4 does not play an appreciable role in our experiments because slow degrees do not contribute significantly to the heat exchanged @xmath3 , which is dominated by the contribution of the fast spins .
interestingly , @xmath4 might be expected to enter the fr @xcite if only the heat exchanged by slow degrees was considered , ( namely subtracting away the dominant contribution provided by the fast ones in a specially designed experiment ) .
evans , e.g.d .
cohen and g.p .
morriss g p , _ phys .
_ * 71 * , 2401 ( 1993 )
evans and d.j .
searles , _ phys .
_ e * 50 * , 1645 ( 1994 ) .
g. gallavotti and e.g.d .
cohen , _ j. stat
_ * 80 * , 931 ( 1995 ) .
lett . _ * 74 * , 2694 ( 1995 ) .
p. c. hohenberg and b. i. shraiman , physica d * 37 * , 109 ( 1989 ) .
m. c. cross and p. c. hohenberg , rev .
phys . * 65 * , 851 ( 1993 ) . m. s. bourzutschky and m.
c. cross , chaos * 2 * , 173 ( 1992 ) .
m. caponeri and s. ciliberto , physica d * 58 * , 365 ( 1992 ) .
c. godrche and j.m .
luck , j.phys .
a 33 , 1151 ( 2000 ) .
e. lippiello and m. zannetti , phys.rev.e 61 , 3369 ( 2000 ) .
f. corberi , e. lippiello , and m. zannetti , phys .
e * 63 * , 061506 ( 2001 ) ; eur .
j. b * 24 * , 359 ( 2001 ) ; jstat , p12007 ( 2004 ) .
f. corberi , a. de candia , e. lippiello , and m. zannetti , phys .
e * 65 * , 046114 ( 2002 ) .
f. corberi , e. lippiello , and m. zannetti , j. stat .
p12007 ( 2004 ) .
f. corberi , c. castellano , e. lippiello , and m. zannetti , phys .
e * 70 * , 017103 ( 2004 ) .
f. corberi , c. castellano , e. lippiello , and m. zannetti , phys .
e * 65 * , 066114 ( 2002 ) . | this paper is a contribution to the understanding of the thermal properties of aging systems where statistically independent degrees of freedom with largely separated timescales are expected to coexist .
focusing on the prototypical case of quenched ferromagnets , where fast and slow modes can be respectively associated to fluctuations in the bulk of the coarsening domains and to their interfaces , we perform a set of numerical experiments specifically designed to compute the heat exchanges between different degrees of freedom .
our studies promote a scenario with fast modes acting as an equilibrium reservoir to which interfaces may release heat through a mechanism that allows fast and slow degrees to maintain their statistical properties independent .
pacs : |
the @xmath7cdm model predicts the existence of a cosmological constant @xmath7 , entering the einstein equations and responsible for the dynamical properties of the universe today .
indeed , the model arguably represents the simplest cosmological paradigm to allow the universe to speed up at late times . despite its experimental successes ,
the @xmath7cdm model is unable to explain why the magnitudes of both matter and @xmath7 energy densities are so comparable today .
nevertheless , cosmological constraints indicate a severe difference between the observed @xmath7 and the corresponding quantum field theory s value .
it follows that the standard @xmath7cdm model could represent only a limiting case of more general paradigm , based on the existence of some sort of ( dynamical ) cosmic fluid , driving the universe today , dubbed dark energy @xcite .
dark energy seems to dominate over all cosmological species enabling the universe to accelerate .
dark energy acts as a _ dynamical source _ for speeding up the universe , whereas dark matter is responsible for structure formation .
recently , the need of understanding at which time the decelerating - accelerating transition phase occurs has leads cosmologists to directly measure the corresponding transition redshift @xmath0 @xcite
. there exists a wide consensus , based on robust observational supports , indicating @xmath0 around the unity @xcite . even though the @xmath7cdm value of @xmath0 reproduces the same result of recent observations
, we can not conclude that dark energy is a constant at all stages of universe s evolution .
however , several classes of models depart from the case of @xmath9 , indicating that @xmath0 represents a _
_ n__ew cosmic number which any cosmological model should reproduce to be in accordance with cosmic bounds . among concurring models to @xmath7cdm , modified gravity is considered a realistic alternative to dark energy . in particular , the challenge of finding out the most viable function @xmath1 , which permits to describe the gravity as a whole , passes through the reconstruction of the correct cosmological model . in other words
, determining viable numerical outcomes of @xmath0 means to reproduce the @xmath1 models which satisfy the late - time bounds in a model independent way @xcite . to support this idea
, one can consider _ cosmography _ , which represents a technique for matching cosmic data with observable quantities without imposing a particular cosmological model .
in general , a taylor expansions evaluated at present - time is adopted in cosmography . constraining a class of coefficients
gives rise to the _ cosmographic series _ @xcite .
hence , the use of cosmography may suggest possible departures from the standard @xmath7cdm model and also allows to understand how and whether dark energy evolves in time @xcite . in this paper , we show how to get the hubble rate by assuming the cosmographic constraints on @xmath1 models .
further , we bound numerical intervals for the transition redshift @xmath0 and demonstrate that only particular classes of @xmath1 models allow to get @xmath0 in the correct observable ranges .
afterwards , we propose an effective evolving dark energy term which extends the @xmath7cdm model and provides a logarithmic correction to the dark energy density .
this model predicts that the corresponding acceleration parameter changes sign around @xmath10 .
finally , we describe the corresponding @xmath11 function in terms of the cosmographic coefficients and propose a possible expression for it . the paper is structured as follows : in sec .
ii we describe the cosmographic approach in the context of @xmath1 gravity and compute the corresponding transition redshift .
we also show a comparison with current data and demonstrate that the dark energy corrections are compatible with observations .
we also report numerical outcomes .
finally , sec .
iii is devoted to conclusions and perspectives .
c c c c fit & supernovae & @xmath12 + [ 0.8ex ] @xmath13 value & @xmath14 & @xmath15 + & @xmath16@xmath17 & @xmath18@xmath19 + @xmath20 & @xmath21@xmath22 & @xmath23@xmath24 + @xmath25 & @xmath26@xmath27 & @xmath28@xmath29 + @xmath30 & @xmath31@xmath32 & @xmath33@xmath34 + @xmath35 & @xmath36@xmath37 & @xmath38@xmath39 + @xmath40 & @xmath41@xmath42 & @xmath43@xmath44 + @xmath45 & @xmath46@xmath47 & @xmath48@xmath49 + [ tab : dl2 ]
corrections to the hubble rate , inferred from @xmath1 gravity , can be achieved by means of cosmography .
the technique allows to determine the corresponding @xmath1-transition redshift @xmath0 showing its compatibility with present - time constraints .
in particular , cosmography represents a technique to fix limits on @xmath1 and its derivatives by passing through the use of an auxiliary function @xmath2 .
this means to define @xmath11 and its first derivative through @xmath50 . to this end
, one needs the scale factor expansion into a taylor series around the present epoch @xmath51 : @xmath52 keeping in mind eq .
( [ aexpans ] ) , we are able to evaluate @xmath2 and derivatives at @xmath53 which corresponds to @xmath54 . to relate @xmath2 and derivatives to the cosmographic expansion of @xmath55 , it is necessary to assume the constraint on @xmath56 @xcite @xmath57 with @xmath58 .
simple algebra leads to @xmath59 and @xmath60 . from eq .
( [ aexpans ] ) , we define the cosmographic coefficients as @xmath61 respectively the hubble rate @xmath62 , the acceleration parameter @xmath63 , the variation of acceleration , i.e. the jerk parameter @xmath64 . constraining the above quantities by means of cosmological data permits one to fix the initial settings on the curvature dark energy term and so on the form of @xmath1 .
moreover , once scalar curvature is somehow fixed by geometrical observations , cosmography becomes a pure model - independent treatment to fix cosmic bounds and to understand which corrections due to @xmath1 models are really viable to describe current universe expansion .
furthermore , it is easy to show that the measurements of the cosmographic parameters provide the disadvantage that all coefficients @xmath65 and @xmath64 can not be measured alone .
assuming , in fact , the luminosity distance as function of the cosmographic series , i.e. @xmath66 , and comparing cosmic data with it , we notice that it is impossible to _ decouple _ the terms @xmath67 and @xmath68 , respectively the second and third orders of the taylor expansion of @xmath69 @xcite .
thus , to better limit @xmath70 and @xmath5 independently , alleviating the corresponding degeneracy problem between coefficients , and also reducing the error bars , one can measure @xmath71 alone , by considering the first order of @xmath69 taylor expansion @xmath72 in the observational range @xmath73 . here , we adopt the union 2.1 supernova compilation @xcite and we compare our results with those predicted by the planck mission @xcite . once @xmath71 is fixed , it is possible to determine the corresponding @xmath74 and @xmath5 and it is also possible to obtain @xmath75 finally , we have @xmath76 these expressions are needful for setting priors on @xmath2 in terms of cosmography and to understand which deceleration - acceleration transition redshift @xmath0 is predicted in the context of @xmath1 gravity . with these considerations in mind , we can integrate the modified friedmann equations , directly by using cosmographic measurements as a sort of _ setting conditions_. in so doing , we infer a parameterized cosmological model , which differs from the @xmath7cdm model and shows an effective logarithmic dark energy term @xmath6 of the form @xcite : @xmath77 the whole hubble rate becomes @xmath78 , with @xmath35 and @xmath40 constants to be fixed by present - time bounds .
in fact , to get @xmath79 , terms @xmath35 and @xmath40 are @xmath80\,,\end{aligned}\ ] ] which are consistent with cosmographic bounds .
the corresponding cosmographic @xmath2 function becomes @xmath81 with @xmath82 , @xmath83 , @xmath84 , @xmath85 and @xmath86 and @xmath87 .
this @xmath2 predicts the above modified hubble rate and provides a deceleration - acceleration transition redshift of the form @xcite @xmath88 } { \left[\omega_m(1+z)^3 + \ln(\alpha+\beta z)\right]}\,.\ ] ] by means of eqs .
( [ lkdfj ] ) and ( [ quxz ] ) , the cosmographic reconstruction favors values of the hubble constant in agreement with other previous estimations @xcite and with the planck results on @xmath71 @xcite .
the bounds on @xmath89 and @xmath90 suggest that @xmath63 changes sign at @xmath91 , in agreement with previous measurements @xcite , showing meanwhile that our model provides @xmath92 to be compatible with the @xmath7cdm predictions . in addition , our cosmological model , providing an evolving dark energy , as presented in eq .
( [ kjdgh ] ) estimates viable @xmath30 and @xmath40 , with @xmath93 $ ] , providing slight departures from the standard @xmath7cdm model .
further , our model seems to pass experimental tests , performed using monte carlo analyses and the union 2.1 and hubble rate measurement data sets @xcite all our numerical outcomes have been listed in table i. future improvements will better distinguish any significant deviations from the @xmath7cdm case at higher redshift domains . with our results , in fact , it seems that dark energy slightly evolves also at small redshift , although a definitive conclusion remains object of future developments .
the passage between the deceleration to acceleration phases happens as the acceleration parameter @xmath63 vanishes , corresponding to a precise transition redshift @xmath0 . to understand how well @xmath1 gravities are able to reproduce suitable bounds on @xmath0
, we proposed to reconstruct the class of @xmath1 functions by means of cosmography .
cosmography represents a strategy to fix constraints on the principal observable quantities by means of taylor expansions around present - time .
cosmic data may be directly matched with the cosmographic coefficients and so the cosmographic series permits to understand which models are really favored than others . in the case of @xmath1 gravity , we defined the auxiliary function @xmath2 as @xmath11 and its first derivative as @xmath94 . thus , we showed how to relate them to the cosmographic parameters , i.e. @xmath4 and @xmath5 . in particular , reconstructing the modified dark energy term in terms of @xmath1 , by using as _ initial settings _ the numerical values inferred from cosmography , we found that dark energy seems to evolve as a curvature fluid proportional to a logarithm , at small redshift .
this fact departures from the cosmological standard model , in which dark energy is a constant at all stages of universe s evolution .
however , albeit our model showed an evolving dark energy term , its transition redshift @xmath0 is compatible with present - time observations .
we also found the corresponding @xmath95 . to find out @xmath2 ,
the degeneracy between cosmographic coefficients are alleviated , fixing @xmath71 , by fitting the first order taylor expansion of the luminosity distance in the range @xmath96 .
all numerical results are in agreement with present - time observations and it seems possible that dark energy may be framed in terms of a curvature dark energy fluid . to show that paradigm is capable of well reproducing cosmic bounds , we employed experimental tests , making use of the most recent union 2.1 supernova survey and of the hubble measurement data set .
future developments will assume different redshift domains to better fix the cosmographic coefficients .
thus , the corresponding effective dark energy term would change correspondingly , providing a different transition redshift
. this technique will be useful to understand whether @xmath1 functions are compatible with cosmological data at different epochs .
moreover , another significative aspect will be to directly relate the cosmographic coefficients to the transition redshift itself , opening the possibility to use @xmath0 as a _
cosmographic coefficient . | we propose a strategy to infer the transition redshift @xmath0 , which characterizes the passage through the universe decelerated to accelerated phases , in the framework @xmath1 gravities . to this end , we numerically reconstruct @xmath2 , i.e. the corresponding @xmath1 function re - expressed in terms of the redshift @xmath3 and we show how to match @xmath2 with cosmography . in particular , we relate @xmath2 and its derivatives to the cosmographic coefficients , i.e. @xmath4 and @xmath5 and demonstrate that its corresponding evolution may be framed by means of an effective logarithmic dark energy term @xmath6 , slightly departing from the case of a pure cosmological constant .
afterwards , we show that our model predicts viable transition redshift constraints , which agree with @xmath7cdm . to do so , we compute the corresponding @xmath0 in terms of cosmographic outcomes and find that @xmath8 . finally , we reproduce an effective @xmath2 and show that this class of models is fairly well compatible with present - time data . to do so ,
we get numerical constraints employing monte carlo fits with the union 2.1 supernova survey and with the hubble measurement data set . 2 |
in our experiments , @xmath22rb atoms in the @xmath23 hyperfine state are confined in a time - averaged orbiting potential ( top ) magnetic trap @xcite . evaporative cooling increases the phase space density to near the condensation critical point .
the trap frequencies are then relaxed , and a final stage of cooling ( quench a , b , or c ) induces the phase transition . in our numerical approach , the field consisting of the condensate and the low - energy , highly occupied modes of the gas is coupled to a bath of thermal atoms , parameterised with a chemical potential @xmath7 and temperature @xmath24 above a cut - off energy @xmath25 .
further details regarding our approaches are provided in the methods section below .
* evaporative cooling . * during the main evaporative cooling stages of our experimental procedure , our top trap is created with a spherical quadrupole field that has a vertical magnetic field gradient of @xmath26 g / cm , and a magnetic bias field @xmath27 that has a direction that rotates in a horizontal plane at a frequency of either @xmath28 khz or @xmath29 khz . evaporative cooling proceeds over 72 seconds as @xmath27 decreases from 41 g to @xmath125.2 g , leaving a trapped cloud of atoms just above the condensation critical temperature @xmath30 .
@xmath31 is then adiabatically reduced to @xmath1254 g / cm over 2 seconds , weakening the harmonic oscillator trapping frequencies to a measured radial ( horizontal ) trapping frequency of @xmath32 hz and an axial ( vertical ) trapping frequency of @xmath33 hz . the center - of - mass position of the atom cloud correspondingly sags by approximately 0.6 mm vertically . in the final stage of our cooling cycle for quench a , we use a continuous 6-s ramp of the rf field , which induces the evaporative cooling of the atom cloud from 70 nk to 20 nk , with @xmath34 nk , to create condensates of @xmath35 atoms . for quench
b , the continuous rf evaporative cooling ramp is replaced with a sudden rf jump to a final rf value , followed by a hold of the atomic sample in the trap before release and imaging . in this situation
we find @xmath36 nk and the final condensate number is @xmath37 atoms .
* top trap . * to ensure that the rotating bias field of the top trap plays no significant role in the spontaneous formation of vortices , we measured the @xmath13-component of the net orbital angular momentum @xmath38 of our condensates using surface wave spectroscopy .
we excite a quadrupolar oscillation of the bec in the horizontal plane , and stroboscopically probe the bec with a set of non - destructive in - trap phase - contrast images , obtained by probing along the @xmath13 axis @xcite .
the quadrupolar oscillations will then precess with a rate and direction proportional to @xmath38 . in our measurements
, there was no significant biasing of surface mode precession in a direction corresponding to the top trap rotation direction , an indication that top trap temporal dynamics have little to no influence on spontaneous vortex formation .
this is discussed further in supplementary information
. * toroidal trap . *
a potential - energy barrier was added to the centre of the magnetic trap using a focussed blue - detuned laser beam with a wavelength of 660 nm , @xmath1218 @xmath7w of power , and a 6-@xmath7 m gaussian radius .
the maximum beam intensity corresponds to a potential energy of approximately k@xmath3920 nk , where k@xmath40 is boltzmann s constant .
this can be compared with a @xmath12k@xmath3910 nk chemical potential of our fully formed becs in the purely harmonic trap .
the beam was adiabatically ramped on prior to the final 6-s evaporation ramp , only slightly perturbing the thermal cloud but providing enough additional potential energy along the trap axis to exclude bec atoms from the @xmath13 axis of the trap .
* imaging . *
our main imaging procedure involves the sudden removal of the magnetic trap , and the subsequent ballistic expansion of the trapped cloud .
after 59 ms of expansion in the presence of an additional magnetic field to support the atoms against gravity , the atomic cloud is illuminated with near - resonant laser light , and the absorption profile of the atomic density distribution is imaged onto a camera . in our grey scale images ,
lighter shades represent higher optical depth , proportional to integrated column density along the @xmath13-direction line - of - sight .
a clear vortex core aligned along the @xmath13 axis appears as a dark hole in the density distribution . *
stochastic gross - pitaevskii theory .
* we denote the condensate and low - energy portion of the trapped gas with the field @xmath41 , and define the gross - pitaevskii operator @xmath42 where @xmath43 is the mass of an atom , @xmath44 is the trapping potential , @xmath45 characterises the strength of atomic interactions , and @xmath46 is the s - wave scattering length .
the equation of motion for the field is @xmath47 which has been derived from first principles using the wigner phase - space representation @xcite .
the first term on the right describes unitary evolution of the classical field according to the gross - pitaevskii equation .
the second term represents growth processes , i.e. collisions that transfer atoms from the thermal bath to the classical field and vice - versa , and the form of @xmath48 may be determined from kinetic theory @xcite .
the third term is the complex - valued noise associated with condensate growth .
the noise has gaussian statistics and is defined by its only non - vanishing moment : @xmath49 ; it is also consistent with the fluctuation - dissipation theorem .
the projection operator @xmath50 restricts the dynamics to the low - energy region @xcite defined by all harmonic oscillator modes with energy @xmath51 for these calculations , which for our parameters gives about three particles per mode at the cutoff . for typical experimental parameters
this method is accurate from slightly above the critical temperature to colder temperatures where there is still a significant thermal fraction @xcite .
the initial states used in our simulations are independent field configurations generated by ergodic evolution of the sgpe at equilibrium with the thermal cloud with @xmath52 and @xmath53 nk for quench a ( b ) , representing the thermalised bose gas above the transition temperature @xcite .
these parameters are then suddenly changed to final values chosen to match the final condensate number and temperature observed in the experiment : @xmath54 @xmath55 @xmath20 and @xmath56 = 34 ( 25 ) nk for quench a ( b ) .
we perform simulations for 300 ( 298 ) sets of initial conditions . by averaging over the different realisations we can calculate any quantum - mechanical observable as a function of time , and in particular we diagonalise the single - particle density matrix to find the number of atoms in the condensate @xcite .
because vortex formation is expected to depend upon the bec growth rate , which is difficult to calculate precisely , we adjust the coupling rate describing bose - enhanced collisions between the classical field and thermal cloud to obtain a close match of the experimental bec growth curves .
we choose a spatially constant rate for the dimensionless coupling @xmath57 .
the noise at each time step then has the explicit form @xmath58 , where @xmath59 are the single particle modes below the cutoff and the real gaussian variables @xmath60 , @xmath61 , are independent and have zero mean and unit variance . in principle @xmath62
is specified by a quantum boltzmann integral , but here we treat it as an experimental fitting parameter for the condensate growth rate ; it is never more than a factor of two different from the result of eq .
( a11 ) in bradley _
et al_. @xcite .
the effect of these parameter choices is discussed further in supplementary information .
we thank david roberts , boris svistunov , ewan wright , and wojciech zurek for useful discussions .
the experimental work was supported by the us national science foundation under grant 0354977 , and by the army research office .
the theoretical work was supported by the australian research council and the university of queensland . _
@xmath63college of optical sciences , university of arizona , tucson , az 85721 , usa + @xmath64arc centre of excellence for quantum - atom optics , school of physical sciences , university of queensland , + brisbane , queensland 4072 , australia _ * supplementary video legends * + here we describe the main features of a selection of movies of our condensate formation simulations .
movie files are available online at * http://www.physics.uq.edu.au / people / mdavis / spontaneous_vortices*. ten example movies are given : five corresponding to condensate formation in the purely harmonic trap ( quench a ) , and five in the toroidal trap ( quench c ) . as specified in the title frames ,
all simulations in this section have initial chemical potential @xmath52 and initial temperature @xmath65 nk , and corresponding final values @xmath66 and @xmath67 nk .
quench c has an additional central repulsive gaussian barrier of height @xmath68 .
all movies begin at time @xmath69 corresponding to the instant that the chemical potential and temperature are suddenly changed in order to induce condensation .
the left panel of each movie shows a rendering of a three - dimensional isosurface of the density of the classical field , with magenta ( cyan ) dots indicating the presence of positive ( negative ) phase windings about the @xmath13 direction .
we indicate only the phase windings located in a spatial region where there is a significant condensate density , as determined by the penrose - onsager criterion . at earlier times , these points of fluid circulation seem to be distributed almost at random and at high spatial density , however at longer times these points can often be seen to form vortex lines that extend through the condensate .
the top - right panel of each frame plots the column density along the @xmath13-axis of the classical field .
the greyscale colourmap is fixed in time as indicated by the gradient bar at the far right ( arbitrary units ) , and the numerical data appear similar to the type of images obtained in the experiment , although without the time - of - flight expansion .
vortices can often be seen as holes or low density regions in the column density image depending on their orientation with respect to the @xmath13-axis .
the lower right panel shows the phase of the field in the @xmath11 plane .
positive ( negative ) phase windings are indicated as magenta crosses ( cyan circles ) in the regions of significant condensate density .
it should be noted that many additional @xmath1 windings in this panel occur where there is no condensate density , and are hence not to be interpreted as vortices .
+ * condensate formation movies in a harmonic trap : quench a. * from 300 simulations of quench a , 229 contained no vortex cores after 1.5 s. the following five movies show a selection of outcomes , including cases with 0 , 1 , and 2 long - lived vortices .
+ * supplementary video 1 : * the first movie is an example of bec formation in which there are no vortices trapped in the final condensate , and is provided for comparison purposes . at early times there is some indication of vortices in the condensate ; however , these do not survive at longer times .
in this situation the condensate appears to grow adiabatically in its ground state , as was assumed in the condensate formation calculations of refs .
qkprli , qkprlii , lee2000a , davis2000a , bijlsma2000a .
the reader should note that cases such as this are typical for the data set corresponding to quench a. + * supplementary video 2 : * this movie provides an example of bec formation with a single vortex line that survives at long times and remains close to the centre of the condensate .
the vortex line remains approximately vertical , and is usually easily visible in the column density plot .
the reader should note that there are many examples similar to this where the vortex line is not so close to the centre , and instead slowly spirals towards the boundary of the condensate over a time scale of several seconds .
+ * supplementary video 3 : * in this example there are three clear vortices at early times .
two of these arise near the edge of the condensate and these damp out relatively quickly , leaving a single vortex line of opposite charge near the centre that survives to the end of the simulation .
this is an example of a vortex that does not directly align with the @xmath13 axis , and the column density often shows an elongated density dip .
this movie uses the same simulation data as in figure 5 of the main text . +
* supplementary video 4 : * this simulation results in two oppositely charged vortices that remain well separated , precessing about the centre in opposite directions .
it should be noted that only 3 out of 300 simulations in this data set clearly exhibited more than one core at long times .
+ * supplementary video 5 : * the final movie for the data set of quench a shows a number of vortex cores that undergo some rather complicated dynamics as they move about within the condensate , including vortices that cross each other and reconnect in a different configuration , and an example where a vortex line flips ( and hence changes colour ) .
this example shows the most complicated vortex dynamics observed in this data set . +
* condensate formation movies in a toroidal trap : quench c. * from the 300 simulations of quench c , 147 contained no vortices after 1.5 s. the following movies show five examples where vortices arose .
+ * supplementary video 6 : * the first movie from this data set shows the formation of a single vortex .
this is not readily seen in the isodensity surface , as the @xmath1 phase winding is located in the centre of the trap where there is no condensate density .
however , the phase winding about the entire condensate is clear in the bottom - right plot of the phase in the @xmath11 plane .
whereas in the harmonic trap all vortices will eventually make their way to the edge of the condensate and disappear , here the gaussian barrier pins the vortex to the centre this is an example of a persistent current . +
* supplementary video 7 : * in this example it is possible to see a pair of oppositely charged vortices at @xmath70 s , one of which becomes pinned to the central barrier while the other ends up precessing about the centre in the region of maximum condensate density . +
* supplementary video 8 : * here there are some rather complicated early dynamics that result in a single vortex precessing about the outside of the condensate , but with no persistent current .
+ * supplementary video 9 : * there are a number of vortices early on , and near 0.6 s it seems that there is a doubly charged persistent current . however
, only one vortex ends up being trapped on the central gaussian barrier and the other vortex of the same charge remains within the bulk of the condensate , precessing rapidly about the centre .
the rate of precession should be compared with the example from supplementary video 7 , where the vortex near the edge is of the opposite charge to that pinned by the barrier . + * supplementary video 10 : * the final example shows a stable doubly - charged persistent current .
this is particularly interesting given the energy difference between this and the thermodynamic ground state of the condensate with no current .
this is the only example from the data set where a stable doubly - charged persistent current was seen . +
* supplementary discussion * + * vortex distribution in the toroidal trap .
* we observe in the simulated bec growth dynamics for the toroidal trap that the central potential can pull in and pin a single spontaneously formed vortex in some cases . by examining the distribution of vortex core positions for both the experimental and simulated data we see that the toroidal trap induces a central clustering of vortex cores , whereas in the harmonic trap the core positions are more evenly spread throughout the becs .
a visual comparison of these cases is given in supplementary figs .
[ fig4]a d .
supplementary figures [ fig4]e f show a histogram of vortex core displacements away from the mean core position for the two different trap geometries for theory and experiment . it should be noted that the experimental images in supplementary figs .
[ fig4 ] are taken after the removal of the central barrier of the toroidal trap ( if present ) , and after time - of - flight imaging , whereas the simulated images correspond to in - trap results .
hence , the data in this figure should be taken as indicative only of trends between harmonic versus toroidal trap results . despite this
, the vortex position data show some agreement between the simulations and the experimental data . due to the observation that the cores in the experimental toroidal trap data are likely to be found within a small region within the bec
, we interpret our results as indicating that the observed cores are likely to have been pinned to the central barrier prior to expansion .
this suggests possibilities for future controlled studies of spontaneous vortex formation , perhaps with multiple sites at which vortices may be pinned , in order to better understand the density and numbers of vortices created at early times in bec growth .
* vortex core pinning . * * a , c * , representation of the experimentally measured positions of the vortex cores relative to the thomas - fermi radius @xmath16 ( outer circles ) for * a * , the harmonic trap and * c * , the toroidal trap . *
b , d * , corresponding theoretical results , crosses and circles indicate oppositely charged vortices .
* e , f * , comparison of the statistics of the vortex displacements @xmath71 , binned in steps of @xmath72 , for the experimental data ( left bar in each pair ) and theoretical simulations ( right bars ) .
harmonic trap results are shown in * e * , toroidal trap results are shown in *
f*. for the experimental data , only images clearly showing a single core within 0.8 of the thomas - fermi radius were considered . for the data represented in this figure , the experimental data are obtained from _ expanded _ images , and the numerical data are obtained from _ in - trap _ simulations . in * d * , the individual vortex cores pinned to the central barrier can not be separately resolved ; nevertheless , the total number of cores pinned to the barrier significantly outweigh the number that are not pinned .
thus for both experimental and simulated toroidal , this figure shows that the toroidal trap strongly influences the vortex core positions within the bec , and that cores are most likely to be found within a smaller region of the bec for the toroidal trap than for the harmonic trap .
, width=396 ] we also note that the presence of a repulsive light field to create a toroidal potential could in principle lead to mechanisms for vortex formation other than spontaneous vortex formation during the bec phase transition , particularly if mechanical stirring exists ( for instance , if the beam or harmonic potential is not stationary ) .
we do not believe that a stirring mechanism is at work in our experiment . through phase - contrast images
, we confirmed that residual center - of - mass motion of the bec ( once formed ) is minimal , with bec position varying by approximately @xmath732 @xmath7 m at most , and usually much less .
this may be due to our procedure of evaporatively cooling directly in a final weak trap , rather than making a bec in a tighter trap and then relaxing to a weak trap ( larger residual center - of - mass oscillations may be observed when relaxing the trap after making a condensate ) .
furthermore , our laser beam position is relatively stable with respect to the center of the harmonic potential throughout the course of a full data set , and fluctuates or drifts less than 4 @xmath7 m during an experimental run .
we therefore believe that unintended stirring of the bec by the laser beam is unlikely . +
* vortex statistics . *
the table below shows a numerical comparison between the experimental and numerical vortex observation statistics in greater detail than provided in the main text . for each quench condition with total number of runs listed , @xmath74 shows the number of times @xmath75 vortices were observed , and @xmath76 shows the corresponding probability of observing @xmath75 vortices in a single run . in the experimental toroidal
trap data , 12 images , or 2%4% , showed 3 cores .
this is not represented in the table .
+ from fig .
2b in the main text and from supplementary table 1 it can be seen that there is some difference between the experimentally estimated probability of vortex observation compared to that from the simulations .
in particular the simulation estimate is slightly higher than experiment for the sudden cooling of quench b , and slightly lower than experiment for the 6-s evaporative cooling of quenches a and c at the time of observation . due to the limited number of realisations of the experiment for these three runs , it is difficult to say whether this is real difference or due to statistical fluctuations .
the theoretical modelling of the evaporative cooling is quite basic , and there is a strong possibility that this could affect the theoretical results in some manner .
however , further simulations ( discussed below ) suggest that the simulations results are quite robust to the time scale of the quench of @xmath7 and @xmath24 .
finally , we again note that the condensate growth curve was not measured experimentally for quench c , which shows the largest discrepancy in vortex statistics between theory and experiment .
it is possible that our simulations may not accurately represent the rate of growth of the condensate or the final condensate number for this case , giving a further plausible explanation for the difference . * vortices in phase transitions in ultra - cold gases . *
the description of defect formation in the kibble - zurek scenario is appealing due to its simplicity and universality .
however , a somewhat more complete description for the quantum gas is the evolution from weak turbulence to a quasi - condensate and superfluid turbulence as described by svistunov and co - workers [ 9,18 - 21 ] .
( for a summary see ref .
svistunov2000 ) . in this scenario , as the system cools ( but before condensation ) , the low - energy modes of the gas become highly occupied , and can be treated as classical waves .
the superposition of a number of low - energy modes with large amplitudes and random phases ( as appropriate for a thermal distribution ) naturally gives zeros in the density of the gas via destructive interference , which can be considered as precursors to vorticity .
a rapid quench leading to condensate formation can freeze these defects into the resulting superfluid , giving a tangle of vortices that then relaxes over a relatively long time scale . during this period of time ,
the system is not in thermodynamic equilibrium .
the presence of vortices suppresses the establishment of long - range order , and the system is said to be in a _ quasi - condensate _ state .
however , the thermodynamic state of the system that would be realised if the cooling proceeded more slowly is a condensate without vortices it is the rapid dynamics of condensate growth that can trap defects in the superfluid .
vortices are present in a two - dimensional ( 2d ) bose gas at cold temperatures ( but above any superfluid critical temperature ) for a similar reason there is a superposition of highly occupied classical modes with random phases .
this is why classical field models ( related to our theoretical techniques ) have had success in modelling berezinskii - kosterlitz - thouless ( bkt ) experiments in bose gases @xcite , see for example refs .
simula2006a , simula2008a , bisset2008a .
however , in a homogeneous system at least , the thermodynamic state of the system below the bkt superfluid temperature contains vortices that arise in bound pairs .
the superfluid - to - normal fluid transition occurs when the temperature is sufficiently high that these pairs start to unbind , resulting in free vortices .
the case of a trapped gas seems to be more complicated than this , and is only now beginning to be understood ; see for example bisset _
@xcite for a recent discussion of vortices , superfluidity , condensation and long - range order in 2d trapped systems . because vortices and vortex dynamics are an important element of the bkt transition , our work on spontaneous vortex formation may at first appear somehow related to the bkt transition .
however , the main difference between the bkt transition and spontaneous vortex formation can be briefly summarized as follows . both above and below the critical temperature of the bkt transition
, the thermodynamic state of a cold homogeneous 2d gas contains vortices ( either free or bound in pairs ) ; the bkt transition involves a transition from one equilibrium state ( with paired vortices ) to another equilibrium state ( with free and unbound vortices ) . in our work involving spontaneous vortex formation in a three - dimensional gas , vortices
are seen because they are frozen into the evolving gas ( the quasi - condensate ) during a fast transition when the system is out of equilibrium .
the thermodynamic equilibrium state below the critical point is free of vortices , and vortices are only observed in our work due to rapid - enough cooling in which the system falls out of equilibrium , and by observing the system during its relaxation from a quasi - condensate to a true condensate . +
* estimating the correlation length .
* here we provide a brief summary of our procedure to estimate the correlation length @xmath0 relevant for the bec phase transition of our experiments .
this method follows the procedure outlined by anglin and zurek [ 8 ] . in the kz model ,
a system undergoing a phase transition will fall out of thermal equilibrium when the equilibration ( or relaxation ) timescale @xmath77 exceeds the time remaining before the critical point is reached .
this essentially means that the decreasing equilibration rates can not keep up with the cooling rate near the phase transition critical point . defining time @xmath69 as the time when the system reaches the critical point , the time @xmath78 is defined as the point at which the system falls out of equilibrium .
the time @xmath79 can be estimated through the relationship @xmath80 , where @xmath81 is the quench time ( @xmath3 is the quench rate as defined in our paper ) , and @xmath82 is the scattering time of atoms in the gas . for our data , @xmath83 s , and
@xmath84 s for quench a and @xmath125 s for quench b , where @xmath81 is obtained from a linear fit of the function @xmath85 to the temperature data of figure 2a ( inset ) of the main text .
this leads us to an estimate of @xmath86 s. in other words , the correlation length @xmath0 of the gas at @xmath87 s becomes essentially frozen into the system because the gas can not regain equilibrium until a time @xmath88 s _ after _ passing through the critical point .
we estimate @xmath0 via the relation @xmath89 , where @xmath90 is the thermal debroglie wavelength of the system at @xmath78 . for our parameters ,
@xmath91 @xmath7 m for quenches a and b. + * experimental vortex observations . * in our experiments on condensate formation
we have not performed any uncommon experimental techniques in order to observe spontaneously formed vortices .
the bose gas of @xmath22rb atoms was evaporatively cooled in a similar manner to those of most other dilute - gas bec experiments [ 27 ] .
one might assume that rapid cooling is necessary in order for distinct regions of coherence and hence spontaneous vortices to form , in contradiction with the seemingly long cooling time scales for quenches a and c. it is then natural to wonder why our experiments have yielded observations of spontaneous vortex formation during the bec phase transition given that this has not been reported previously .
we believe that the major factors contributing to our observations are as follows : 1 .
we evaporatively cool our gas to near - degeneracy in a tight trap , and then relax the trap to the final trapping frequencies before our final cooling ramp .
this relaxation reduces the atomic collision rate at the point at which condensation occurs , and potentially decreases the ratio of the correlation length to the size of the ground state of the trap ( the harmonic oscillator length ) and thus increases the number of vortices expected to form . in the kz mechanism ,
the correlation length @xmath0 depends on the quench time @xmath81 , so specific details of the ratio of correlation length to the harmonic oscillator length will depend upon specific cooling trajectories , as well as trap frequencies .
regardless of the initial number of vortices formed , we expect that vortices may survive for longer times in weaker traps , where the lower atomic density and longer trap oscillation time scale generally decrease the rates of dynamical processes .
our experiment is performed in an oblate trap with an approximately 2:1 aspect ratio , in which it is favourable for the vortices to align along the vertical symmetry axis of the trap @xcite .
all analogous condensate formation dynamics experiments reported to - date in the literature have been performed in cigar - shaped traps @xcite .
we introduce a linear magnetic field gradient during expansion that supports the condensate against gravity and allows for long expansion times before imaging . during ballistic expansion of the condensate ,
the vortex cores expand relative to the condensate size such that our imaging system can optically resolve the cores .
unless one is actively looking for vortex cores , such long expansion times ( 59 ms in our case ) are not necessarily needed or desired in other experiments .
we have the ability to acquire images of trapped atoms along both the vertical axis ( the direction of strongest confinement ) as well as the horizontal axis , the more standard imaging axis for top trap experiments .
the vertical imaging axis allows us to detect vortices that are well - aligned with the vertical trap axis .
it seems entirely possible that spontaneous vortices have been present in other experiments , but simply could not be resolved , damped quickly , or were not sought . + * influence of the top field . *
using the method of quadrupolar surface mode excitation and precession detection [ 31,32 ] , early experimental results from this work appeared to show a significant bias of the direction of fluid circulation for the observed spontaneously formed vortices .
our initial measurements seemed to indicate that the direction of vortex fluid flow for the spontaneously formed vortices was most often found to be circulating in the same azimuthal direction as the rotating magnetic field that forms our top trap .
this suggested that some form of stirring or other influence from the top trap was responsible for the apparent directional bias .
however , our later and more detailed measurements on condensates created with quench a revealed that the apparent bias was a result of a scissors mode oscillation that was unintentionally excited .
coupled with the times that we initially chose to stroboscopically probe the surface mode precession , the scissors mode oscillation appeared as a bias to the precession direction .
such directional biasing is not detected in our more recent measurements using quench a , and we do not believe that the rotating field of the top trap plays a role in vortex formation ( these measurements were not performed with the toroidal trap data ) .
furthermore , there is no observable difference in vortex observation statistics between data obtained with a 2 khz or with a 4 khz rotating bias field for the top trap ; if the field were somehow stirring the cloud , it is likely that a difference in vortex observation statistics for the two cases should be detectable due to different angular momentum coupling between the rotating magnetic field and the trapped atoms . + * relation to simulations of berloff and svistunov . * here we briefly compare the simulations of berloff and svistunov [ 22 ] to our own numerical calculations .
the fact that our simulations are for the trapped gas rather than the homogeneous gas means that there are many quantitative differences compared to those of berloff and svistunov [ 22 ] however the basic picture of strongly nonequilbrium dynamics resulting in a vortex tangle and superfluid turbulence before subsequent relaxation is the same .
berloff and svistunov [ 22 ] simulated the transition from weak to superfluid turbulence in a homogenous bose gas using a discretised gross - pitaevsii equation on a large cubic grid of dimension @xmath92 .
their initial condition was a superposition of plane waves according to @xmath93 where the phases of the complex amplitudes @xmath94 were randomly distributed as appropriate for the weakly - interacting bose gas in the kinetic regime [ 21 ] .
the subsequent dynamics resulted in the redistribution of the mode occupation numbers leading to a growth in the low - momentum modes and the formation of a quasi - condensate while conserving energy and particle number .
the lines of zero density present in the initial state evolved into a vortex tangle as the quasi - condensate formed , which subsequently relaxed further and resulted in a number of vortex lines in the final state . in the visualisation of the resulting vortex networks ,
berloff and svistunov either filtered out the high momentum modes , or performed short - time averaging of the density , resulting in a smoothing of the isodensity surfaces that were plotted . in some sense
the high - momentum modes provide a mechanism for dissipation as the low - momentum modes of the system evolved towards thermal equilibrium . 1 .
our simulations begin with an equilibrium state for the low - energy classical modes above the critical temperature for condensation .
the quenching of our thermal reservoir to a lower temperature and higher chemical potential results in the classical region evolving to a strongly non - equilibrium state similar to the initial condition of berloff and svistunov [ 22 ] appropriate to a trapped gas .
this subsequently evolves through the quasi - condensate and superfluid turbulent stages towards equilibrium , sometimes trapping vortices .
the thermal reservoir provides the mechanism for dissipation in our simulations .
we explicitly separate the high - energy modes from the classical field represented by the sgpe , and represent them as a thermal reservoir .
berloff and svistunov include a large number of what we would classify as high - energy modes in their microcanonical gpe approach .
if the occupation number of these modes is not sufficiently high , this is not an entirely quantitatively accurate representation of the full system .
however , the dynamics of the high - energy modes has little qualitative impact on the dynamics of the low - momentum modes that form the superfluid , and provide an effective dissipative mechanism for relaxation . in our opinion ,
the key innovations differentiating our simulations from those of berloff and svistunov [ 22 ] are the efficient implementation of harmonic oscillator basis in our representation of the classical field , and the inclusion of the high - energy modes using grand canonical classical field theory [ 11,12,14,15 ] .
+ * dependence of results on simulation fitting parameters .
* in the main text we have compared our experimental results to simulations in which the chemical potential @xmath7 and temperature @xmath24 underwent an instantaneous quench to their final values . in this section
we discuss the dependence of the simulations modelling quench a on the parameters that characterise our idealisation of the cooling process . the final temperature chosen for the simulations
was based on the approximate value measured experimentally once the slope of the condensate growth curve began to flatten out .
we have also performed a set of 100 simulations where the final temperature was @xmath95 nk instead of @xmath96 nk , and found no difference in the vortex probabilities within statistical noise .
this implies that for our simulations the time - scale of the temperature decrease and the exact final value has little effect .
we have also performed the same simulations for 100 trajectories using a linear ramp of the temperature and chemical potential from their initial to final values over time scales of @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , and @xmath102 radial trap periods , and the results are shown in supplementary figures [ sifig2]a b .
we find that the first four of these give condensate growth curves that can not be distinguished , apart from an increasing delay in the initiation of growth .
the vortex observation probability appears to drop slightly with increasing ramp time , but this is difficult to distinguish within the statistical error . as we are not actually modeling the evaporative cooling trajectory of the experiment , and the time axis of our simulations is shifted to best fit the experimental data , it seems that the time scale of the quench is of little consequence for our simulations .
( thick solid blue line ) , @xmath97 ( thick red dashed line ) , @xmath98 ( thick green dot - dashed line ) , @xmath99 ( thick black dotted line ) , @xmath100 ( thin magenta solid line ) , @xmath101 ( thin blue dashed line ) , and @xmath102 ( thin red dot - dashed line ) radial trap periods .
the ramp begins at @xmath69 , and ends at the times indicated by the vertical black dotted lines .
* c , d * , condensate growth and vortex probability for different scattering rates : @xmath103 ( red dashed line ) , @xmath104 ( blue solid line ) , @xmath105 ( green dot - dash line ) , and @xmath106 ( black dotted line ) . , title="fig:",width=294 ] ( thick solid blue line ) , @xmath97 ( thick red dashed line ) , @xmath98 ( thick green dot - dashed line ) , @xmath99 ( thick black dotted line ) , @xmath100 ( thin magenta solid line ) , @xmath101 ( thin blue dashed line ) , and @xmath102 ( thin red dot - dashed line ) radial trap periods .
the ramp begins at @xmath69 , and ends at the times indicated by the vertical black dotted lines .
* c , d * , condensate growth and vortex probability for different scattering rates : @xmath103 ( red dashed line ) , @xmath104 ( blue solid line ) , @xmath105 ( green dot - dash line ) , and @xmath106 ( black dotted line ) . , title="fig:",width=294 ] this brings into question the relation between the kz mechanism of formation of topological defects and the simulations . in the kz mechanism ,
the quench time @xmath81 is a crucial parameter for estimating the initial density of spontaneously formed vortices , and it describes the rate at which the temperature @xmath24 is lowered relative to the critical temperature @xmath30 .
however , given that the condensate forms at the same rate even for different @xmath7 and @xmath24 ramp times in our simulations ( and hence the condensate `` quench '' is occurring on a similar time scale ) , similar vortex observation probabilities among these simulations seems reasonable . for the longer time
scale ramps of @xmath7 and @xmath24 over @xmath101 and @xmath102 radial trap periods , the initial sections of the condensate growth curves are much more rounded , and the maximum growth rate is reduced .
fewer vortices are observed after the condensate reached its final population , and it thus seems that over these time scales the `` effective quench time '' of the system is increased .
the other fitting parameter for the simulations was the dimensionless coupling of the classical field to the thermal cloud , @xmath104 , which is essentially a measure of the collision rate within the thermal cloud .
as a check of our methods , we have generated data sets of 200 trajectories for @xmath107 and @xmath108 with all other parameters unchanged .
doubling @xmath62 essentially halves the time scale for condensate formation as shown in supplementary figure [ sifig2]c , and so might lead one to predict that larger @xmath62 would lead to a higher probability of vortex formation and observation .
however , supplementary figure [ sifig2]d shows that while this is sometimes confirmed , the results are complicated by the fact that a larger @xmath62 also cause vortices to disappear at a faster rate . to make a definitive statement about the validity of the kibble - zurek scenario for trapped bose gases , it will be necessary to better control the quenching of the thermal cloud relative to the condensate .
we are currently working towards techniques to achieve this . | * phase transitions are ubiquitous in nature , ranging from protein folding and denaturisation , to the superconductor - insulator quantum phase transition , to the decoupling of forces in the early universe .
remarkably , phase transitions can be arranged into universality classes , where systems having unrelated microscopic physics exhibit identical scaling behaviour near the critical point . here
we present an experimental and theoretical study of the bose - einstein condensation phase transition of an atomic gas , focusing on one prominent universal element of phase transition dynamics : the spontaneous formation of topological defects during a quench through the transition @xcite . while the microscopic dynamics of defect formation in phase transitions are generally difficult to investigate , particularly for superfluid phase transitions @xcite , bose - einstein condensates ( becs ) offer unique experimental and theoretical opportunities for probing such details .
although spontaneously formed vortices in the condensation transition have been previously predicted to occur @xcite , our results encompass the first experimental observations and statistical characterisation of spontaneous vortex formation in the condensation transition . using microscopic theories @xcite
that incorporate atomic interactions and quantum and thermal fluctuations of a finite - temperature bose gas , we simulate condensation and observe vortex formation in close quantitative agreement with our experimental results .
our studies provide further understanding of the development of coherence in superfluids , and may allow for direct investigation of universal phase - transition dynamics .
* spontaneous vortex formation in superfluids is intimately connected to superfluid growth . in one model , illustrated in fig .
[ suppfig ] , isolated superfluid regions of characteristic size @xmath0 independently form as the system nears the critical point of the phase transition .
these regions with random relative phases merge together during the transition , leading to a continuous phase _
gradient _ in the merged fluid . due to wave function continuity requirements ,
the merging process may trap phase loops of @xmath1 if the merging regions have suitable relative phases , as illustrated in fig .
[ suppfig ] .
the superfluid density at the center of these @xmath1 phase loops is topologically constrained to be zero , resulting in the formation of a _ quantised vortex _ ; the absence of superfluid at the vortex core may be viewed as arising from destructive interference between merging regions . although cast here in the context of superfluid growth ,
spontaneous topological defect formation is a fundamental component of the kibble - zurek ( kz ) mechanism @xcite .
based on universality classes for second - order phase transitions , the kz mechanism provides a prescription for estimating a correlation length @xmath0 and hence the density of defects , proportional to @xmath2 , that may form . for a continuous phase transition that proceeds quasistatically
, @xmath0 diverges at the critical point and therefore no defects are expected to form .
however , in the kz mechanism the phase transition occurs over a finite time , and the system falls out of equilibrium when the thermalisation ( or relaxation ) rate drops below a quench rate @xmath3 . at this point @xmath0 is frozen in , essentially remaining constant through the critical point .
a principle result is that faster quenches lead to an earlier freeze - in time , and hence smaller values of @xmath0 and higher defect densities .
* @xmath4 schematic of spontaneous vortex formation . *
( left ) as a thermal gas ( mottled grey shade ) is cooled through the bec transition , isolated coherent regions of approximate size @xmath0 and unpredictable phase may form @xcite .
quantum phase ranges from @xmath5 to @xmath1 , represented here by shades of grey as indicated by the gradient bar at the right .
( right ) initial coherent regions eventually merge to form a single bec ( continuous greyscale region ) , potentially forming quantised vortices . here , a positive ( negative ) vortex is labeled with a cross ( circle ) , with the phase winding direction corresponding to the direction of superfluid flow and phase gradient around the vortex core . ]
the kz mechanism is appealing due to its potential for characterizing a wide variety of phase transitions , irrespective of the microscopic processes involved . a model of condensation in a homogeneous bose gas describing the transition from a weak - turbulent ( kinetic ) stage to strong - turbulent ( coherent ) state has been proposed by svistunov and co - workers @xcite . in this scenario ,
as energy is removed from a system , the low - energy atomic field modes become macroscopically occupied .
destructive interference between these essentially classical modes leads to nodes in the field , which appear as lines of zero atomic density .
subsequently , a _ quasi - condensate _ having local coherence but no long - range coherence grows around the lines of zero density , which simultaneously evolve into well - structured vortex cores .
eventually the superfluid relaxes into equilibrium and a true condensate with global phase coherence is achieved .
berloff and svistunov numerically demonstrated the validity of this scenario for the _ homogeneous _ bose gas in simulations of the gross - pitaevskii equation @xcite .
our work involves an experimental and theoretical exploration of similar phenomena in the condensation of trapped gases .
our approach also has the potential to investigate the relationship of the kz mechanism to bec phase transition dynamics . in previous work we demonstrated that vortices can form during the _ controlled _ merging of three independent becs with uncorrelated phases @xcite , an analog of the kz mechanism .
here we study vortex formation by evaporatively cooling an atomic gas through the bec phase transition in a single axially symmetric oblate harmonic trap ( see methods ) . in order to probe condensate growth dynamics under varying cooling conditions , we utilise two temperature quenches :
quench a uses a 6-s radio - frequency ( rf ) evaporative cooling ramp , and quench b uses a sudden rf jump to a final value .
plots of temperature and condensate number versus time for both quenches are shown in fig .
[ fig1]a . following anglin and
zurek @xcite ( see supplementary information ) , we estimate a correlation length of @xmath6 @xmath7 m near the critical point for both quenches . because @xmath0 is about a factor of 5 _ smaller
_ than our radial harmonic oscillator length @xmath8 @xmath7 m characterising a condensate radius for small atom numbers , we would not expect global phase coherence just below the critical temperature , suggesting that spontaneous vortex formation could be observable in our experiments .
* @xmath4 condensate formation and vorticity .
* * a * , condensate number @xmath9 versus time .
blue squares ( red circles ) indicate experimental data for quench a ( b ) , and lines indicate corresponding theoretical simulations .
the green dot - dashed line is the theoretical result for the toroidal trap ( quench c ) .
vertical dotted lines indicate the observation times for which experimental statistics are generated .
inset : experimentally measured temperatures for quenches a and b ( @xmath10 7 s and 5 s , respectively ) .
* b * , the probability of finding at least one vortex passing through the @xmath11 plane plotted for all three simulated quenches .
grey bars indicate the experimental measurement range for each data set . ] to look for vortices , we suddenly remove the trapping potential after the 6-s evaporative cooling ramp of quench a or @xmath121.5 s after the rf jump for quench b. each bec ballistically expands and is then imaged along the vertical direction ( the @xmath13 axis ) , which coincides with the trap s symmetry axis .
vortex cores well - aligned with the @xmath13 axis appear as holes in the column - density distribution , as shown in fig .
[ fig2]a .
we emphasise that our procedure does not impart net angular momentum to the atomic cloud , such as through phase engineering @xcite or stirring @xcite ; our observations thus represent a new regime for the study of quantised vortex nucleation in becs ( see supplementary information for further discussion ) . * @xmath4 vortices in the harmonic and toroidal traps . *
* a * , 200-@xmath7m - square expansion images of becs created in a harmonic trap , showing single vortices ( left , centre ) and two vortices ( right ) .
* b , c * , sample simulation results from quench b , showing in - trap integrated column densities along @xmath13 ( in * b * ) and associated phase profiles in the @xmath11 plane ( in * c * ) , with vortices indicated by crosses and circles at @xmath14 phase windings . *
d * , left m - square phase - contrast experimental image of a bec in the toroidal trap .
remaining images : vortices in 200-@xmath7m - square expansion images of becs created in the toroidal trap . *
e , f * , simulations of bec growth in the toroidal trap show vortices ( as in * b*,*c * ) and persistent currents . ]
we simulate condensate formation using the stochastic gross - pitaevskii equation ( sgpe ) formalism @xcite that describes the highly - occupied , low - energy modes of a bose gas with a classical field .
the field evolves according to a generalised gross - pitaevskii equation that includes dissipation and thermal noise describing collisions between the partially condensed matter waves and the high - energy atoms in the thermal cloud .
because evaporative cooling is difficult to simulate realistically @xcite , and the details are often qualitatively unimportant , we implement an idealised cooling model with a sudden jump in chemical potential and temperature of the thermal cloud through the condensation critical point .
this leaves the sgpe classical field out of equilibrium with the thermal cloud ; the subsequent return to equilibrium results in condensate formation .
figure [ fig1]a shows the growth in condensate number for the simulations of both quenches . in our simulations , the initial and final thermal cloud parameters have been chosen to match the experimental results , and the coupling between the thermal cloud and the classical field is then adjusted to give good agreement with the experimentally observed bec growth curves .
this approach allows a meaningful comparison of other observables such as vortex statistics with the experimental data .
further discussion can be found in methods and supplementary information . as shown in figs .
[ fig2]b c , vortices spontaneously form in our simulations , where each realisation can be interpreted as the numerical analog of a single experimental run .
we therefore study vortex dynamics in each growing condensate to compare vortex formation statistics with our experimental results . in both our laboratory and numerical procedures ,
for each quench we repeat the bec creation procedure and analyse statistics of vortex observations . for each data
set described below , we extract the fraction of images showing at least one vortex core within a displacement of @xmath15 from the bec centre , where @xmath16 is the bec thomas - fermi radius in the @xmath11 plane @xcite .
this fraction serves as our estimate of the probability of observing spontaneously formed vortices in a single run . because localised decreases in the density profile of an experimentally obtained image may not always clearly indicate the presence of a core ( e.g. due to tilting or bending with respect to the @xmath13 axis ) our experimental uncertainty ranges are defined by our ability to visually determine whether or not an image shows a vortex .
for quench a , 23% to 28% of 90 total images contain at least one visible vortex core . for quench
b , 15% to 20% of 98 total images show at least one core .
although the two quenches utilise quite different rf evaporation trajectories , they exhibit similar cooling and bec growth rates .
we can thus expect statistical similarities between the two data sets .
further statistical details , including results of observing multiple cores per image , are given in supplementary information . from our simulations ,
we additionally analyse vortex observation probabilities as a function of time for each quench . to determine the presence of a vortex we consider an instantaneous slice of the classical field in the @xmath11 plane of the trap , and detect all phase - loops of @xmath17 within a displacement of @xmath15 from the bec centre ( here
, @xmath16 is based on the time - dependent condensate number ) .
we find that the majority of vortices are aligned with the @xmath13 axis of the trap .
the vortex observation probabilities obtained from @xmath12300 simulation runs for each quench are plotted against time in fig .
[ fig1]b , with comparison to the experimental statistics . according to our simulations
the number of vortices decreases with time , consistent with our model of a thermal bath that is independent of time and that has no angular momentum ; the thermodynamic final state should therefore be a condensate without vortices . in this respect
the simulations diverge from our experimental observations , where there is no significant variation of vortex observation probabilities over a timescale of a few seconds .
this low damping rate is consistent with the comparatively small thermal atomic component observed , indicating that a kinetic theory of thermal cloud dynamics may be needed to fully account for the long - time behaviour of the experiment .
we therefore compare our simulation results at @xmath18 s for quench a , and @xmath19 s for quench b , based upon experimental observations of negligible vortex damping once the bec is nearly fully formed . by focusing a blue - detuned laser beam propagating along the @xmath13 axis into the centre of the trap ( see methods )
, we experimentally studied bec growth in a toroidal potential in which a bec may display both persistent superfluid current @xcite about the central barrier , as well as free vortices circulating around the barrier .
the pinning of superfluid flow may influence both vortex dynamics during bec growth and observations of vortices after the bec is formed : a vortex pinned to the barrier reduces the likelyhood of complete self - annihilation between pairs of spontaneously formed vortices of opposite charge , thereby increasing the probability of finding a vortex in a bec .
we implement a 6-s final evaporative cooling ramp identical to quench a , and identify this data set as quench c. an _ in - situ _ image of a bec in the toroidal trap is given in the leftmost image of fig .
[ fig2]d .
note that the dark region in the bec centre indicates atoms displaced by the laser beam ; vortices are not visible in this image . after creating each bec
, we ramp down the laser power over 100 ms and immediately thereafter allow the bec to expand from the trapping potential . for these conditions
, we found 56% to 62% of 52 images contained at least one visible core ; examples are shown in fig .
[ fig2]d .
condensate formation rates were not experimentally measured for quench c ; for the simulations we use the parameters of quench a but with an additional repulsive gaussian barrier with a height of 33 @xmath20
. simulated condensate growth versus time resulted in smaller condensates compared to quench a ( also observed experimentally ) as shown as the green dot - dashed curve in fig .
[ fig1]a .
examples of numerically obtained column density and phase are shown in figs .
[ fig2]e f .
vortex observation statistics for 300 runs are plotted as a green dot - dashed line in fig .
[ fig1]b ; we find that the vortex observation probability is approximately twice that of the harmonic trap of quench a , as is also the case with the experimental data , but somewhat lower in overall magnitude than the experimental observations .
in contrast to the harmonic case , the curve does not exhibit decay below 40% this corresponds to vortices pinned by the central barrier .
additional toroidal trap statistics are provided in supplementary information . in the supplementary information , we provide movies of simulated condensate formation for quenches a and c. here we describe one run in which a single vortex persists to the final time step for a quench a simulation in the harmonic trap .
after the system temperature is initially lowered , the atomic field density profile fluctuates temporally and spatially , as illustrated in fig .
[ figmovieframes]a .
a bulk bec then begins to grow , and a tangle of vortices is trapped within the bec as shown in fig .
[ figmovieframes]b , in qualitative agreement with the models of superfluid turbulence @xcite and the kz mechanism @xcite . at later stages , a nearly uniform condensate exists with clear vortex cores , as shown in fig . [ figmovieframes]c .
this state eventually damps to a single core , seen in fig .
[ figmovieframes]d .
* @xmath4 bec growth dynamics . *
* a d * , four snapshots during the simulated growth of a bec showing isodensity surfaces ( in light red ) in a three - dimensional rendering .
vortex cores of opposite charges about the @xmath13 axis are indicated as magenta and cyan lines .
the corresponding times are * a * , 0.13 s ; * b * , 0.45 s ; * c * , 0.57 s ; * d * , 1.57 s , where @xmath210 s is the time when the quench is initiated in the simulation .
the full movie from which these images were taken is provided as supplementary video 3 . ] in examining the relationship between the bec transition and the kz mechanism , one would ideally study vortex formation with widely varying bec growth rates in order to test the predictions of the scaling of the vortex density @xcite .
however , in our harmonic traps , we have only succeeded in increasing bec growth rates by a factor of two to three , resulting in a factor of @xmath122 increase in vortex formation compared with quenches a and b. in simulations , faster growth rates can potentially result in more vortices , as discussed in supplementary information .
in addition to providing new experimental observations , our work places spontaneous topological defect formation on a theoretical foundation that has not been available in analogous studies in other systems .
related experimental investigations of spontaneous symmetry breaking of quenched ferromagnetic spinor becs @xcite may also yield insight into phase transition dynamics .
the quantitative agreement between our experimental and theoretical results is of primary importance for their mutual interpretation : even in the ultra - cold bec phase transition , thermal fluctuations can play an important role , and spontaneous topological defect formation may be virtually unavoidable in some situations .
our continuing work will explore in greater detail exactly how a condensate forms in this regime ; while the superfluid turbulence model describes vortex formation during condensation , to what extent does the kz mechanism s universality relate to the bec transition ?
with further simulations and experiments , new understanding of the development of coherence in the birth of a superfluid may be uncovered , a tantalizing prospect addressing the interface between the classical and quantum worlds . |
the single - band hubbard model , ref .
[ 1 ] , plays in the solid state physics a similar principal role as the hydrogen atom in the atomic physics .
this explains a continuous interest in its properties .
the hubbard hamiltonian reads @xmath6 here @xmath7 @xmath8 are annihilation ( creation ) operators of an electron with spin @xmath9 , @xmath10 in the wannier representation at the lattice site @xmath11 and @xmath12 .
moreover , @xmath13 is the hopping integral between different lattice sites @xmath14 and @xmath15 ( @xmath16 and @xmath0 is the intrasite coulomb repulsion .
the bloch conduction band energy @xmath17 is given by @xmath18 in the following we restrict ourselves to the simple cubic ( sc ) lattice and assume that @xmath19 then @xmath20 the hopping parameter @xmath3 is simply related to the bandwidth @xmath21 of the bloch band ( 2 ) , e. g. @xmath22 for the sc lattice@xmath23 the interplay between the two model parameters , @xmath21 and @xmath24 is decisive for the properties of the model resulting in strong electron correlations , leading to band magnetism ( see e.g. refs [ 2 ] , [ 3 ] for a review ) , insulator - to - metal transition ( see e.g. refs [ 3 ] , [ 4 ] and papers cited therein ) and high-@xmath25 superconductivity ( negative @xmath0-model , see e.g. ref . [ 5 ] ) .
the hubbard model very often plays also a role of a submodel for many other more complicated models ( as e.g. anderson model ( see ref . [ 6 ] ) , s - f model ( see refs [ 7,8 ] ) and so on ) .
especially interesting , but difficult to handle are the properties of the hubbard model in such a regime of the model parameters where the bandwidth @xmath21 is comparable to the coulomb repulsion @xmath0 . in a large number of papers [ 9 - 30 ] the authors tried to solve this model using many sophisticated methods .
the exact solution , however , does not exist till now .
many authors tried to change the situation in this field by introducing the expansion parameter @xmath2 @xmath26 .
this idea ( cf refs [ 31,32 ] ) consists in replacing difficult physics connected with the model by
difficult mathematics obtained by a laborious expansion with respect to @xmath27 different methods connected with this problem have been applied as e.g. the perturbation expansion ( see refs [ 18 ] , [ 21 ] ) , canonical transformation ( see refs [ 9 ] , [ 12 ] , [ 17 ] , [ 22 ] , [ 23 ] ) or ab initio derivations ( see refs [ 25 ] , [ 28 - 30 ] ) .
most of the methods , leading to the @xmath28 model ( or generalized @xmath1 model ) are also summarized in refs [ 27 ] , [ 4 ] .
the goal of the present paper is just to show that we can take another , nonperturbative way . in the first step we divide the crystal lattice into a set of interacting dimers . in other words
, we can rewrite the hubbard hamiltonian ( 1 ) for the sc lattice ( see fig .
1 ) in the equivalent form:[multiblock footnote omitted ] @xmath29 where @xmath30 the indices @xmath31 and @xmath32 enumerate the dimers and @xmath33 is the dimer hamiltonian .
the second term in ( 4 ) describes the hopping between nearest dimers in the @xmath34-direction , the third one represents the hopping between nearest dimers ( ( @xmath35)-plane ) and between different dimer planes ( see fig .
the equivalent form of the hubbard hamiltonian ( 4 ) is especially suitable because we can apply in the following the exact solution of the dimer problem ( 5 ) . in the next step
we express the construction operators @xmath36 ( @xmath37 ) as linear combinations of the transition operators between different dimer states .
this , in turn , allows to find the exact dimer representation of the whole hamiltonian ( 4 ) .
the space of the dimer eigenvectors consists , however , of two subspaces .
one of them corresponds to the lowest lying energy levels , the second one contains the levels with energies which in the large @xmath0 limit take on large , positive values .
these levels in a reasonable temperature range can not be occupied by electrons and therefore we can exclude them from further considerations .
it is interesting to note that this approch , without any additional assumptions , applied to the dimer hamiltonian ( 5 ) produces the analogy of the well known @xmath1 model ( cf refs [ 31 ] , [ 32 ] ) .
a similar approach can be applied to the hamiltonian ( 4 ) , describing the crystal as a whole . by eliminating from the considerations the unoccupied dimer energy levels in the large @xmath0 limit ( it is the only assumption ) with the use of the projection technique we can find in a straightforward way the final form of the hamiltonian in this limit without using perturbation expansion or canonical transformation .
the resulting hamiltonian , obtained in this way , is very complicated .
it , however , explicitly shows all possible magnetic , nonmagnetic or more complex competitive interaction processes very deeply hidden in the original hamiltonian written in the site representation ( 1 ) .
it is the aim of this paper just to reveal these important but normally invisible elementary interactions .
one important advantage might be that a given approach to the unsolvable hubbard problem can be tested with respect to the types of neglected interaction processes .
besides , the new and straightforward method , presented in this paper , can easily be adopted to clusters of any size and also to another types of model hamiltonians .
the paper is organized as follows . in sec .
2 we find the exact solution of the hubbard dimer ( 5 ) and give the exact expressions for the annihilation operators @xmath38 in the dimer representation . in sec .
3 we derive the hubbard dimer hamilonian ( 5 ) in the large @xmath0 limit . with the use of the projection technique onto the lowest lying dimer states we derive in sec .
4 the hubbard hamiltonian for the crystal in this limit ( central formula of this paper ) . a simplified version of the derived hamiltonian with the use of the taylor expansion with respect to @xmath2 @xmath26 and the case @xmath5 is discussed in sec .
the eigenvalues and eigenvectors of the dimer hamiltonian ( 5 ) can be found by using a standard procedure ( cf refs [ 33 - 35 ] ) .
we start with the vectors @xmath39 @xmath40 ; @xmath41 which constitute the fock basis of the single - dimer space of states : @xmath42 starting with the vectors ( 6 ) we easily get the eigenvalues @xmath43 and the eigenvectors @xmath44 of the hubbard dimer ( 5 ) . here , we only mention that the space of 16 eigenvectors @xmath44 can be devided into some subspaces numbered by @xmath45 .
the subspace , belonging to @xmath46 and @xmath47 is 1-dimensional ( @xmath48 @xmath49 @xmath50 @xmath51 .
there are , however , two 2-dimensional subspaces corresponding to @xmath52 ( as e.g. @xmath53 @xmath54 etc . ) and two 2-dimensional subspaces corresponding to @xmath55 ( as e.g. @xmath56 @xmath57 etc . ) .
the subspace , belonging to @xmath58 consists of one 4-dimensional subspace ( as e.g. @xmath59 @xmath60 etc . ) and two 1-dimensional subspaces ( @xmath61 @xmath62 ) .
the complete set of the eigenvalues @xmath63 and eigenvectors @xmath44 of the hubbard dimer is given in the appendix a. it allows to express the dimer hamiltonian ( 5 ) in the equivalent form @xmath64 the next important step in our calculations is the possibility to express the annihilation operators @xmath65 as linear combinations of transition operators between dimer states @xmath66 this procedure can easily be performed when acting with the annihilation operators on the basis vectors ( 6 ) and using the reciprocal relations to ( a.1 ) . in this way we can obtain the dimer representation of the annihilation ( creation ) operators , given in appedix b. after this operation we can insert so prepared @xmath38 @xmath67 into the hubbard hamiltonian ( 4 ) to obtain the hubbard model in the dimer representation for the sc lattice .
this representation will be used later to derive the hubbard hamiltonian in the large @xmath0 limit ( see sec .
looking at the eigenvalues ( a.1 ) of the hubbard dimer it is easy to see that for large @xmath0 @xmath68 the energies @xmath69 and @xmath70 take on large , positive values , producing in the partition function the terms which can practically be neglected .
it means that the mentioned energies are not occupied in the reasonable temperature range ( @xmath71 @xmath72 @xmath73 ) and can be excluded from our considerations .
therefore the dimer hamiltonian , given by ( 7 ) , reduces to @xmath74 to bring this expression into a compact ( second quanization ) form we introduce the hubbard operators @xmath75 and spin operators @xmath76 where @xmath77 @xmath78 @xmath79 , @xmath80 with the use of ( 9)-(13 ) the hamiltonian of the hubbard dimer ( 8) for large @xmath81 can be presented in the form ( we introduce the omitted earlier dimer index @xmath31 ) @xmath82 \nonumber \\ & & \nonumber \\ & & + \frac{4t^2}{u\sqrt{1+(\frac{4t}u)^2}}[\overrightarrow{s}_{i,1}\cdot \overrightarrow{s}_{i,2}-\frac{n_{i,1}^an_{i,2}^a}4 ] \nonumber \\ & & \nonumber \\ & & + \frac{t^2(1-\sqrt{1+(\frac{4t}u)^2})}{u(1+\sqrt{1+(\frac{4t}u)^2})\sqrt{% 1+(\frac{4t}u)^2}}[2(b_{i,1,\uparrow } ^{+}a_{i,1,\downarrow } ^{+}a_{i,2,\downarrow } b_{i,2,\uparrow } \nonumber \\ & & \nonumber \\ & & + b_{i,2,\uparrow } ^{+}a_{i,2,\downarrow } ^{+}a_{i,1,\downarrow } b_{i,1,\uparrow } ) \\ & & \nonumber \\ & & + ( 1-n_{i,1}^a)n_{i,2}^b+(1-n_{i,2}^a)n_{i,1}^b - n_{i,1}^bn_{i,2}^b ] \nonumber \\ & & \nonumber \\ & & + \frac{t(1-\sqrt{1+(\frac{4t}u})^2)}{2\sqrt{1+(\frac{4t}u)^2}}% \sum\limits_{\sigma } \sum\limits_{\alpha = 1,2}[a_{i,\alpha , \sigma } ^{+}b_{i,% \overline{\alpha } , \sigma } + b_{i,\alpha , \sigma } ^{+}a_{i,\overline{\alpha } % , \sigma } ] \nonumber\end{aligned}\ ] ] where @xmath83 @xmath84 @xmath85 , @xmath86 when @xmath87 and @xmath88 when @xmath89 .
it is very important to stress that the formula ( 14 ) has been obtained in a nonperturbative way , starting from the exact form of the dimer hamiltonian ( see ( 5 ) or ( 7 ) ) , and excluding from the considerations unoccupied dimer energy levels in the large @xmath0 limit .
let us note that using the taylor expansion in ( 14 ) with respect to @xmath2 and retaining the terms proportional to @xmath4 we obtain @xmath90+\frac{4t^2}u[\overrightarrow{s}% _ { i,1}\cdot \overrightarrow{s}_{i,2}-\frac{n_{i,1}^an_{i,2}^a}4].\ ] ] the formula ( 15 ) is the well - known @xmath1 model for the hubbard dimer where the first part in ( 15 ) , similarly to ( 14 ) , represents the exact form of the dimer hamiltonian ( 5 ) in the limit @xmath5 . the same result ( 14 )
can also be obtained applying a more general approach which will be used later to derive the hubbard hamiltonian for large @xmath0 in the case of a crystal .
let us note that after elimination of the unoccupied levels the subspace of the eigenvectors of the hubbard dimer ( a.1 ) for large @xmath0 consists of the following eigenvectors : @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , and @xmath99 .
it means that we can define a projection operator onto this subspace @xmath100 which , in the second quantization form , reads @xmath101 \end{array } \\ & \\ & + \frac t{u\sqrt{1+(\frac{4t}u)^2}}\sum\limits_{\sigma } \sum\limits_{\alpha = 1,2}(a_{i,\alpha , \sigma } ^{+}b_{i,\overline{\alpha } , \sigma } + b_{i,\alpha , \sigma } ^{+}a_{i,\overline{\alpha } , \sigma } ) .
\end{array}\ ] ] now , the hamiltonian ( 14 ) can also be obtained from ( 5 ) ( or ( 7 ) ) with the use of the projection operator ( 16 ) or ( 17 ) using the relation @xmath102 and applying a straightforward but laborious algebraic calculation .
the hamiltonian of the whole crystal ( 4 ) can be expressed ( similar to ( 7 ) ) in the form @xmath103 with unknown energies @xmath104 and eigenvectors @xmath105 .
we can , however , expand the eigenvectors @xmath106 in the series of the dimer eigenvectors ( see ( a.1 ) ) @xmath107 assuming that the crystal consists of @xmath108 dimers . using ( 19 ) and ( 20 )
we obtain @xmath109 it is clear that to obtain the hubbard hamiltonian for large @xmath0 we have to project ( 21 ) onto the subspace of the lowest lying dimer states with the use of the projection operator @xmath110 where @xmath111 is given by ( 16 ) or ( 17 ) .
in analogy to ( 14 ) we denote the hubbard hamiltonian in the large @xmath0 limit by @xmath112 .
similar to ( 18 ) we write @xmath113 and instead of the form ( 21 ) for the hamiltonian @xmath114 we can use ( 4 ) . taking into account that @xmath115 @xmath116 @xmath117 = 0)$ ] we obtain @xmath118\equiv p\overline{\overline{h } } \end{array}\ ] ] where @xmath119 is given by ( 14 ) and @xmath85 @xmath120 @xmath121 applying the projection operator ( 16 ) or ( 17 ) to ( b.1 ) - ( b.4 ) and introducing hubbard- and spin operators ( 9 ) - ( 13 ) we obtain @xmath122 -\delta \left [ s_{i,1}^za_{i,2,\uparrow } + s_{i,1}^{-}a_{i,2,\downarrow } \right ] , \ ] ] @xmath123 + \delta \left [ s_{i,1}^za_{i,2,\downarrow } -s_{i,1}^{+}a_{i,2,\uparrow } \right]\ ] ] where the corresponding expressions for @xmath124 ( @xmath125 ) can easily be obtained by changing the internal dimer index @xmath126 in ( 27 ) and ( 28 ) .
the new operators @xmath127 are introduced to obtain a relatively compact form of ( 27 ) and ( 28 ) .
they are defined as follows @xmath128 \end{array}\ ] ] where @xmath129 @xmath130 and @xmath131 @xmath132 to write down the explicit form of the hubbard hamiltonian @xmath133 ( see ( 24 ) ) we have to insert ( 27 ) and ( 28 ) into @xmath133 .
this operation leads , however , to a very complicated form of @xmath134 given in the appendix c ( a simplified form of this hamiltonian is discussed in the next section ) . here
again the only assumption made to derive @xmath135 was the reduction of the whole dimer space ( a.1 ) to the subspace of the dimer eigenvectors @xmath136 , @xmath137 , @xmath138 , @xmath139 , @xmath140 , @xmath141 , @xmath142 , @xmath143 and @xmath144 , corresponding to the lowest lying dimer energy levels because in the large @xmath0 limit only these levels can be occupied . the hamiltonian @xmath135 , obtained in this way , contains many competing , magnetic , nonmagnetic and more complex interactions . among them
we can find a direct antiferromagnetic interaction generated by the term @xmath145 ( heisenberg exchange interaction ) multiplied by the positive coupling constant .
such a term appears in @xmath119 ( see ( c.6 ) and ( 14 ) ) . inside of @xmath135 ( see ( c.6 ) and ( c.2 ) - ( c.5 ) )
a kind of ferromagnetic interactions between spins from different dimers , represented by ( c.3 ) , appears within the terms proportional to @xmath146 and @xmath147 ( negative coupling constants ) .
the antiferromagnetic interactions , however , appear again in terms proportional to @xmath148 .
there are also many other magnetic , more exotic interactions , represented by ( c.2 ) , ( c.4 ) and ( c.5 ) , entering into ( c.6 ) .
the situation is , however , much more complicated when we consider the total hamiltonian @xmath112 ( see ( 24 ) ) in the large @xmath0 limit .
@xmath112 differs from @xmath135 by the multiplicative factor @xmath149 ( a product of the projection operators @xmath150 ( see ( 22 ) and ( 17 ) ) , standing on the left .
inside of each @xmath150 the mentioned antiferromagnetic interaction also appears ( see ( 17 ) ) . in other words , the total hamiltonian @xmath112 ,
we are interested in , is actually a sum of the products of many competitive ferromagnetic , antiferromagnetic and more exotic interactions .
the thermodynamic properties of the system , described by the hamiltoniam @xmath112 ( 24 ) , are then a result of the competition between all of them .
which interaction wins in such a competition it certainly depends on temperature , model parameters ( @xmath151 ) and on the average number of electrons per lattice site , determining the chemical potential of the system .
the formalism presented in this paper is also applicable to a more complicated decomposition of the hubbard hamiltonian ( 1 ) into a set od interacting clusters consisting e.g. of one central atom and @xmath34 its nearest neighbours .
we , however , know ( see e.g. refs [ 36]-[39 ] and papers cited therein ) that , unfortunately , the mathematical problems in this case exponentially grows up with the size of the cluster .
the complicated form of the hubbard hamiltonian @xmath112 in the large @xmath81 limit ( see ( 24 ) , ( c.6 ) ) where @xmath149 is given by ( 22 ) ( see also ( 17 ) ) can essentially be reduced when applying the taylor expansion with respect to the parameter @xmath152 . to do it we have to expand all the coefficients in ( 14 ) , ( 17 ) and ( c.6 ) including also @xmath153 and @xmath154 ( see ( 32 ) , ( 33 ) ) .
such an expansion can be performed to any power of @xmath155 , however , the most simple form we obtain when we restrict ourselves to the linear approximation , resulting in the terms proportional to @xmath3 and @xmath156 the accuracy of this expansion can easily be verified assumming e.g. a typical value of the ratio @xmath157 ( or less ) . because the bandwidth of the conduction band for the sc lattice is @xmath158 it results in a small value of the expansion parameter @xmath159 in this case .
it , however , means that the linear approximation is quite reasonable because all higher terms in the expansion , proportional to @xmath160 ( @xmath161 produce @xmath162 times smaller contribution . to present the results of the taylor expansion including all the terms proportional to @xmath3 and @xmath4
let us first define several auxiliary quantities @xmath163 @xmath164 where @xmath165 @xmath166 and ( @xmath167 @xmath168 ) @xmath169 @xmath170 where @xmath171 is given by ( 31 ) .
the hamiltonian @xmath135 ( c.6 ) , including all the terms proportional to @xmath3 and @xmath4 , takes on the form @xmath172 where @xmath173 \nonumber \\[0.2 cm ] & & -t\sum\limits_{i,\sigma } \left [ \widetilde{a}_{i,2,\sigma } ^{+}\widetilde{% a}_{i+1,1,\sigma } + \widetilde{a}_{i+1,1,\sigma } ^{+}\widetilde{a}% _ { i,2,\sigma } \right ] \\[0.2 cm ] & & -t\sum\limits_{i\neq j,\sigma } \sum\limits_{\alpha = 1}^2\widetilde{a}% _ { i,\alpha , \sigma } ^{+}\widetilde{a}_{j,\alpha , \sigma } \nonumber\end{aligned}\ ] ] and @xmath174 \nonumber \\ & & -\frac{t^2}u\sum\limits_{i,\sigma } [ \widetilde{a}_{i,2,\sigma } ^{+}% \widetilde{\widetilde{a}}_{i+1,1,\sigma } + \widetilde{\widetilde{a}}% _ { i,2,\sigma } ^{+}\widetilde{a}_{i+1,1,\sigma } \nonumber \\ & & + \widetilde{a}_{i+1,1,\sigma } ^{+}\widetilde{\widetilde{a}}_{i,2,\sigma } + % \widetilde{\widetilde{a}}_{i+1,1,\sigma } ^{+}\widetilde{a}_{i,2,\sigma } ] \nonumber \\ & & -\frac{t^2}u\sum\limits_{i\neq j,\sigma } \sum\limits_{\alpha = 1}^2[% \widetilde{a}_{i,\alpha , \sigma } ^{+}\widetilde{\widetilde{a}}_{j,\alpha , \sigma } + \widetilde{\widetilde{a}}_{i,\alpha , \sigma } ^{+}\widetilde{a}% _ { j,\alpha , \sigma } ] \\ & & + \frac{2t^2}u\sum\limits_i[\overrightarrow{s}_{i,2}\cdot ( \overrightarrow{% \underline{\underline{s}}}_{i,1;i+1,1}+\overrightarrow{\underline{s}}% _ { i+1,1;i,1 } ) \nonumber \\ & & + \overrightarrow{s}_{i+1,1}\cdot ( \overrightarrow{\underline{\underline{s}}% } _ { i+1,2;i,2}+\overrightarrow{\underline{s}}_{i,2;i+1,2 } ) ] \nonumber \\ & & + \frac{2t^2}u\sum\limits_{i\neq j}[\overrightarrow{s}_{i,1}\cdot ( % \overrightarrow{\underline{\underline{s}}}_{i,2;j,1}+\overrightarrow{% \underline{s}}_{j,1;i,2 } ) \nonumber \\ & & + \overrightarrow{s}_{i,2}\cdot ( \overrightarrow{\underline{\underline{s}}}% _ { i,1;j,2}+\overrightarrow{\underline{s}}_{j,2;i,1 } ) ] .
\nonumber\end{aligned}\ ] ] the operators @xmath175 and @xmath176 in ( 42 ) retain their forms introduced in ( c.1 ) but @xmath177 in ( c.1 ) should actually be replaced by @xmath178 , defined by ( 38 ) .
the total hamiltonian @xmath112 in the large @xmath81 limit ( 24 ) including all terms proportional to @xmath3 and @xmath4 can thus be written in the form @xmath179 the first part , @xmath180 , contains the terms proportional to @xmath3 whereas the term in the parentheses is proportional to @xmath181 ( see ( 35 ) and ( 42 ) ) . in the expression for @xmath182(see ( 42 ) )
the first term describes the antiferromagnetic , heisenberg intradimer interaction ( see also the second term in ( 15 ) ) .
the apearence of this term may suggest that such an interaction should also arise between different dimers .
this is of course the case . however , because of applied procedure such terms do not appear explicitly . in the last analysis
our method treats all interactions within and between dimers , on the same quality level , i.e. all interactions are taken into account .
the magnetic interdimer interactions , represented by the fourth and fifth term in ( 42 ) , have formally the same structure as the heisenberg interactions but instead of the scalar products of spin operators there are the products of spin operators and hopping spin operators ( defined in c.1 ) .
all the terms presented in ( 42 ) are correct because they originate from the exact decomposition of the hubbard hamiltonian ( 1 ) into a set of interacting dimers ( 4 ) where each dimer problem has been exactly solved ( exact dimer representation of the construction operators after applying the projection procedure , given by ( 27 ) and ( 28 ) ) .
the total hamiltonian @xmath112 ( 43 ) , we are interested in , is much more complicated than @xmath183 and @xmath184alone ( see ( 41)-(43 ) ) because of the presence of the projection operators @xmath185 and @xmath186 ( see ( 34)-(37 ) ) in the expression ( 43 ) .
it is also interesting to see what happens in the special case when taking the limit @xmath5 .
the second term in the parentheses of eq.(43 ) vanishes in this limit ( cf ( 35 ) and ( 42 ) ) . besides
, each lattice site can not be at the same time occupied by two electrons what is equivalent to the assumption ( @xmath187 @xmath188 @xmath189 and ( cf ( 38 ) ) @xmath190 the total hamiltonian ( 43 ) is thus given by a simple formula @xmath191 \\ & \\ & -t\sum\limits_{i,\sigma } \left [ a_{i,2,\sigma } ^{+}a_{i+1,1,\sigma } + a_{i+1,1,\sigma } ^{+}a_{i,2,\sigma } \right ] \\ & \\ & -t\sum\limits_{i\neq j,\sigma } \sum\limits_{\alpha = 1}^2a_{i,\alpha , \sigma } ^{+}a_{j,\alpha , \sigma } .
\end{array}\ ] ] going back to the original lattice ( cf ( 4),(5 ) and ( 1 ) ) it is easy to see that the exact hubbard hamiltonian in the limit @xmath5 takes on the form @xmath192 where @xmath14 and @xmath15 ( as before ) number the lattice points and @xmath193 @xmath194 are the hubbard operators , defined by ( 9 ) .
using a new , nonperturbative approach , basing on the equivalent form of the hubbard hamiltonian , represented by the collection of interacting dimers ( 4 ) where each dimer problem has been exactly solved , we have expressed the annihilation ( creation ) operators ( b.1)-(b.4 ) as linear combinations of transition operators between different dimer states ( dimer representation ) .
this method made it possible to exclude from the considerations the unoccupied dimer energy levels in the large @xmath0 limit by means of the projection technique resulting in the final hamiltonian for the dimer itself ( 14 ) , as well as , for the crystal as a whole ( see ( 23 ) , ( 24 ) and ( c.6 ) ) .
it is important to stress that the elimination of the unoccupied dimer energy levels was the only assumption to derive the hubbard hamiltonian ( 24 ) in the large @xmath0 limit .
therefore we can be sure that expanding ( 24 ) with respect to @xmath195 ( taylor expansion ) we obtain absolutely all terms proportional to @xmath160 @xmath196 in other words all the coefficients proportional to @xmath160 are easy to control what is not always the case when using another methods .
the final form of the obtained hamiltonian in the large @xmath0 limit ( see ( 23 ) , ( 24 ) and ( c.6 ) ) visualizes high degree of complexity of the model , deeply hidden in its original version , forming a mixture of many multiple ferromagnetic , antiferromagnetic and more complex interactions competing one with another .
this fact seems to be decisive for our final conclusion . because we are still dealing with the approximate solutions of the model ( the exact solution does not exist till now )
it may happen that we underestimate in this way some important interactions and overestimate the others .
it is the reason why the resulting thermodynamic properties of the hubbard model , obtained in an approximate way , so strongly depend on the quality of applied approximations .
we note in passing that the exact dimer solution may serve as a novel alloy analogy for the hubbard model , which could be treated by coherent potential approximation ( cpa ) .
it is well - known that the standard alloy analogy , based on the atomic limit , does not allow for ferromagnetism in the hubbard model .
this may change by application of the dimer solution which already accounts for a restricted hopping of the band electrons .
a corresponding study is in preparation .
@xmath205{ll } c_{1,\uparrow } = & \frac 1{\sqrt{2}}\left ( p_{0,11}+p_{0,12}\right ) + \frac 1{% \sqrt{2}}p_{11,25}-\frac 1{\sqrt{2}}p_{12,25 } \\ & + \frac 12\left ( p_{13,21}+p_{13,22}\right ) + \frac 1{\sqrt{2}}\left ( bp_{13,23}+ap_{13,24}\right ) \\ & -\frac 12\left ( p_{14,21}-p_{14,22}\right ) + \frac 1{\sqrt{2}}\left ( ap_{14,23}-bp_{14,24}\right ) \\ & + \frac 12\left ( p_{21,33}-p_{21,34}\right ) -\frac 12\left ( p_{22,33}+p_{22,34}\right ) \\ & + \frac 1{\sqrt{2}}\left ( ap_{23,33}+bp_{23,34}\right ) -\frac 1{\sqrt{2}% } \left ( bp_{24,33}-ap_{24,34}\right ) \\ & + \frac 1{\sqrt{2}}\left ( p_{26,31}-p_{26,32}\right ) + \frac 1{\sqrt{2}% } p_{31,4}+\frac 1{\sqrt{2}}p_{32,4 } , \end{array } \tag{b.1}\ ] ] @xmath205{ll } c_{1,\downarrow } = & \frac 1{\sqrt{2}}\left ( p_{0,13}+p_{0,14}\right ) + \frac 1{\sqrt{2}}p_{13,26}-\frac 1{\sqrt{2}}p_{14,26 } \\ & + \frac 12\left ( p_{11,21}-p_{11,22}\right ) -\frac 1{\sqrt{2}}\left ( bp_{11,23}+ap_{11,24}\right ) \\ & -\frac 12\left ( p_{12,21}+p_{12,22}\right ) -\frac 1{\sqrt{2}}\left ( ap_{12,23}-bp_{12,24}\right ) \\
& -\frac 12\left ( p_{21,31}-p_{21,32}\right ) -\frac 12\left ( p_{22,31}+p_{22,32}\right ) \\ & + \frac 1{\sqrt{2}}\left ( ap_{23,31}+bp_{23,32}\right ) -\frac 1{\sqrt{2}% } \left ( bp_{24,31}-ap_{24,32}\right ) \\ & -\frac 1{\sqrt{2}}\left ( p_{25,33}-p_{25,34}\right ) -\frac 1{\sqrt{2}% } p_{33,4}-\frac 1{\sqrt{2}}p_{34,4 } , \end{array } \tag{b.2}\ ] ] @xmath205{ll } c_{2,\uparrow } = & \frac 1{\sqrt{2}}\left ( p_{0,11}-p_{0,12}\right ) -\frac
1{% \sqrt{2}}p_{11,25}-\frac 1{\sqrt{2}}p_{12,25 } \\ & -\frac 12\left ( p_{13,21}+p_{13,22}\right ) + \frac 1{\sqrt{2}}\left ( bp_{13,23}+ap_{13,24}\right ) \\ & -\frac 12\left ( p_{14,21}-p_{14,22}\right ) -\frac 1{\sqrt{2}}\left ( ap_{14,23}-bp_{14,24}\right ) \\ & -\frac 12\left ( p_{21,33}+p_{21,34}\right ) + \frac 12\left ( p_{22,33}-p_{22,34}\right ) \\ & + \frac 1{\sqrt{2}}\left ( ap_{23,33}-bp_{23,34}\right ) -\frac 1{\sqrt{2}% } \left ( bp_{24,33}+ap_{24,34}\right ) \\ & -\frac 1{\sqrt{2}}\left ( p_{26,31}+p_{26,32}\right ) + \frac 1{\sqrt{2}% } p_{31,4}-\frac 1{\sqrt{2}}p_{32,4 } , \end{array } \tag{b.3}\ ] ] @xmath205{ll } c_{2,\downarrow } = & \frac 1{\sqrt{2}}\left ( p_{0,13}-p_{0,14}\right ) -\frac
1{\sqrt{2}}p_{13,26}-\frac 1{\sqrt{2}}p_{14,26 } \\ & -\frac 12\left ( p_{11,21}-p_{11,22}\right ) -\frac 1{\sqrt{2}}\left ( bp_{11,23}+ap_{11,24}\right ) \\ & -\frac 12\left ( p_{12,21}+p_{12,22}\right ) + \frac 1{\sqrt{2}}\left ( ap_{12,23}-bp_{12,24}\right ) \\ & + \frac 12\left ( p_{21,31}+p_{21,32}\right ) + \frac 12\left ( p_{22,31}-p_{22,32}\right ) \\ & + \frac 1{\sqrt{2}}\left ( ap_{23,31}-bp_{23,32}\right ) -\frac 1{\sqrt{2}% } \left ( bp_{24,31}+ap_{24,32}\right ) \\
& + \frac 1{\sqrt{2}}\left ( p_{25,33}+p_{25,34}\right ) -\frac 1{\sqrt{2}% } p_{33,4}+\frac 1{\sqrt{2}}p_{34,4 } \end{array } \tag{b.4}\ ] ] the operators @xmath215 , @xmath216 , @xmath175 , are not strictly the spin operators , they , however , show some similarities to the true spin operators ( as e.g. ( 11)-(13 ) ) and therefore they can be called hopping spin operators .
moreover , the following abreviations have to be used : @xmath221 \\ & \\ & -2t\beta \sum\limits_i\left [ q\left ( i,1;i+1,1\right ) + q\left ( i+1,2;i,2\right ) \right ] \\ & \\ & + 2t\delta \sum\limits_i\left [ q\left ( i,2;i+1,1\right ) + q\left ( i+1,1;i,2\right ) \right ] \end{array}\ ] ] @xmath226 \\ & \\ & + 2t\beta \delta \sum\limits_{i\neq j}\sum\limits_{\mu = 1}^2[r\left ( i,\mu ; j,\overline{\mu } \right ) + r^{z,-}\left ( i,\mu ; j,\overline{\mu } % \right ) \\ & \\ & + r^{z,+}\left ( i,\mu ; j,\overline{\mu } \right ) + r^{-,+}\left ( i,\mu ; j,% \overline{\mu } \right ) ] \end{array}\ ] ] where @xmath119 is given by ( 14 ) .
it should also be noted that the operators @xmath227 @xmath228 entering in ( c.1 ) and ( c.2 ) , are in fact , dependent on @xmath153 and @xmath154 ( see ( 29 ) ) .
the decomposition of the hamiltonian @xmath135 ( c.6 ) according to the terms proportional to @xmath229 @xmath154 , @xmath146,@xmath147 and @xmath230 has thus only a formal character in order to keep the presentation of the hamiltonian @xmath135 in a compact form . | we formulate the hubbard model for the simple cubic lattice in the representation of interacting dimers applying the exact solution of the dimer problem . by eliminating from the considerations unoccupied dimer energy levels in the large @xmath0 limit ( it is the only assumption )
we analytically derive the hubbard hamiltonian for the dimer ( analogous to the well - known @xmath1 model ) , as well as , the hubbard hamiltonian for the crystal as a whole by means of the projection technique . using this approach
we can better visualize the complexity of the model , so deeply hidden in its original form .
the resulting hamiltonian is a mixture of many multiple ferromagnetic , antiferromagnetic and more exotic interactions competing one with another .
the interplay between different competitive interactions has a decisive influence on the resulting thermodynamic properties of the model , depending on temperature , model parameters and assumed average number of electrons per lattice site .
a simplified form of the derived hamiltonian can be obtained using additionally taylor expansion with respect to @xmath2 ( @xmath3-hopping integral between nearest neighbours , @xmath0-coulomb repulsion ) . as an example , we present the expansion including all terms proportional to @xmath3 and to @xmath4 and we reproduce the exact form of the hubbard hamiltonian in the limit @xmath5 . the nonperturbative approach , presented in this paper , can , in principle , be applied to clusters of any size , as well as , to another types of model hamiltonians . |
observed relationships in galaxies between central black hole mass and stellar mass ( e.g. magorrian et al . 1998 ) , velocity dispersion ( i.e. the @xmath4-@xmath5 relation ; e.g. gebhardt et al .
2000 ; ferrarese & merritt 2000 ) , or galaxy structural properties ( e.g. the black hole fundamental plane : hopkins et al . 2007a ,
b ) indicate a co - eval nature in supermassive black hole growth and star formation in galaxies .
these results have prompted a number of investigations in recent years to quantify this apparent self - regulation in star formation and black hole growth in galaxies , and their relationship to the formation and evolutionary history of the host system ( e.g. kauffmann & haehnelt , 2000 ; hopkins et al .
2006a , b ; 2007c , d ) . over the last two decades , observations of local galaxies have painted a compelling picture in which galaxy mergers provide the link between massive starbursts and central black hole growth and activity .
ulirgs ( ultraluminous infrared galaxies ) , for example , are a class of starburst galaxies with elevated infrared luminosities ( @xmath6 10@xmath7 ) which typically show signs of interactions ( e.g. downes & solomon , 1998 ; sanders et al .
1988a ; scoville et al .
2000 ) . while the intense infrared luminosity in these sources certainly owes in large part to the merger induced starbursts ( e.g. sanders et al .
1988a , b ; sanders & mirabel , 1996 ) , in some cases a contribution from a buried active galactic nucleus ( agn ) may be non - negligible .
for example , many ulirgs show spectral energy distributions ( seds ) , infrared color ratios ( e.g. f[25/60@xmath8 m ] ) , polycyclic aromatic hydrocarbon emission ( pah ) deficits , and emission line fluxes ( e.g. [ ne v ] at 14.3@xmath8 m ) consistent with central agn activity ( e.g. armus et al .
2004 , 2006 ; farrah et al . 2003 ; de grijp et al .
indeed , 35 - 50% of ulirgs with luminosity above of 10@xmath9 show optical and nir spectra consistent with agn activity ( kim , veilleux & sanders , 2002 ; tran et al .
2001 ; veilleux , kim & sanders , 1998 ) .
these results suggest that these merging systems are simultaneously undergoing a massive star formation and central black hole growth phase
. observed similarities such as these between starburst galaxies , ulirgs and quasars prompted soifer et al .
( 1987 ) and sanders et al .
( 1988a , b ) to propose an empirically derived evolutionary sequence which connects galactic starbursts to quasars through galaxy mergers . in this picture , the fueling of the central black hole and agn growth phase is realized during the major merger when gaseous inflows trigger nuclear starbursts as well as central black hole growth .
the dusty galaxy transitions from a cold , starburst dominated ulirg , to a warm , agn - dominated ulirg ( where cold and warm refer to f[25/60@xmath8 m ] ratios ) , and , as supernovae and stellar winds clear the obscuring gas and dust , to an optical quasar . in this sense , galaxy mergers provide a unique laboratory for studying the possible co - evolution of starbursts , black hole growth and activity , and spheroid formation .
recent models have provided a theoretical foundation and further evidence for a merger - driven starburst - agn connection in galaxies .
specifically , simulations by springel , di matteo & hernquist ( 2005a ) have shown that galaxy mergers can fuel large - scale gaseous inflows ( e.g. barnes & hernquist , 1991 , 1996 ) which trigger nuclear starbursts ( mihos & hernquist , 1996 ; springel et al .
2005a ) as well as promote central black hole growth ( di matteo , springel & hernquist , 2005 ) .
subsequent winds associated with the growth of central black holes can lift the veil of obscuring gas and dust , and along several sightlines produce a quasar with comparable lifetimes , luminosity functions , and observed @xmath10-band and x - ray properties to those observed ( cox et al .
, 2006b ; hopkins et al .
2005a - d ; 2006a - d ) .
the merger remnants quickly redden owing to gas depletion and the impact of feedback from star formation and black hole growth ( e.g. springel et al .
2005b ) and resemble elliptical galaxies in their kinematic and structural properties .
indeed , the population of stars formed during the starbursts accounts for the central `` excess light '' seen in ongoing mergers ( e.g. rothberg & joseph 2004 , 2006 ) and provides a detailed explanation for the luminosity profiles of old ellipticals ( e.g. kormendy et al . 2007 ; see mihos & hernquist 1994a ; hopkins et al .
2007e , f , g ) .
a consensus picture has thus been borne out from these observations and simulations over the last two decades in which so called feedback processes associated with winds from star formation and black holes act to self - regulate the growth of both the stellar and black hole masses in galaxies ( e.g. fabian , 1999 ; silk & rees , 1998 ) .
indeed , the effects of starburst and agn feedback - driven winds have been observed in local galaxies ( e.g. heckman et al .
2000 ; martin , 2005 ; rupke , veilleux & sanders , 2005a - c ; rupke & veilleux , 2005 ; tremonti , moustakas & diamond - stanic , 2007 ) , as well as those at high- ( e.g. narayanan et al . 2004 ; pettini et al . 2002 ; shapley et al .
2003 ) by way of absorption line outflows .
however , the direct effect of feedback processes ( especially from the highly efficient central agn ) on the emission properties of galaxies is not yet well characterized ( see veilleux , cecil & bland - hawthorn , 2005 for an extensive review and associated references ) . in this sense , it is important to relate theories of galaxy formation and evolution which incorporate physically motivated models of feedback processes to observational signatures of winds across the electromagnetic spectrum .
in particular , observations of molecular gas in galaxies have proven valuable in characterizing the physics related to nuclear star formation and central agn fueling in galaxy mergers as the molecular gas serves as fuel both for star formation as well as the central black hole(s ) .
for example , interferometric observations of molecular gas in ulirgs show that most typically harbor @xmath010@xmath11of molecular gas within the central 1.5 kpc ( scoville et al .
1986 ; bryant & scoville , 1999 )
. moreover , high resolution maps of dense molecular gas at submillimeter wavelengths in ulirgs have revealed kinematic structures of double nuclei in mergers ( scoville , yun & bryant , 1997 ) , bar - driven inflows ( sakamoto et al .
2004 ) , and the density structure of gas fueling the central agn ( iono et al . 2004 ) . with the increased spatial resolution and sensitivity afforded by the latest generation of ( sub)mm - wave interferometers ( e.g. the sma , carma , pdbi ) ,
a number of recent observations have been able to pioneer investigations as to the effects of galactic winds on molecular gas emission in galaxies ( e.g. iono et al .
2007 ; sakamoto , ho & peck , 2006 ; walter , wei & scoville , 2002 ) .
detections such as these are expected to become more routine in upcoming years as the alma interferometer becomes available .
an important complement to these current and forthcoming observations of molecular line emission from starburst galaxies and agn are physical models which directly relate co emission properties to galactic scale winds . in this context
, it is our aim to investigate the role that galactic winds can play on co emission properties from starburst galaxies and agn via numerical simulations .
in particular , we focus on specific signatures of winds imprinted on co morphologies and emission line profiles . in this work ,
we present self - consistent radiative transfer calculations for the emission properties of co molecular gas in gas - rich galaxy mergers which account for the winds associated with both star formation and black hole growth . in
[ section : numericalmethods ] , we describe the hydrodynamic simulations , and radiative transfer methodology . in [ section : spiral ] we provide an example of our methods by applying our radiative transfer calculations to a model of a star - forming disk galaxy . in [ section : morphology ]
, we discuss the effect of winds on observed co morphologies .
we explore the response of emission line profiles to winds in
[ section : lineprofiles ] , present a broader discussion of these results with respect to observations in [ section : observations ] , and summarize in [ section : conclusions ] . throughout this paper ,
we assume a @xmath12cdm cosmology with @xmath13=0.7 , @xmath14=0.7 , @xmath15=0.3 .
simulating molecular line emission on galaxy - wide scales has a relatively short history owing to the computational costs associated with the hydrodynamics and radiative transfer .
early works by silk & spaans ( 1997 ) and combes , maoli & omont ( 1999 ) focused on simulating molecular emission from high- objects , using idealized galaxy systems .
wada & tomisaka ( 2005 ) pioneered incorporating three - dimensional non local thermodynamic equilibrium ( non - lte ) radiative transfer calculations into hydrodynamic simulations of agn , focusing specifically on predicted emission from the circumnuclear molecular torus .
more recent works have folded non - lte radiative transfer codes into self - consistent hydrodynamic simulations of galaxy mergers .
these works have been performed in mergers scaled for low redshift ( narayanan et al .
2006a ) , as well as those appropriate for higher- systems ( greve & sommer - larsen , 2006 ; narayanan et al . 2007 ) .
here , we employ a similar methodology to that of narayanan et al .
( 2006a , 2007 ) . in this section
, we describe our methods for incorporating 3d non - lte molecular line radiative transfer calculations into hydrodynamic simulations of galaxy mergers which account for a multi - phase ism , star formation , growing black holes , and the effects of winds associated with star formation and central agn .
the hydrodynamic simulations used for this study employed a modified version of the publicly available @xmath16-body / sph code , gadget-2 ( springel , 2005 ) , adopting the fully conservative formulation of sph developed by springel & hernquist ( 2002 ) .
the methods used to construct the progenitor galaxies , as well as a detailed description of the algorithms used to simulate the physics of feedback from star formation and accreting black holes is detailed in springel , di matteo & hernquist ( 2005a ) , and we direct the reader toward this work for further information . here , we briefly summarize these methods , and describe the specifics of the modeling most pertinent to this work .
the interstellar medium ( ism ) is modeled as a multi - phase medium in which pressure feedback from supernovae heating is treated through an effective equation of state ( eos ) ( springel , di matteo & hernquist , 2005a ) .
the disk galaxies used as progenitors in these simulations employed a softened eos with softening parameter @xmath17=0.25 ( see figure 4 of springel et al .
the ism is modeled to include cold , dense gas surrounded by a hot ism , and is realized numerically through `` hybrid '' sph particles ( springel & hernquist , 2003 ) . in this formulation
, cold ism is allowed to grow through radiative cooling of the hot ism , and conversely feedback associated with star formation can evaporate cold clouds into diffuse , hot gas .
star formation follows the prescription of springel & hernquist ( 2003 ) and is constrained to fit the schmidt / kennicutt observed star formation laws ( kennicutt , 1998a , b ; schmidt , 1959 ) . in order to explore the effects of galactic winds on the molecular ism
, we include a formulation for both accreting black holes , and winds associated with massive starbursts . in our sph formalism , the black holes are included as sink particles which accrete gas from the surrounding ism .
the accretion is treated with a bondi - lyttleton - hoyle parametrization ( bondi & hoyle , 1944 , hoyle & lyttleton , 1939 ) with a fixed maximum rate corresponding to the eddington limit .
the black hole radiates such that its bolometric luminosity is given by @xmath18=@xmath19 with accretion efficiency @xmath20=0.1 .
we further assume that 5% of this energy couples to the surrounding ism such that 0.5% of the accreted mass energy in our simulations is reinjected isotropically into the ism as thermal energy ( a parameter choice which allows the merged galaxy to reproduce the local @xmath4-@xmath5 normalization [ di matteo , springel & hernquist , 2005 ] ) .
hence , the black hole energy deposition rate into the ism may be expressed as @xmath21 where @xmath22=0.005 ( cox et al . 2007 ; springel , di matteo & hernquist . , 2005a ) .
starburst winds are treated utilizing the constant wind models of springel & hernquist ( 2003 ) . in this formulation , the mass loss rate ( denoted by @xmath23 )
is assumed to be proportional to the star formation rate ( denoted by @xmath24 ) , such that : @xmath25 where @xmath26 quantifies the wind mass loading factor .
the wind is assumed to have a speed constrained by a fraction of the supernova energy ( @xmath27 ) : @xmath28 in a recent study of galactic winds in galaxy mergers , cox et al .
( 2007 ) found that winds with high mass loading factors ( @xmath291 ) may be unphysical in that they tend to prevent any starburst after the major merger .
we thus choose a moderate mass loading factor of @xmath26=0.5 , and constant wind velocity @xmath30=837 for this study .
these numbers are consistent with the favored momentum - driven wind scalings for the galaxy masses we simulate found by oppenheimer & dav ( 2006 ) . for more details concerning constraints on wind velocities and efficiencies in galactic scale simulations , we refer the reader to cox et al .
( 2007 ) .
the energy input rate into the ism from star formation may be expressed as @xmath31 where @xmath32 is the energy imparted by supernovae , per solar mass , taken here to be 1.4 @xmath3310@xmath34 ergs @xmath35 ( springel & hernquist 2003 ) .
the progenitor galaxies in our models contain a dark matter halo initialized with a hernquist ( 1990 ) profile , concentration index @xmath36=9 , spin parameter @xmath37=0.033 , circular velocity @xmath38=160 , and are bulge - less .
the galaxy s exponential disk is rotationally supported and comprises 4.1% of the total mass .
we utilize 120,000 dark matter particles , and 80,000 total disk particles , 40% of which are gaseous in nature , the remainder serving as collisionless star particles .
the total masses of the progenitor galaxies were 1.4@xmath3310@xmath7 , and the final merger produced a central black hole with mass @xmath05@xmath3310@xmath39 .
the softening lengths were 100 pc for baryons , and 200 pc for dark matter .
we have conducted five binary merger simulations .
each of the progenitor galaxies in our simulations was constructed with the aforementioned physical parameters , and the varied parameters include initial orbit and types of wind included ( e.g. starburst and/or agn ) .
identical progenitor galaxies were used in each simulation .
the initial conditions for the merger simulations are summarized in table [ table : ics ] , and we will henceforth refer to the simulations by their model name listed in column 1 of table [ table : ics ] . in order to simplify analysis throughout this paper we will largely analyze the properties of the four simulations which share the same merger orientation and utilize the coplanar simulations ( model co - bh ) for only specific comparisons .
cccccccccc bh & e & 30 & 60 & -30 & 45 & 160 & 0.4 & yes & no + no - winds & e - no & 30 & 60 & -30 & 45 & 160 & 0.4 & no & no + sb & & 30 & 60 & -30 & 45 & 160 & 0.4 & no & yes + sbbh & & 30 & 60 & -30 & 45 & 160 & 0.4 & yes & yes + co - bh & h & 0 & 0 & 0 & 0 & 160 & 0.4 & yes & no + submillimeter and millimeter wave radiation from molecules is dependent on the distribution of level populations which depend both on collisions with other molecules and atoms , as well as the incident radiation field .
the excitation of molecular gas is highly sensitive to both variations in the temperature and density distribution , as well as the incident radiation field .
there are usually large differences in the collisional densities necessary to excite different energy levels .
for example , in the case of co , the critical density to collisionally populate the j=1 state is @xmath4010@xmath41 - 10@xmath42 @xmath43 while the j=3 state typically requires @xmath4410@xmath45 @xmath43 .
the latter density is characteristic of dense cores in gmcs , whereas the former is typical of diffuse gmc atmospheres .
the non - lte radiative transfer calculations for this work were calculated during post - processing of the hydrodynamic simulations .
the spatial range considered for the radiative transfer was 12 kpc .
the simulation outputs were smoothed to a resolution of @xmath0250 pc in order to defray computational costs associated with the radiative transfer simulations . in order to model the strongly density - dependent excitation rates in our simulations ,
we have expanded the bernes ( 1979 ) non - local thermodynamic equilibrium ( lte ) monte carlo radiative transfer algorithm to include a mass spectrum of gmcs in a subgrid manner .
the molecular gas fraction in galaxies is determined by the metallicity , dust content , interstellar radiation field , density and temperature ( hollenbach , werner & salpeter , 1971 ; pelupessy , papadopoulos & van der werf , 2006 ) . however , owing to limited spatial resolution , this calculation can not be done explicitly as the location of individual stars and clouds are not know .
we therefore assume that half of the cold neutral gas mass in each grid cell is in atomic form and half in molecular , as motivated by local volume surveys of star forming galaxies ( e.g. keres , yun & young 2003 ) though note that an estimate of the molecular fraction can be obtained via sub - grid prescriptions ( e.g. pelupessy et al . 2006 ; greve & sommer - larsen 2006 ; robertson & kravtsov 2007 ) . in our formulation , the cold molecular gas is assumed to be exist in a mass spectrum of gmcs in each grid cell @xmath46 where we take @xmath47=1.8 ( blitz et al .
the gmcs are modeled as power - law spheres where the density ( @xmath48 ) is given by : @xmath49^\alpha\ ] ] the radius of the cloud is determined by the galactic gmc mass - radius relation ( rosolowsky 2005 , 2007 ; solomon et al .
1987 ) , such that the sum of the cloud density distributions in a grid cell takes the form : @xmath50 \label{eq : globaldist}\ ] ] which is used to set the initial level populations , as well as the densities involved in the collisional rate calculations . for this work
we utilize a cloud power - law index of @xmath51=2 ( walker , adams & lada 1990 ) .
observational evidence suggests that a range of power law indices in gmcs from @xmath51=1 - 2 may be appropriate ( andre , ward - thompson & motte 1996 ; fuller & myers 1992 ; ward - thompson et al .
1994 ) , although our results are not very sensitive to the choice of power law index within this range .
formally , we build an emergent spectrum by integrating the equation of radiative transfer along the line of sight : @xmath52e^{-\tau_\nu(\rm tot)}\ ] ] where @xmath53 is the frequency dependent intensity , @xmath54 is the source function , @xmath55 is the optical depth , and @xmath56 is the position along the line of sight .
if the level populations are in lte , then the source function , @xmath54 can be simply replaced by the planck function . however ,
when considering the propagation of lines through a medium with insufficient density for collisions to thermalize the level populations ( e.g. , @xmath57 ) , the effects of radiative excitation and de - excitation must be considered . in this case , the source functions must be calculated explicitly . the source function from each cloud , @xmath58 , for a given transition from upper level to lower level @xmath59 is given by @xmath60 where the level populations are assumed to be in statistical equilibrium , and determined through the rate equations : @xmath61= \\
\sum_{k > l}n_{k , m}a_{kl}+\sum_{k\neq l}n_{k , m}(b_{kl}j_\nu+c_{kl } ) \ , .\end{aligned}\ ] ] @xmath62 , @xmath63 and @xmath64 are the einstein rates for spontaneous emission , stimulated emission , and absorption , respectively , @xmath65 are the collisional rate coefficients , the indices @xmath66 and @xmath67 represent different energy levels , and @xmath68 is the mean intensity through a given grid cell : @xmath69 as is evident by the previous three equations , the problem is circular : the observed intensity depends on the source function which is determined by the level populations . when in a non - lte regime ( @xmath70 ) , collisions alone do not determine the level populations , but rather the mean intensity field ( e.g. radiation from gas in other grid cells ) plays a role as well .
the solution is achieved by means of iteration .
the initial level populations in a grid cell are estimated based on the global density distribution for all clouds in a grid cell ( equation [ eq : globaldist ] ) .
we emit model photons which represent many real photons from the mass spectrum of gmcs in each grid cell .
the model photons are given a weight @xmath71 proportional to the total number of molecules in the upper state of a given transition in a cell , and the einstein @xmath62 rate coefficient for the transition .
the photons are emitted isotropically in 3 dimensions and have line frequency randomly drawn from the line profile function : @xmath72 where the effects of the local kinetic temperature and microturbulent velocity field are accounted for via @xmath73^{\frac{1}{2}}\ ] ] the photon then takes a step over distance @xmath74 through the next grid cell . after taking a step ,
the weight is attenuated by a factor @xmath75 where the opacities are given by @xmath76 and @xmath77 the column seen by a photon in a grid cell is calculated via a monte carlo approach .
specifically , we calculate the distribution of potential column densities a photon could see through a given grid cell by simulating a sample of the subgrid cells individually on higher resolution ( 512@xmath42 ) grids . in these higher resolution grids
, we place gmcs at random locations with masses drawn from the power - law mass spectrum , and radii given by the galactic mass - radius relation .
we do this numerous times until the distribution of potential columns a photon entering at random angle sees converges .
the distribution of columns is found to scale in a self - similar manner with grid cell ( molecular ) mass .
we then explicitly draw from this distribution to determine the column seen by a model photon as it leaves and enters given grid cells . to illustrate this
, we plot an example of the distribution of column densities in figure [ figure : column_dist ] .
the photon continues to propagate through grid cells in this manner until the weight is negligible , or the photon has left the grid .
distribution of columns is derived by randomly placing clouds drawn from a mass spectrum until the distribution converges .
photons entering a grid cell see a column randomly drawn from a distribution such as this in order to simulate the actual column seen by a photon in a statistical manner .
[ figure : column_dist ] ] once a sufficiently large number of model photons have been emitted to simulate the mean radiation field , @xmath68 , the level populations can be updated via the rate equations . in the rate equations ( also commonly called the equations of statistical equilibrium ) , the terms involving the mean intensity represent the radiatively induced excitations ( absorption ) and stimulated emission in a given cell of clouds @xmath79 .
we may rewrite these equations by defining the number of excitations of a molecule in lower state @xmath66 by @xmath80 where @xmath81 is the volume of the cell and @xmath82 is the original weight of the photon .
following bernes ( 1979 ) , we then rewrite the equations of statistical equilibrium as @xmath83= \\
\sum_{k > l}n_ka_{kl}+\sum_{k\neq l}n_k\left [ \frac{g_l}{g_k}s_{lk , m'}+c_{kl}\right ] \label{eq : stateq2}\end{aligned}\ ] ] where @xmath84 are the statistical weights of the level .
after a single generation of photons has been emitted , the @xmath85 is calculated for each grid cell , @xmath79 .
we can then use the analog of the previous equation to calculate the updated level populations for the @xmath16 individual radial cells that make up the subgrid spectrum of gmcs .
the relative contribution of each of the @xmath16 cells to the total number of excitations ( @xmath85 ) in a given grid cell is determined via the column across that region .
new level populations are calculated for each of the @xmath16 cells in the spectrum of gmcs via matrix inversion and then summed to give the total populations for their parent grid cell .
new weights @xmath71 are given to a new generation of emergent photons , and the process is iterated upon until the level populations across the grid are converged .
we typically emitted 12@xmath3310@xmath78 model photons per iteration and the boundary conditions for the radiative transfer included the 2.73k microwave background .
the radiative transfer and excitation depend sensitively on the accuracy of the rate coefficients . we have obtained our coefficients from the _ leiden atomic and molecular database _ ( schoier et al .
we have tested our radiative transfer codes per publicly available benchmarks published in van zadelhoff et al .
these tests comprise `` inside - out '' collapsing spherical molecular clouds and the details of the performance of our codes under these standard tests can be found in figure 1 of narayanan et al .
( 2006b ) . owing to its relatively high abundance ( co/=1.5@xmath3310@xmath86 in the galaxy ; lee , bettens & herbst , 1996 ) compared to other common molecules ( e.g. hcn , cs and hco@xmath87 ) , @xmath7co ( hereafter , co ) is the most commonly observed molecular tracer of .
we therefore focus our modeling efforts for this work on the predicted co emission properties of galaxy mergers in the context of galactic winds , and defer investigations of emission from other molecules to future work .
this methodology has the distinct advantages of being able to simulate the effects of dense cores as well as diffuse atmospheres of clouds .
because we divide the gmcs into a series of sub - grid cells , we are not constrained to using a single density in our rate equations , and thus the collisional excitation characteristic of dense cores can be well - represented while the influence of radiation on more diffuse regions can likewise be accounted for .
moreover , we are able to include these features in our calculations without expanding the grid to computationally prohibitive sizes .
our subgrid formulation is dependent on a series of assumptions as well .
first , owing to limited spatial resolution , we are forced to assume a constant gas fraction constrained by observational surveys .
this molecular gas in a given grid cell is assumed to be all in gmcs .
the gmcs are realized as power - law spheres in our model whereas clouds are understood observationally to be fractal in nature ( e.g. elmegreen & falgarone , 1996 ) .
the clouds within each grid cell are all at the same temperature , and individually isothermal ( though temperatures are allowed to vary from grid cell to grid cell , and thus temperature gradients exist across the galaxy ) .
the temperatures in the cold phase of the ism in the hydrodynamic models are fixed at 1000k ( springel & hernquist , 2003 ) .
this choice is arbitrary and has no effect on the hydrodynamic simulations .
typically , temperatures in molecular clouds range from 10 - 100k ; temperatures of 1000k would systematically put too much weight on the collisional coefficients in the rate equations , resulting in overly excited level populations . while it is indeed tractable to run galaxy - wide hydrodynamic models which allow the temperatures to cool to lower values than those set here ( e.g. robertson & kravtsov 2007 ) , it is computationally infeasible for the large numbers of simulations presented in this work ( table [ table : ics ] ) .
thus , a subgrid model for refining the temperature structure of the ism is required .
we construct a relatively simple model for the ism temperature structure by considering the dominant heating mechanism for the gas as heating from o and b stars ( assuming a salpeter imf ) .
we sample individual grid cells on a finer ( 512@xmath42 ) resolution and place o and b stars formed in the hydrodynamic simulations in the power - law sphere clouds randomly throughout the higher resolution grid cells .
these higher resolution cells are utilized to calculate the mean column of gas and dust ( which are assumed to have the same spatial distribution in the clouds ) seen by the stellar uv flux , and the mean blackbody heating rate is derived for each grid cell .
when no o stars are present , lower mass stars dominate the heating of the gas .
owing to the large grid cell size ( @xmath0250 pc ) , and the high densities of the molecular cores in the subgrid clouds ( @xmath8810@xmath89 ) , blister regions from the few formed o stars have a negligible impact on the total molecular gas content ( hollenbach & tielens 1999 ) .
the neutral gas is further allowed to cool via metal line cooling which is calculated using a mean escape probability radiative transfer code ( escape ; kulesa 2002 ; kulesa et al .
the dominant ism coolants considered are cii , nii and atomic oxygen and carbon with abundances set at galactic values ( lee , bettens & herbst 1996 ) .
heating and excitation from a central agn ( when included ) is not considered , and is deferred to a future study . utilizing this methodology
, we find typical isothermal cloud temperatures ranging from 10k-120k ( nominally 10 - 30 k for simulations scaled for the local universe such as those presented in this work , though temperatures can approach @xmath0120 k for models of higher redshift galaxies ; narayanan et al .
it is additionally worth noting as well that variations in the temperature structure of the ism within this range are ancillary with respect to determining the excitation patterns of the molecular ism . as an example , doubling the temperature from 10k to 20k only changes the collisional rates of excitation of e.g. co j=2 to co j=3 by @xmath010% ( utilizing values taken from the leiden atomic and molecular database ; schoier et al .
the dominant effect is instead the density structure of the gmcs .
tests have shown that the excitation conditions of these models are robust while the assumed cloud power - law index and/or gmc mass spectrum index remain within observational constraints ( blitz et al .
2006 ; rosolowsky 2007 ; see also narayanan et al .
2007b ) .
120 pc ) to lower resolution runs .
the test snapshot is model bh during the peak of the quasar phase .
the emission from higher lying transitions may be underestimated in cases of extremely poor spatial resolution.[figure : sed_restest ] ] in order to test the convergence properties of the presented results , we have taken a snapshot of merger bh ( black hole winds only ) at the peak of the quasar phase , and examined the simulated co properties at a variety of spatial resolutions . in figure
[ figure : restest_image ] , we plot the co ( j=1 - 0 ) images of the sample galaxy at three resolutions , and the normalized co ( j=1 - 0 ) spectra at the same resolutions .
the detailed structure of the galaxy s morphology begins to become evident at resolutions of @xmath0240 pc , though of course shows smaller - scale features at finer resolution .
similarly , at 240 pc resolution the unresolved spectra begin to show detailed kinematic signatures in the line profile that are smoothed over in the lower spatial resolution simulations .
thus , for both image analysis and spectral line profile analysis we utilize simulations with @xmath0240 pc resolution in order to glean the maximum information while minimizing the extensive computational costs associated with galaxy - wide non - lte radiative transfer .
knowledge of the excitation properties of co requires a slightly higher spatial resolution .
while the work presented here does not nominally investigate the excitation properties of the co molecular gas to great detail , other investigations utilizing models similar to these ( e.g. narayanan et al .
2007b ) examine the excitation properties in greater detail .
we therefore briefly review the convergence of the co excitation .
the excitation properties require a higher spatial resolution for convergence , and we therefore include a higher resolution simulation ( 120 pc ) for comparison . in figure [ figure : sed_restest ] , we plot the ratio of the intensity of the highest resolution run ( 120 pc ) versus lower resolution simulations for a series of co transitions .
the results for higher lying transitions ( co@xmath904 ) are only converged for higher spatial resolution ( @xmath0160 pc ) than is typically required for the morphology or spectral line profile studies .
therefore higher resolution models must be accounted for in studies concentrating on the molecular excitation ( e.g. narayanan et al .
for the few excitation condition investigations in this work , we employ targeted higher resolution simulations . ] in the binary merger simulations presented here , after the initial passage of the merging galaxies , gas is driven into the nuclear regions owing to tidal torquing ( barnes & hernquist , 1991 , 1996 ) .
this begins a major black hole growth phase .
while the galaxies approach toward and undergo final coalescence1.6 gyr , though varies slightly from model to model . ]
( hereafter defined as the in - spiral stage ) they go through a massive starburst and lirg / ulirg phase while the central black hole grows as an enshrouded agn following the methodology developed by springel et al .
( 2005a ) . in the models which include agn feedback - driven winds , after nuclear coalescence ,
the thermal energy associated with the black hole accretion drives a powerful wind into the surrounding ism and strongly reduces the star formation rate . during much of this time , along several sightlines , the object may be viewed as an optical quasar ( hopkins et al .
2005a - d ; 2006a - d ; 2007h ) .
the galaxy then proceeds to evolve passively into a red elliptical galaxy ( springel , di matteo & hernquist , 2005b ) . in figure
[ figure : vc3vc3e_2b_bharsfr ] , we show as a reference the sfr , black hole accretion rate , and bolometric luminosity of model bh throughout the time period considered in this work . the galaxies and quasars formed via this formulation for agn feedback in galaxy merger simulations have proven consistent with observed quasar luminosity functions and lifetimes ( hopkins et al .
2006a - d , 2006a , c , d ; though see richards et al . , 2006 ) , as well as the locally observed relation ( di matteo et al .
2005 ; hopkins et al .
2007a , b ; robertson et al .
2006a , c ) .
these simulations have also shown successes in reproducing observed x - ray , ir and co patterns characteristic of ulirgs and quasars ( cox et al .
2006b chakrabarti et al .
2007a , b ; narayanan et al .
2006a , respectively ) , the bimodal galaxy color distribution ( springel , di matteo & hernquist , 2005b ; hopkins et al .
2006b ) , seyfert galaxy luminosity functions ( hopkins & hernquist , 2006 ) , the kinematic structure of merger remnants ( cox et al .
2006a ) , the sfr - co and sfr - hcn relation in galaxies ( narayanan et al .
2007b ) , and observed properties of 6 quasars ( li et al .
2007a , b ; narayanan et al . 2007a )
. 0.3 gyr period of peak star formation and black hole activity which will be considered for the remainder of this paper .
while the integrated feedback from star formation and black holes is roughly equivalent , the star formation delivers its energy in a slow and steady manner from simulation time t=0 spread out over much of the galaxy , in contrast to the near point explosion characteristic of the black hole winds .
the arrow in the bottom panel denotes the point of nuclear coalescence .
model results first published in cox et al .
( 2007).[figure : fb_energy],title="fig : " ] 0.3 gyr period of peak star formation and black hole activity which will be considered for the remainder of this paper .
while the integrated feedback from star formation and black holes is roughly equivalent , the star formation delivers its energy in a slow and steady manner from simulation time t=0 spread out over much of the galaxy , in contrast to the near point explosion characteristic of the black hole winds .
the arrow in the bottom panel denotes the point of nuclear coalescence .
model results first published in cox et al .
( 2007).[figure : fb_energy],title="fig : " ] in principle , there are three mechanisms for driving winds in our major merger simulations : via shock heated gas ( cox et al .
2004 , 2007 ) , starburst - driven winds ( cox et al . , 2006c , 2007 ) and agn - feedback driven winds ( cox et al . , 2007 ; hopkins et al . , 2005a - d ; springel , di matteo & hernquist 2005a , b )
in practice , however , outflow rates in the simulations are dominated by winds driven by starbursts and agn feedback .
this was explicitly shown by cox et al . , ( 2007 ) who found that during the active phase , the outflow rate owing to shock heated gas was at most an order of magnitude less than that from starbursts and/or agn .
thus for the rest of this work , we will focus primarily on the effects of winds from star formation and agn - feedback , and rely on the model without these winds ( though with shock - heated winds ) as the fiducial comparative model .
we note that table 1 refers to this model as no - winds though in practice winds owing to shock heated gas do exist , albeit with minimal impact on the molecular ism .
we next review some of the broader physical effects generated by agn feedback - driven and supernovae - driven winds . in a recent comprehensive study of galactic winds in mergers utilizing ( among others ) identical models to those presented here , cox et al .
( 2007 ) analyzed the decoupled properties of starburst and agn feedback - driven winds in galaxy merger simulations .
we refer the reader to this work for a thorough analysis of these effects , and highlight here the wind properties most relevant to the discussion of this paper . as a reference
, we first show the instantaneous and integrated feedback energies for model sbbh in figure [ figure : fb_energy ] ( e.g. equations [ eq : bhenergy ] and [ eq : sbenergy ] ; model results first published in cox et al .
we highlight the period of peak starburst and agn activity which we will focus on for the remainder of this study , and denote as the `` active phase '' .
we focus on this period as the models exhibit their peak wind activity during this time .
the first passage of the galaxies triggers a starburst event , fueling star formation rates to @xmath050 , irrespective of the presence of a central black hole . at this point ,
starburst - driven winds begin to disrupt the central cold gas supplies , limiting to some degree ( dependent on the wind efficiency and speed ) the magnitude of the starburst . in the simulations presented here , during the in - spiral stage prior to nuclear coalescence , @xmath090% of the star formation ( and
thus @xmath090% of the starburst generated wind mass ) occurs ( cox et al .
2007 ) .
the growth of the central agn and associated feedback winds develops along a markedly different story line . while the initial passage and in - spiral stage of the galaxy merger fuels central black hole growth , the majority of black hole growth occurs during the active phase highlighted in figure [ figure : fb_energy ] , typically around nuclear coalescence .
the exact percentage of black hole growth that occurs during the active phase is dependent on galaxy mass ( cox et al .
2007 ) , but the generic trends are robust .
after the active phase , the total integrated feedback energy deposited into the ism from star formation and black holes is roughly the same .
however , the instantaneous properties are different .
while the energy input from star formation winds is spread out over the entire in - spiral and active phase , the black hole offers a relatively short impulse of energy .
the slow buildup of the stellar mass and relatively rapid black hole growth phase results in black holes contributing over five times the energy input to the ism as stars during the active phase ( cox et al .
this will have important consequences on the relative impact of winds on co emission from starbursts and agn . throughout this work
, we will focus on the rough peak of the active phase , specifically [email protected] - 1.8 gyr .
the panel is 12 kpc on a side , and the scale is on top.[figure : cdens ] ] ] to illustrate the generic properties of our methods , we have allowed a fiducial 40% gas fraction progenitor disk galaxy to evolve as outlined in springel et al .
( 2005a ) . for reference , in figure [ figure : cdens ] , we plot the column density map through the model disk galaxy . in figure
[ figure : spiral ] we present the co ( j=1 - 0 ) through ( j=10 - 9 ) intensity contours for comparison with observations where the intensity units are k@xmath91 km s@xmath35 where the temperature unit is a rayleigh - jeans temperature .
the co ( j=1 - 0 ) emission traces the star forming molecular gas through the spiral arms and becomes more intense near the starbursting nuclear region .
additionally visible are individual concentrations of cold gas and star forming regions emitting as discrete pockets of co ( j=1 - 0 ) emission throughout the spiral arms . higher density co transitions probe gas of a higher critical density .
thus the higher - lying transitions probe the higher density peaks , and become more centrally concentrated , following the density distribution in the galaxy . at first glance , the emission contours from e.g. the co ( j=1 - 0 ) image appear to peak at higher temperatures ( @xmath92 ) than are typically recorded from observations of local star forming spirals .
this effect owes to the spatial resolution of the radiative transfer models ( @xmath0250 pc ) . when simulating more practical observing conditions , the results from our modeling are quite comparable to observations of local universe spirals . as an example
, we compare our models to 50@xmath93 observations by scoville & young ( 1983 ) of co ( j=1 - 0 ) emission from m51 . when our model disk galaxy was scaled to be at the distance of m51 and the co ( j=1 - 0 ) emission convolved to the same resolution as the scoville & young observations , the integrated intensity in our model was comparable to the @xmath0 60 k - observed in m51 ( scoville & young , 1983 ) . in figure [
figure : cosed ] we present the normalized co line spectral energy distribution ( sed ) of the model face - on disk galaxy .
the co sed represents the total integrated flux density emitted from each line and serves as a diagnostic for the excitation conditions in the molecular gas .
these are plotted as the co integrated flux density versus the upper rotational quantum number of the transition .
co seds are particularly useful in studies of unresolved high-@xmath56 sources ( e.g. narayanan et al . 2007a ; wei et al . 2005a , 2007 ) .
we have also decomposed the co sed by plotting the contribution to the flux from the starbursting nucleus ( central kpc ) and from the rest of the galaxy .
the co flux density rises up to the co ( j=5 - 4 ) transition , and then drops toward higher excitation lines .
the intensity from the highly excited component of the co gas ( e.g. j>6 ) comes primarily from the warm and dense conditions in the nuclear starburst regions , though some contribution from subthermally excited gas in the outer regions of the galaxy is visible ( narayanan et al .
this is in good agreement with observations of local starbursts .
for example , in a co sed decomposition of m82 , wei et al .
( 2005b ) found that the central regions peak at the co ( j=6 - 5 ) transition whereas the total sed peaks near the co ( j=4 - 3 ) transition .
similarly , the co line sed from the nuclear region of ngc 253 peaks near j=7 or j=8 ( bayet et al . 2004 ; bradford et al .
2003 , 2005 )
our approach in combining non - lte radiative transfer calculations with sph simulations allows us to uniquely explore , for the first time , a variety of the decoupled effects of starburst and agn feedback - driven winds on the molecular gas emission from galaxy mergers .
in particular , a powerful tool for understanding the spatial distribution of molecular gas in local mergers is through high resolution velocity - integrated intensity contour maps . these can serve to describe the morphology and kinematics of the star - forming molecular gas , as well as characterize the nature of star - forming regions through observations of multiple excitation levels ( e.g. iono et al . 2004 ) .
indeed , such maps have revealed , for example , the compact nature of molecular emission in mergers ( bryant & scoville , 1999 ) , the double - nucleus in the central regions of arp 220 ( sakamoto et al .
1999 ) , and the large scale distribution of atomic and molecular gas in colliding galaxy systems ( iono , yun & ho , 2005 ) .
in this section , we build on the current understanding of co morphologies in interacting systems by describing the response of the co morphology to starburst and agn feedback - driven winds .
we first discuss general properties of the co morphology in mergers .
we then continue with a discussion of signatures of galactic winds apparent in co emission maps via outflows detectable in the co morphology .
we conclude the section with a brief discussion of secondary effects of galactic winds in affecting the spatial extent of co gas . in figure
[ figure : sb10bh_morph ] , we show the evolution of the co ( j=1 - 0 ) emission during a major merger of two disk galaxies simulated in run sbbh ( model with starburst and black hole winds ) .
the properties discussed in this subsection ( [ section : generalmorph ] ) are generic to each of our models , regardless of wind parameter choice .
the images in figure [ figure : sb10bh_morph ] start after the first passage of the galaxies , when they are approaching for their final coalescence .
the tidal interaction of the first passing of the galaxies causes ` streamer ' gas to emit in the intergalactic space between the two galaxies .
the co morphologies are disturbed by the first passage itself , and the co emission is extended in nature until the nuclei have fully merged .
the first passage of the galaxies induces a burst of star formation , peaking in the nuclear regions of each galaxy .
this is reflected in the luminous co emission at the center of each galaxy .
because the co ( j=1 - 0 ) can be thermalized at relatively low critical densities , and lies only @xmath05 k above ground , it serves as a good tracer for total molecular gas content .
both diffuse envelopes of clouds as well as dense cores can emit co ( j=1 - 0 ) flux .
conversely , emission from higher lying lines of co , and high dipole moment molecules ( e.g. hco+ , hcn , cs ) arises primarily from dense cloud cores .
thus , higher lying co lines ( e.g. co j=3 - 2 ) from the merger broadly follow the distribution of the co ( j=1 - 0 ) line , but remain more compact in nature , generally emitting only from the dense starbursting nucleus .
this was explicitly seen in the disk galaxy in figure [ figure : spiral ] as well .
the nuclear co intensity peaks soon after the progenitor galaxies lose individual identity , and there is only one surface brightness peak . during this time , stars formed in the merger - induced starburst heat up the molecular gas such that collisions readily excite the co in the warm ( t@xmath030 k ) and dense ( @xmath9410@xmath95 ) gas .
the columns are still quite large through the nucleus as the starburst has not had time to consume the available molecular gas . while the nuclear co intensity peaks near coalescence , the globally averaged velocity - integrated co intensity across all transitions drops through the evolution of the merger .
this can be attributed to global consumption of cold molecular gas by star formation over time .
this result is consistent with observations by gao & solomon ( 1999 ) who found from co ( j=1 - 0 ) emission that the total molecular gas content of a merger decreases as a function of projected nuclear separation in galaxy mergers .
starburst and agn feedback - driven winds can leave an imprint on the molecular line emission via large scale molecular outflows seen in the co morphology .
these winds are seen to entrain large masses of molecular gas in outflows which appear prominently as a secondary surface brightness peak in velocity - integrated emission contour maps ( narayanan et al . 2006a ) . in this section ,
we discuss the properties of the co morphology in the context of these outflows .
the outflows discussed here are only those which are large enough to be visible via co ( j=1 - 0 ) emission maps . for the purposes of understanding
general trends of detectable outflows in this section , we will find it convenient to define a detectable outflow as any co emission peak at least 50% of the nuclear emission maximum , with a projected distance at least 1.5 kpc away from the galactic nucleus .
the molecular outflows are defined only after nuclear coalescence , and thus also follow the requirement that the co emission peak be separate from the stellar surface density peak . before continuing
, we note that the simulations chosen here are not extreme in any way ( figure [ figure : vc3vc3e_2b_bharsfr ] ) .
the peak b - band luminosity during the quasar phase is @xmath010@xmath96 .
similarly , the peak ir luminosity is @xmath010@xmath7 , allowing the galaxy to be just visible as a ulirg .
more massive mergers ( e.g. cox et al .
2007 ) , and , as will be discussed in
[ section : morphorientation ] , more coplanar mergers will drive stronger winds , leaving the simulations in this section representative of lower limits . here
, we discuss the general properties of both starburst and agn feedback - driven outflows which appear in co emission maps . while there are differences in some specific properties of the outflows based on the type or strength of the driving wind , we defer those comparisons to later subsections and focus this subsection on the outflow properties shared by all the simulations presented here
. in figure [ figure : vc3vc3e_outflow ] , we show an example of molecular gas entrained in the outflow from model bh ( agn winds only ; outflow first described in narayanan et al .
the cold gas in the outflow is not completely ablated by starburst and agn heating . while the co emission from cold gas in the outflows closest to the nucleus is faint owing to lower columns of cold molecular gas , clouds which remain more deeply embedded in the outflow continue to exhibit active star formation . during short ( @xmath05 myr ) time intervals
, the co ( j=1 - 0 ) velocity - integrated intensity from galactic outflows can be seen to be comparable ( within a factor of @xmath02 ) to the galaxy s nuclear emission .
the detectable outflows typically correspond to entrainments of mass @xmath010@xmath97 - 10@xmath98 , with peak column of @xmath010@xmath99 - 10@xmath100 ( within the constraints of our definition of a detectable outflow , [ section : outflowmaps ] ) .
the more massive outflows in this mass range tend to be seen closer to the point of nuclear coalescence , with a trend toward decreasing outflow mass as the merger progresses .
this owes to more loosely bound gas near the time of coalescence , as well as the feedback energy rate being near its peak ( figure [ figure : fb_energy ] ) .
these outflow masses are typically 5 - 15% of the total molecular gas mass . in the mergers
e , sb , and sbbh , with random disk orientation angles , the outflows do not occur in any preferred direction . as we will discuss later ,
this is not the case for the coplanar model .
the entrained clouds can be relatively excited , allowing outflows to be imageable at higher - lying co transitions ( e.g. j@xmath904 ) . in figure
[ figure : vc3vc3e_outflow_excitation ] , we show the outflow from the first panel of figure [ figure : vc3vc3e_outflow ] in co transitions j=1 - 0 , j=3 - 2 and j=4 - 3 .
the latter two transitions have a relatively high critical density of @xmath9010@xmath45 meaning they are typically thermalized only in the dense cores of gmcs , and may rely on radiative pumping for excitation in other parts of the ism .
the warm and dense gas heated by active star formation in the deeply embedded clouds in the outflows collisionally excite the co gas into higher levels .
in fact , the co flux density in the outflow peaks at j=4 , consistent with actively star - forming gas and multi - line measurements of excitation patterns in starburst - driven outflow of m82 ( walter , wei & scoville , 2002 ) .
it should be noted that the relative visible lifetimes of outflows in higher - lying transitions may be much smaller than at co ( j=1 - 0 ) .
for example , the outflow presented in figure [ figure : vc3vc3e_outflow ] appears as a unique secondary surface brightness peak for @xmath035 myr at co ( j=1 - 0 ) . in the co ( j=3 - 2 ) transition ,
the same outflow is distinguishable from the background for only @xmath020 myr . as an illustrative point demonstrating the relative contributions of starburst and agn winds in driving detectable molecular outflows , in figure [ figure : outflows_differentmodels ] we plot three snapshots during the period of nuclear coalescence for models no - winds , bh , sb , and sbbh ( thus varying only the form of winds included , and not geometry of the merging galaxies ) .
the wind - dependent morphological differences in the models are typically minimal near the point of nuclear coalescence , and thus the outflows imaged at this time serve as a reasonable comparison for the capacity of different wind models to drive outflows . here , loosely bound gas near the inner regions is seen to be entrained in nuclear galactic winds . for reference , these outflows are the same as those shown in figures [ figure : vc3vc3e_outflow ] and [ figure : vc3vc3e_outflow_excitation ] .
model bh is able to drive strong outflows ( 4th row , figure [ figure : outflows_differentmodels ] ) with properties described generally in
[ section : outflowmapgeneral ] .
the black hole feedback is tightly associated with the accretion rate , and thus the agn wind is produced over a very short period of time and peaks soon after nuclear coalescence .
relative to the evolution of the galaxy , the agn wind generically acts like a point explosion , and the bulk of the energy is imparted to the ism nearly instantaneously .
indeed , hopkins et al .
( 2006c ) and hopkins & hernquist ( 2006 ) have shown explicitly that these agn - driven winds are well - characterized by generalized sedov - taylor solutions for point explosions .
the outflows in model bh are primarily seen soon after the major merger , when the gas is still highly dynamical and loosely bound .
they are also seen near the height of quasar activity , when the agn feedback - driven winds are strongest .
the starburst driven winds are also able to power outflows ( 2nd row , figure [ figure : outflows_differentmodels ] ) , albeit less strongly than in the agn - wind models ; consequently , these outflows are detectable via emission mapping for shorter periods of time .
the typical lifetime of an outflow such that it is detectable in co ( j=1 - 0 ) in the sb simulation ( with only starburst winds ) is @xmath020 - 25 myr ( compared with @xmath035 myr in the agn only model ) . on average ,
3 - 5@xmath3310@xmath97are entrained , although outflows as small as @xmath010@xmath101are seen to occur , consistent with observations of starburst driven outflows by sakamoto et al .
( 2006 ) and walter et al .
it should be noted that these values are derived within the constraints of the parameters chosen for our starburst winds ( @xmath26=0.5 , @xmath30=837 ; [ section : numericalmethods ] ) . in principle , a more efficient wind might increase the mass outflow rate , but , as discussed in cox et al .
( 2007 ) , such parameter choices were found to quench merger - induced starbursts , and are thus probably not representative of e.g. ulirgs .
the relative weakness of starburst - driven outflows compared to the agn only model is a direct consequence of the lower peak energy input from the starburst - driven winds , and hence lower power . during the active phase , the feedback from black holes inputs more than five times the energy as that imparted by star formation .
this can be attributed to two causes .
first , black hole growth is far more efficient at converting accreted mass into thermal energy than star formation .
for example , by utilizing equations ( [ eq : bhenergy ] ) and ( [ eq : sbenergy ] ) , cox et al .
( 2007 ) noted that the relative efficiency of black holes to star formation in converting gas mass to energy deposited into the ism was of order @xmath0640 ( see , also , lidz et al .
second , during the active phase , the relative black hole growth rate is much larger than the concomitant sfr . during this time , while the bh grows by a factor of @xmath04 , only @xmath08% of the final stellar mass is formed ( cox et al .
in contrast , @xmath090% of the stellar mass ( and consequently , stellar winds ) is produced during the in - spiral stage , prior to the active phase .
the effect of including both starburst and agn winds is demonstrated in the third row of figure [ figure : outflows_differentmodels ] .
the outflows produced are more massive and longer lived than the starburst - only wind models , but less so than the agn - only wind models .
the greater capacity to drive outflows over the starburst winds only model ( sb ) can be attributed to the extra energy input given by the accreting black hole(s ) .
however , the outflows in figure [ figure : outflows_differentmodels ] are seen to be less powerful than those in the black hole only model .
this effect owes to the starburst winds that occur prior to nuclear coalescence . as shown explicitly by cox et al .
( 2007 ) , starburst winds lower the amount of rotationally supported gas early during the in - spiral stage , which limits the amount of gas available for immediate funneling into the central nucleus via tidal torquing ( hernquist , 1989 ; mihos & hernquist , 1994b , 1996 ) .
this gas does eventually rain into the nuclear potential , fueling black hole growth and producing a black hole with the same mass as in the agn wind only model ( bh ) and a galaxy that lies on the m@xmath102-@xmath103 line ( cox et al .
2007 ; di matteo , springel & hernquist , 2005 ; robertson et al .
2006a ) , as well as the black hole fundamental plane ( hopkins et al .
2007a , b ) . however , the black hole growth is integrated over significantly longer time , and thus the impact of the associated winds on molecular gas emission is diminished from the agn - only model .
the relative effects of agn feedback and starburst - driven winds on powering molecular outflows can be summarized in figure [ figure : imageable_clumps_mass ] , where we plot the mass of detectable outflows as a function of time for models bh , sb and sbbh .
the masses of gas each model is able to drive in an outflow is comparable to within a factor of @xmath02 - 3 . agn feedback - driven winds drive outflows clustered primarily around nuclear coalescence and the peak period of black hole accretion . in comparison , the models with starburst driven winds ( sb and sbbh ) power detectable outflows for relatively shorter periods . ]
the strength of outflows in merger models depends on the initial orientation of the merging galaxies . in models where
the disks are aligned , strong tidal torquing can cause a rapid inflow of gas thus providing a sudden burst of agn feedback energy injected into the ism .
alternatively , geometries such as those in the bh and no - winds models by comparison experience less tidal torquing on the gas , causing the major black hole accretion phase to be longer lived . in order to investigate the dependence of co outflows on merger orbit
, we have run simulation co - bh ( table [ table : ics ] ) in which the progenitor galaxies are coplanar with the orbital plane ( @xmath104=@xmath105=@xmath106=@xmath107=0 ) .
the stronger tidal torquing in the coplanar configuration results in a more efficient blow - out of material .
the gas in the galaxy remains confined in a thin configuration in the plane of the merger , consistent with previous simulations which showed the possibility of disk formation from major mergers ( springel , & hernquist , 2005 ; robertson et al .
2006b ) . in figure
[ figure : vc3vc3h_2c_outflow ] , we show a typical outflow from the coplanar model .
similar to the other models in table [ table : ics ] , the primary molecular outflows occur at or near the point of nuclear coalescence . unlike models sb / sbbh / bh , which showed no preferred direction for outflows , the outflows in coplanar model co - bh are seen to occur almost exclusively in the plane of the merger with an expanding shell - like structure .
the fraction of time in our simulations outflows are visible is dramatically increased as well in this merger model . in the sb / sbbh / bh models ,
the outflows typically involve loosely bound gas entrained in agn winds which fall back into the nuclear potential , tending the emission toward a centrally concentrated morphology .
conversely , the outflows in the coplanar model appear to continue expanding within a 6 kpc ( radius ) region , resulting in an increasing half light radius throughout the evolution of the post - merger galaxy . the continued presence of detectable outflows in the coplanar model is seen explicitly in figure [ figure : imageable_clumps_mass2 ] , where we show the mass of detectable outflows for models bh and co - bh ( thus varying only the orientation angle ) as a function of time .
the outflows are most dramatic for face on viewing angles as the outflows are preferentially within the plane of the merger .
while at some inclined angles outflows continue to be visible , broadly , the more edge - on the viewing angle becomes , the more difficult outflows become to detect against nuclear emission .
that said , outflows may be preferentially detectable along these sightlines via spectral line signatures which we will discuss further in
[ section : lineprofiles ] . within the plane of the merger
, outflows can be viewed in outflow , or falling back into the nucleus throughout the entire @xmath0200 myr peak of the active phase investigated here .
the peak black hole accretion rate and integrated energy deposited into the ism are similar for both the coplanar model and model bh .
consequently , the typical physical conditions of the agn feedback - driven outflows are roughly the same in the two simulations . in the coplanar model discussed in this section ,
the typical mass of an outflow is a few @xmath3310@xmath1 , similar to the outflows observed in model bh which had a more generic inclination angle for the progenitor galaxies . ]
1.5 - 2 kpc , and becoming more compact at higher lying transitions .
the arrow in the third panel denotes the point of nuclear coalescence .
( see text for details.)[figure : halflight_2models ] ] ] galactic winds can play a role in the spatial extent of co emission in galaxy mergers . after nuclear coalescence ,
the co ( j=1 - 0 ) emission in the simulation without winds ( model no - winds ) is seen to be more spatially extended than the emission from mergers with winds by a factor of @xmath02 .
the co ( j=1 - 0 ) emission from model no - winds has a typical halflight radius of 1 - 2 kpc ( figure [ figure : halflight_2models ] , top panel ) .
the co emission remains on these extended scales throughout the evolution of the merger , and star formation simply consumes the available molecular gas as fuel .
the predicted lack of extended co ( j=1 - 0 ) emission in galaxy mergers with winds is consistent with observations . in a high resolution
co survey of five mergers , bryant & scoville ( 1999 ) found the primary co emission to be compact and concentrated within a 300 - 400 pc radius of the nucleus in three sources , and within 1.6 kpc in all of the objects in the sample .
moreover , high resolution observations and radiative transfer modeling of local mergers have confirmed the compact nature of co emission in mergers ( downes & solomon , 1998 ) .
winds associated with growing black holes and/or starbursts may help to explain this . in the models with winds ,
as the galaxies merge , the sudden inflow of gas into the nucleus rapidly drives up the nuclear star formation rate , as well as the accretion rate of the black hole(s ) .
the feedback associated with star formation and the central agn is effective at heating the gas in the nuclear regions , and dispersing lower density co ( j=1 - 0 ) gas , leaving behind only a dense core of cool gas in the central regions in the winds models .
the average co ( j=1 - 0 ) half light radius after the nuclei have coalesced for the models with sne and/or agn is @xmath0700 pc .
this was also seen in figure [ figure : sb10bh_morph ] where we showed that the major co ( j=1 - 0 ) intensity for the model with sne and agn winds originates in a compact region , in agreement with observations of local ulirgs ( bryant & scoville , 1999 ; downes & solomon , 1998 ) . in principle
, these model results can be directly tested through observations of higher lying co lines from local galaxies .
the middle and bottom panels of figure [ figure : halflight_2models ] show the halflight radius as a function of time for models bh and no - winds for the co ( j=3 - 2 ) and co ( j=7 - 6 ) transitions .
these transitions probe denser gas than co ( j=1 - 0 ) , and have a critical density of @xmath9010@xmath108 - 10@xmath89 .
figure [ figure : halflight_2models ] shows that the emission from the model with no winds is stratified such that tracers of dense gas [ e.g. co ( j=7 - 6 ) ] are more compact by nearly an order of magnitude than tracers of lower density gas [ e.g. co ( j=1 - 0 ) ] . in figure
[ figure : halflight ] we show the mean half light radius for each model plotted as a function of co transition . here
, we have averaged the half light radius after nuclear coalescence in each simulation .
the co emission in the model without winds becomes more compact in higher lying transitions ( e.g. those with higher critical densities ) . in the models with sne and/or agn winds
, the co emission shows a similar behavior , but the halflight radius for the lowest transitions is about a factor of two smaller than in the models without winds . with these predictions ,
we caution that specific values quoted here are dependent on a number of assumptions included in parameter choices ( such as , for example , the specific multi - phase ism breakdown [ e.g. springel & hernquist , 2003 ] or constant co abundances ) , and thus the normalization of the plots presented are to be taken within those confines . however , the general trends are likely to be more robust . in this sense ,
the relatively shallow slope in half - light radius versus transition ( figure [ figure : halflight ] ) remains a valid observational test for the role of galactic winds in shaping observed co spatial extents .
finally , it is important to note that the weak trend in average halflight radius as a function of transition in figure [ figure : halflight ] is an ensemble - averaged trend over three orthogonal sightlines and numerous points in the post - coalescence evolution of the galaxies .
thus , observations examining this relationship must select samples of galaxies after the progenitors have coalesced - in other words , after the merger has only a single nuclear emission peak in the co morphology .
generically , when the nuclei of the progenitor galaxies are still resolvable , prior to coalescence , the unresolved co emission line exhibits multiple emission peaks corresponding to the nuclear regions of each galaxy ( e.g. greve & sommer - larsen , 2006 ) .
when the molecular gas in galaxy mergers becomes dynamically relaxed , the co emission line is reasonably well represented by a single broad gaussian . alongside this general trend , at various times throughout the evolution of the merger , secondary ( and potentially , though less
commonly , tertiary and quaternary ) peaks can be seen superposed on the broad co gaussian - like line . in figure
[ figure : map_spectra ] , we show a representative evolution of co spectral lines , and how they correspond to the co morphology .
these secondary peaks superposed on the broad gaussian originate in massive co gas clumps moving along the line of sight . in figure
[ figure : map_spectra ] , the secondary peak in the third panel represents emission from the outflowing clump visible in the corresponding co contour image . while these superposed secondary emission peaks are a general feature of all models in our simulations , the physical origin of these peaks is dependent on the strength of the winds included in a given merger model .
in particular , secondary emission peaks with velocities greater than the circular velocity typically represent loosely bound gas which has not virialized ( peaks at velocities lower than the circular velocity may also be loosely bound gas , though they may also represent gas in rotational support ; for brevity we hereafter refer to secondary peaks at velocities greater than the circular velocity as high velocity peaks ) .
as we will show more quantitatively , in models which include winds , this high velocity gas is typically entrained in outflow during periods of activity ( e.g. elevated star formation rate or black hole growth ) , and thus give rise to high velocity peaks in the emission line representative of outflowing gas .
this is analogous to the situation seen in the co morphologies presented in figure [ figure : outflows_differentmodels ] when loosely bound gas simply fell into the central potential in the model without winds , but was entrained in outflow in models with winds ( here , models sb , sbbh and bh ) .
this effect of winds on co emission profiles was first shown explicitly in similar calculations by narayanan et al .
( 2006a ) .
this work demonstrated that once the galaxies merged , large columns of gas outflowing along the line of sight can leave their imprint on the broad single - gaussian emission line characteristic of galactic co emission by introducing an additional superposed emission peak , red or blueshifted at the line of sight velocity of the outflow .
the relative significance of high velocity peaks compared to the main emission line decreases as the gas virializes in the merger . to illustrate this , in figure [ figure : chi ] we plot the evolution of the goodness of fit ( taken here as the median of the square of the residuals ) for a single gaussian fit to the emission lines from model bh , although the results are common across all models .
the impact of winds in driving high velocity peaks decreases as the merger progresses . in our models , high velocity peaks in co emission lines
are typically representative of clumps of gas moving along the line of sight with masses @xmath905@xmath3310@xmath101 .
gas clumps with masses much less than this are typically not able to emit strongly enough to be seen in the emission spectrum over the main gaussian . in the remainder of
[ section : lineprofiles ] , when we discuss clumps of outflowing and infalling gas , we will be referring to those above this minimum mass that are able to impact the emission line profile . because high velocity peaks appear in our simulations even in spatially unresolved observations , they may be a powerful diagnostic for agn and starburst induced outflows both in the local universe and at earlier epochs .
however , at this point in the discussion , there are open questions regarding the uniqueness of this line profile , as well as how these secondary peaks specifically relate to starburst and agn activity . in the remainder of this section , we attempt to quantify the use of molecular emission line profiles as a diagnostic for outflowing gas in galaxy mergers .
emission lines representing gas with a strong line of sight velocity component can originate from either outflowing or infalling gas . as demonstrated by narayanan et al .
( 2006a ) , agn feedback - driven molecular outflows can generate high velocity secondary peaks in the co emission lines when the outflow has a significant line of sight velocity component .
any gas falling into the central potential with a negative radial velocity with respect to the center of mass of the system ( e.g gas that is in infall , or pre - coalescence mergers ) may also drive high velocity peaks in the co emission lines , and are thus degenerate with high velocity peaks that originate in outflows .
it is therefore necessary to establish the uniqueness of high velocity peaks in co emission lines with respect to an actual outflow or infall origin .
any attempt to break this degeneracy requires a more detailed knowledge of the physical properties and observational characteristics of the galaxy during periods when high velocity peaks are observable . in order to assess how the evolutionary state of the galaxy merger may determine the origin of high velocity peaks ( i.e. whether they owe to outflowing or infalling gas )
, we investigate the physical conditions during times when the galaxy exhibits high velocity peaks . to this end , we created synthetic co ( j=1 - 0 ) spectra along three orthogonal sightlines for the models in table [ table : ics ] .
these spectra were made by integrating over the entire co emitting area . to identify candidate emission lines with high velocity peaks driven by winds ( as opposed to rotation ) , we require the emission line to have a peak near the systemic velocity of the galaxy , and an additional peak with velocity greater than @xmath109160 .
this fiducial requirement is set in place to filter out multiple emission peaks that arise from rotation as @xmath0160 is the circular velocity of the post - merger galaxy at the spatial extent of the gas , and thus the maximum velocity of rotating gas .
we consider all high velocity peaks whose peak intensity is at least 20% the peak line intensity . upon identifying a line with a high velocity peak
, we then determine whether the clump of gas generating the high velocity peak is outflowing or infalling . in order to simplify the analysis , in the few cases with multiple high velocity peaks , we focused on the peak with highest velocity . in figure
[ figure : vc3vc3e_2b_outflowinfall ] , we plot a histogram of the number of high velocity peaks seen as a function of time in the simulation with black hole winds only ( model bh ) , summed over 3 orthogonal lines of sight . we have further broken the line profiles down into having an outflow or infall origin . as a reference for the evolutionary state of the merger , we have overplotted the sfr , black hole accretion rate , and bolometric luminosity .
figure [ figure : vc3vc3e_2b_outflowinfall ] demonstrates that while the number of high velocity peaks is roughly constant after the major merger , the origin of these line profiles is a function of the evolutionary state of the galaxy . during phases of heavy black hole growth , the majority of the high velocity peaks owe to outflowing gas . towards the end of the quasar phase , when much of the blowout has already occurred , gas raining from outer parts of the galaxy give rise to multi - peaked profiles with an infall origin .
conversely , similar analysis from the model without winds shows a more random origin for high velocity peaks in the emission profiles , reflecting the more randomly oriented velocity vectors of loosely bound clumps .
this is explicitly shown in figure [ figure : vc3vc3e_noc_outflowinfall ] , where we show the origin of high velocity peaks in a merger without winds compared with the bolometric and ir luminosity .
another way of saying this is that the winds in model bh do not _ cause _ the high velocity peaks to exist in the co spectra .
rather winds typically entrain the clumps of gas behind the high velocity peaks in outflow , thus giving physical significance to secondary peaks .
this is not the case in the model with no winds , where the molecular gas representative of the high velocity peaks has random radial velocities with no preferred direction ( figure [ figure : vc3vc3e_noc_outflowinfall ] ) . _
high velocity peaks are seen in models both with and without winds ; in wind models , though , these peaks preferentially owe to outflowing gas during peaks of starburst activity and black hole growth . _
observationally , many of the criteria that are used in creating samples of local galaxy mergers may select galaxies which are at the peak of their starburst or agn activity . high velocity peaks in the emission profiles ( e.g. higher than the circular velocity of the host ) observed in these galaxies may be likely , then , to owe to outflowing clumps of gas ( figure [ figure : vc3vc3e_2b_outflowinfall ] ) .
for example , the extreme bolometric and ir luminosities trace not only the prodigious star formation owing to the merger induced starburst , but also reflect a large contribution from the growing supermassive black hole .
the winds associated with this black hole growth are behind much of the outflowing molecular gas during the most luminous phase of the galaxy s evolution .
similarly , ir colors which select objects with embedded agn , such as the 25@xmath8m/60@xmath8 m flux ratio , may preferentially show objects which owe their outflow profile to outflows as opposed to infalling gas clumps . finally , the hard x - ray flux is dominated by the central agn , and thus traces the general growth of the central black hole .
our models suggest that the phase during which the x - ray luminosity is at its greatest will be marked by high velocity peaks in the co emission originating in outflowing molecular gas . a natural question is whether or not one can observationally infer the nature of high velocity peaks in co emission lines as discussed in
[ section : lineoverview ] .
specifically , high velocity peaks can originate in both wind - driven outflows , as well as owing to dynamical effects in merging galaxies ( e.g. figures [ figure : vc3vc3e_2b_outflowinfall ] and [ figure : vc3vc3e_noc_outflowinfall ] ) .
distinguishing characteristics between the various origins for high velocity peaks may be evident when considering limiting velocities for the high velocity peak much greater than the virial velocity .
figures [ figure : vc3vc3e_2b_outflowinfall ] and [ figure : vc3vc3e_noc_outflowinfall ] showed the presence of all high velocity peaks beyond a limiting velocity being set at the virial velocity of the galaxy ( here , @xmath110 ) . when considering the origin of high velocity peaks driven by dynamical effects , few emitting clumps of gas are able to regularly sustain velocities much greater than the escape velocity of the galaxy ( or @xmath111 ) . in contrast , when included , winds are able to drive outflows at higher speeds .
we show this effect quantitatively in figure [ figure : vspike_noutflows ] where we plot the number of high velocity peaks as a function of limiting velocity for both two models with agn winds ( models bh and co - bh ) and the model with no winds . at velocities greater than @xmath01.7 times the virial velocity ,
the number of high velocity peaks driven by purely dynamical effects in model no - winds drops rapidly while winds from the agn in model bh and co - bh are still able to drive clumps in outflow at these velocities .
the range of the numbers of high velocity peaks in the models with agn winds owes to a merger - orientation dependent range in instantaneous black hole feedback energies .
more efficient fueling of the central agn in the coplanar model allows for a stronger impact of the winds on the molecular line profiles ( cf .
[ section : morphorientation ] ) . in either case , both models which consider agn feedback show that agn feedback - driven winds can drive high velocity peaks at larger velocities than dynamical effects can typically account for . in this sense ,
one may be able to discriminate between high velocity peaks with an outflow origin versus random kinematics when restricting the limiting velocity of the peaks to @xmath901.7 times the virial velocity of the galaxy .
it is important to note , however , that this is specific to the single merger with no winds examined in this work , and may not necessarily be general .
the dispersion seen between the no winds model and the models with agn winds ( bh and co - bh ) in figure [ figure : vspike_noutflows ] is in principle a lower limit .
the galaxies in figure [ figure : vspike_noutflows ] are relatively low mass - @xmath112 .
cox et al . ( 2007 ) found that the energy input by accreting black holes in galaxy mergers increases as a function of galaxy mass for a variety of progenitor gas fractions ( ranging from 0.05 to 0.8 ) and initial merger angles .
specifically , these authors found @xmath113 , suggesting that higher mass galaxies have the capacity to drive co peaks at even higher speeds compared to their virial velocity . as an example
, a merger with similar gas fraction and orientation angle as model bh , though with final stellar mass @xmath114 would have roughly an order of magnitude greater black hole energy input ( cox et al .
2007 ) .
figure [ figure : vspike_noutflows ] suggests that the effect of agn feedback - driven winds on co emission line profiles should be discernible by comparing a merger s virial velocity to observed high velocity features in the co line profile .
a comparably clear ( and observationally more tractable ) test may be an analysis of the velocity offset of high velocity peaks with respect to the centroid velocity of the emission line .
when examining high velocity peaks defined with respect to the centroid velocity , we find the curves presented in figure [ figure : vspike_noutflows ] are robust .
this owes to a narrow distribution of centroid velocities in all the models , with a typical standard deviation in the distribution @xmath115 .
many of the model results presented here have the potential to be directly tested in the near future , either with existing datasets , or with observations from millimeter - wave interferometers such as carma and pdbi , and high resolution submillimeter telescopes such as the submillimeter array ( sma ) or atacama large millimeter array ( alma ) .
first , it is worth briefly revisiting the evolutionary status of the galaxies studied here , and placing them into a larger context . in a merger - driven scenario for quasar formation , gaseous inflows fueled by tidal torquing on the gas ( barnes & hernquist 1991 , 1996 ; hernquist , 1989 ) fuel nuclear starbursts ( mihos & hernquist , 1996 ) , allowing the galaxy to be viewed as a dusty starburst system similar to local ulirgs ( chakrabarti et al .
the gaseous inflows fuel central black hole growth ( hopkins et al .
2005a - d ; 2006a - d ; 2007h ; di matteo et al . 2005 ; springel et al . 2005a ) which makes the galaxy first visible as a ulirg with warm infrared colors ( chakrabarti et al .
2007a ) , and then as an optically selected quasar ( hopkins et al . 2005a - d ; 2006a - d ; 2007h ) .
a key physical process in driving the host galaxy along this sequence is galactic winds .
namely , winds driven by agn feedback ( as well as nuclear starbursts ) contribute to clearing the veil of obscuring gas and dust surrounding the central black hole , and allow the quasar to become visible across numerous sightlines ( hopkins et al
. 2005a ) .
these winds are most prominent during the @xmath0200 myr associated with major black hole growth .
the models presented in this work therefore examine simulated galaxy mergers during this time period of heightened wind activity , starting near nuclear coalescence , and ending slightly after the quasar light curve has declined . during this time
, the galaxy is seen to have increased ir , x - ray and @xmath10-band luminosities , indicative of the starburst event and central black hole growth ( figure [ figure : vc3vc3e_2b_outflowinfall ] ) . in this sense , these simulations are designed to provide analysis for local galaxies selected for nuclear activity via elevated x - ray and ir fluxes ( e.g. ulirgs ) .
second , we have shown that the effects of winds can shape the molecular morphology of galaxy mergers .
for example , distinct differences are seen in the simulations with winds versus without when considering the spatial extent of the co emitting gas ( [ section : halflight ] ) .
the predicted weak evolution of average half - light radius as a function of transition for evolved mergers can be tested in the co ( j=1 - 0 ) , ( j=2 - 1 ) , ( j=3 - 2 ) and ( j=7 - 6 ) transitions with current technology using a millimeter - wave interferometer ( e.g. carma , pdbi ) and the sma .
perhaps the most dramatic effects of galactic winds on the co morphology lie in signatures of entrained molecular outflows in emission contour maps .
co outflows driven by starburst - dominated winds have been imaged in a handful of nearby galaxies .
recent mapping of starburst galaxies m82 and ngc 3256 have shown that molecular gas can indeed survive in outflows entrained in starburst - driven winds ( walter et al , 2002 ; sakamoto et al .
both galaxies are known to be interacting systems , the latter a major merger .
the outflow masses are recorded to be @xmath010@xmath39 - 10@xmath1 , consistent with the model results presented in [ section : morphology ] .
the outflows have been detected in co ( j=2 - 1 ) emission ( peaking at co j@xmath03 or 4 ; wei et al .
the outflow observed in m82 appears to be distributed in a roughly spherical halo with radius @xmath01.5 kpc .
this outflow may correspond with an h@xmath51 outflow , as well as m82 s dust halo ( alton , davies & bianchi , 1999 ; walter et al .
this type of morphology is seen along some sightlines in the models bh and co - bh , alongside the more collimated outflows characteristic of e.g. figure [ figure : outflows_differentmodels ] .
outflows which may be driven by a central agn have been imaged in local galaxies as well .
recent sma observations of ngc 6240 by iono et al .
( 2007 ) have revealed an outflow in co ( j=3 - 2 ) morphologically similar to the outflow seen in figure [ figure : vc3vc3e_outflow ] .
ngc 6240 is a major merger near final coalescence with two central x - ray sources , which _
observations have suggested are a binary agn ( komossa et al , 2003 ) . in this sense ,
the outflow imaged in ngc 6240 occurs when our models suggest the most massive outflows may be imageable ( [ section : relativestaragnmorph ] ) .
the mass of the outflow in this system is estimated to be @xmath05@xmath3310@xmath1 , consistent with the typical masses entrained in agn winds in our simulations .
similar morphological features to those associated with molecular outflows have been detected in local lirg / ulirg ngc 985 , though the interpretation of these features is still uncertain ( appleton et al .
2002 ) . while the number of outflows directly imaged in co remains small and limited to only relatively recent observations ,
outflows traced in absorption ( e.g. na i d lines ) have a much more extensive detection history and associated literature .
absorption line outflows have been detected in the uv and optical in both nearby starburst galaxies / agn ( heckman et al .
2000 ; martin , 2005 ; rupke & veilleux , 2005 ; rupke , veilleux & sanders , 2005a - c ) , as well as high - redshift galaxies ( pettini et al . 2002
; shapley et al . 2003 ) and quasars ( ganguly et al .
2006 ; hamann , barlow & junkkarinen , 1997 ; misawa et al .
2005 , 2007a , b ; narayanan et al .
2004 ; trump et al . 2006 ; wise et al .
hi has also been seen in outflow in active galaxies via absorption lines ( morganti , tadhunter & oosterloo ; 2005 ) .
in addition to the neutral absorption line outflows detected , many outflows are relatively high ionization lines ( e.g. civ , nv ; hamann et al . 1997 ; narayanan et al .
2004 ) . in this sense
, it appears as though these may be different populations of outflows from the cold , neutral emission line outflows discussed in this work .
interestingly , in the few systems with co outflow candidates ( as evidenced by their morphology ) , there is an apparent coincidence between the velocity offsets in the molecular outflows and absorption line outflows .
for example , the starburst - driven outflow observed in local merger ngc 3256 by sakamoto et al .
( 2006 ) has a velocity offset ( from the systemic , which in the case of co , typically corresponds to the nuclear emission ) of @xmath0300 km / s , while heckman et al .
( 2000 ) showed a blueshifted na i d absorption line of velocity @xmath0300 km / s .
similarly , appleton et al .
( 2002 ) found two ly@xmath51 and nv absorbers in ngc 985 at coincident velocities as those which co gas is seen projected against the nucleus .
the data bring up the interesting speculative suggestion that perhaps the absorption line outflows seen in these systems are physically associated with the co emitting outflows .
if so , it may be that the neutral and ionized outflows exist more toward the outer envelopes of the entrained complexes of molecular clouds as the conditions toward the central regions of these flows are extremely cold and dense ( @xmath010 - 30 k , @xmath010@xmath89 ; figure [ figure : vc3vc3e_outflow ] ) .
a full investigation into the potential relationship between uv and optical absorbers and co outflows is beyond the scope of this work , but studies into possible physical motivations for absorption line outflows within the context of these models are underway . in principle , unresolved emission line profiles from galaxies are more easily obtainable in large numbers than the high - resolution maps necessary to directly image outflows .
brief inspection by eye suggests that at least some local galaxies appear to espouse the high velocity peaks discussed in
[ section : lineprofiles ] ( e.g. narayanan et al . 2005 ; sanders scoville & soifer , 1991 ; solomon , downes & radford , 1992 ; yao et al .
as previously discussed , robustly determining the physical mechanism driving high velocity peaks is difficult , although tests similar to figure [ figure : vspike_noutflows ] may be possible with some prior knowledge of the host circular velocity .
it is interesting to note that absorption line outflows , which have been attributed to starburst - driven winds , have been detected in many of the ulirgs in the aforementioned co samples ( martin , 2005 ; rupke et al .
2005a - c ) .
apparent high velocity peaks appear prodigiously in 2 submillimeter galaxies ( smgs ; greve et al .
2005 ; tacconi et al . 2006 ; see blain et al .
, 2002 and solomon & vanden bout 2006 for recent reviews ) .
if the high velocity peaks in smgs owe to outflows , they may be explained by the prodigious star formation rates in these galaxies , in combination with a potential contribution from an embedded agn .
for example , if the extreme luminosities owe solely to starbursts , then massive star formation rates of order @xmath01000 are typical of these sources ( e.g. blain et al .
the relative role of agn energy input in smgs is a debated topic , as well as whether they are currently undergoing , or are soon to undergo a heavy black hole accretion phase ( e.g. alexander et al .
2005a , b ; borys et al , 2005 ; chakrabarti et al .
2007b ) . in either case ,
powerful winds produced by the intense @xmath010@xmath42sfr and/or the central black hole could potentially drive high velocity peaks .
that said , we caution against a direct application of these model results toward smgs .
for one thing , the virial properties of the progenitor galaxies in these models have not been explicitly scaled for 2 ( robertson et al .
2006a ) , and it is not clear how that will affect merger - induced star formation rates and black hole growth during the active period .
moreover , it is not completely clear that smgs are the product of binary mergers as this work has focused on , and the kinematic profiles in these sources may be rather complex .
models appropriate for 2 mergers ( e.g. chakrabarti et al . 2007b ; robertson et al .
2006a ) will have to be studied in the context of galactic winds in order to asses the potential relationship between these results and smgs .
we have combined 3d non - lte radiative transfer calculations with sph simulations of galaxy mergers in order to investigate the effects of galactic - scale winds on the molecular line emission in starburst galaxies and agn .
we find that galactic winds are a natural result of merger - induced star formation and black hole growth .
these winds can entrain molecular gas of order @xmath010@xmath1 - 10@xmath2 which imprints generic signatures in both the co morphology as well as unresolved emission line profiles .
the specifics of the morphological and emission line indicators of molecular outflows depend on physical parameter choices within the galaxy merger models .
in particular , the energy source ( i.e. bh accretion or star formation ) , as well as the merger orientation can vary the strength , direction , and duration of molecular outflows in emission contour maps , as well as the velocity separation of high velocity peaks in multi - component emission lines . many of the results including the halflight radius as a function of excitation , or the velocity offsets of high velocity peaks in the emission lines vary enough between physical parameter choices such that these models may be used to constrain physical origins for observed co morphologies and line profiles . in detail , we find the following : 1 .
molecular outflows entrained in agn feedback and starburst - driven winds can be directly imageable via co emission line mapping .
these types of outflows have been recorded in at least some local systems ( e.g. iono et al .
these outflows will be detectable at cosmological distances given the predicted spatial resolution of alma .
molecular outflows entrained in winds driven by agn feedback are typically longer lived than those entrained in starburst driven winds .
this owes to the relative strength of agn feedback - driven winds versus starburst driven winds .
winds from agn feedback in coplanar merger models are typically more powerful than mergers which occur at more random orientations .
this results in outflows in these models being visible for the majority of the @xmath0200 myr active period studied here .
the spatial extent of co emission can be controlled by the presence of galactic winds .
the emission is seen to be stratified with transition in all models such that the co ( j=1 - 0 ) halflight radius is larger than the radius from higher lying transitions ( e.g. co j=7 - 6 ) , though the degree of stratification depends on the inclusion of winds .
the relatively compact nature of observed co emission in local mergers ( r@xmath116 typically confined to the central kpc ; scoville & bryant , 1999 ) may be a consequence of agn feedback or starburst - driven winds .
5 . in all models , high velocity peaks ( peaks at velocities greater than the circular velocity ) can exist superposed on the post - merger galaxy s broad co emission line . in models without winds ,
these peaks owe to random kinematics of molecular gas . in models with winds
, these peaks are seen to originate primarily from gas entrained in outflow , at least during the period of peak black hole accretion / star formation .
6 . for the models studied here ( table [ table : ics ] )
, high velocity peaks driven by random kinematics do not typically appear at velocity offsets ( from systemic ) greater than @xmath01.7 times the circular velocity of the post - merger galaxy .
in contrast , peaks entrained in agn feedback - powered winds can be driven to velocities near 2.5 times the circular velocity .
the centroid velocities of the simulated lines are typically ( 1@xmath5 ) within @xmath020 of the true systemic velocity , and can generally be used as a reliable substitute for the systemic velocity in these models .
thus the above results hold true when measuring the velocity offsets of high velocity peaks with respect to line centroids as well .
we thank john bieging , sukanya chakrabarti , arjun dey , chris groppi , fred hamann , daisuke iono , casey papovich , yancy shirley , and linda tacconi for helpful conversations .
dn was funded for this work by an nsf graduate research fellowship .
the calculations were performed at the harvard - smithsonian center for parallel astrophysical computing .
br was supported in part by nasa through the spitzer space telescope fellowship program .
support for this work was also provided by nasa through grant number hst - ar-10308 from the space telescope science institute , which is operated by aura , inc . under nasa contract nas5 - 26555 .
alexander , d.m . ,
bauer , f.e . ,
chapman , s.c . ,
smail , i. , blain , a.w . , brandt , w.n .
, ivison , r.j . , 2005a , , 632 , 736 alexander , d.m . ,
smail , i. , bauer , f.e . ,
chapman , s.c . ,
blain , a.w . ,
brandt , w.n .
, ivison , r.j .
, 2005b , _ nature _ , 434 , 38 alton , p.b . , davies , j.i . , bianchi , s. , 1999 , a&a , 343 , 51 andre , p. , ward - thompson , d. , motte , f. , 1996 , a&a , 314 , 625 appleton , p. , charmandaris , v. , gao , y. , combes , f. , ghigo , f. , horellou , c. , mirabel , i.f .
, 2002 , , 566 , 682 armus , l. , et al . 2004 , apjs , 154 , 178 armus , l. , bernard - salas , j. , spoon , h.w.w . ,
marshall , j.a . ,
charmandaris , v. , higdon , s.j.u . ,
desai , v. , hao , l. , teplitz , h.i . , devost , d. , brandl , b.r . ,
soifer , b.t . ,
houck , j.r .
, 2006 , , 640 , 204 barnes , j.e . ,
hernquist , l.e . ,
1991 , , 370 , l65 barnes , j.e . , hernquist , l.e . , 1996 , , 471 , 115 bayet , e. , gerin , m. , phillips , t.g . ,
& contursi , a. , 2004 , a&a , 427 , 45 bernes , c. , 1979 , a&a , 73 , 67 scoville , n.z . , 1999 , , 117 , 2632 blain , a.w . ,
kneib , j .-
ivison , r.j . ,
smail , i. , 1999 , , 512 , l87 blitz , l. , fukui , y. , kawamura , a. , leroy , a. , mizuno , n. , rosolowsky , e. , 2006 , `` giant molecular clouds in local group galaxies '' , in _ protostars and planets v _ , [ astro - ph/0602600 ] bondi , h. , hoyle , f. , 1944 , mnras , 104 , 273 borys , c. , smail , i. , chapman , s.c . ,
blain , a.w . ,
alexander , d.m . ,
ivison , r.j . , 2005 , , 635 , 853 bradford , c.m .
, nikola , t. , stacey , g. , bolatto , a.d . ,
jackson , j.m . ,
savage , m.l . ,
davidson , j.a . ,
higdon , s.j .
, 2003 , , 586 , 891 bradford , c.m . , stacey , g.j . ,
nikola , t. , bolatto , a.d . ,
jackson , j.m . ,
savage , m.l . ,
davidson , j.a .
, 2005 , , 623 , 866 bryant , p. , scoville , n. , 1999 , , 117 , 2632 chakrabarti , s. , cox , t.j . , hernquist , l. , hopkins , p.f . ,
robertson , b. , di matteo , t. , 2007a , , 658 , 840 chakrabarti , s. , fenner , y. , hernquist , l. , cox , t.j . , hopkins , p.f . , 2007b , apj , submitted [ astro - ph/0610860 ] combes , f. , maoli , r. , omont , a. , 1999 , a&a , 345 , 369 cox , t.j . , jonsson ,
p. , primack , j.r . ,
somerville , r.s .
, 2004 , , 607 , l87 cox , t.j . , dutta , s.n . , di matteo , t. , hernquist , l. , hopkins , p.f . , robertson , b. , springel , v. , 2006a , , 650 , 791 cox , t.j . ,
di matteo , t. , hernquist , l. , hopkins , p.f . ,
robertson , b. , springel , v. , 2006b , , 643 , 692 cox , t.j . , jonsson , p. , primack , j.r . ,
somerville , r.s .
, 2006c , , 373 , 1013 cox , t.j . , chakrabarti , s. , di matteo , t. , hernquist , l. , hopkins , p.f .
, krause , e. , li , y. , robertson , b. , springel , v. , 2007 , , submitted de grijp , m.h.k .
, miley , g.k .
, lub , j. , de jong , t. , 1985 , _ nature _ , 314 , 240 di matteo , t. , springel , v. , hernquist , l. , 2005 , _ nature _ , 433 , 604 downes , d. , solomon , p.m. , 1998 , , 507 , 614 elmegreen , b. , falgarone , e. , 1996 , , 471 , 816 fabian , a.c .
, 1999 , , 308 , l39 farrah , d. , afonso , j. , efstathiou , a. , rowan - robinson , m. , fox , m. , clements , d. , 2003 , , 343 , 585 ferrarese , l. , merritt , d. , 2000 , apj , 539 , l9 fuller , g.a . , myers , p.c . , 1992 , , 384 , 523 ganguly , r. , sembach , k.r . , tripp , t.m . , savage , b.d . , wakker , b.p . , 2006 , , 645 , 868 gao , y. , solomon , p.m. , 1999 , , 512 , l99 gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s.m . , filippenko , a.v . , green , r. , grillmair , c. , ho , l.c . , kormendy , j. , lauer , t.r
. , magorrian , j. , pinkney , j. , richstone , d. , tremaine , s. , 2000 , 539 , l13 greve , t. , bertoldi , f. , smail , i. , neri , r. , chapman , s.c . ,
blain , a.w .
, ivison , r.j . ,
genzel , r. , omont , a. , cox , p. , tacconi , l. , kneib , j .-
, 2005 , , 359 , 1165 greve , t. , sommer - larsen , j. , 2006 , , in press [ astro - ph/0608683 ] hamann , f. , barlow , t.a . , junkkarinen , v. , 1997 , , 478 , 87 heckman , t.m . , lehnert , m.d . ,
strickland , d.k .
, & armus , l. , 2000 , , 129 , 493 hernquist , l. , 1989 , _ nature _ , 340 , 687 hernquist , l. , 1990 , , 356 , 359 hernquist , l. , quinn , p.j . , 1987 ,
apj , 312 , 1 hernquist , l. , spergel , d.n . , 1992 ,
apj , 399 , l117 hollenbach , d.j . ,
werner , m.w . , & salpeter , e.e .
, 1971 , , 163 , 165 hollenbach , d. j. , & tielens , a. g. g. m. 1999 , reviews of modern physics , 71 , 173 hopkins , p.f . , hernquist , l. , martini , p. , cox , t.j . , robertson , b. , di matteo , t. , springel , v. , 2005a , , 625 , l71 hopkins , p.f . , hernquist , l. , cox , t.j . ,
di matteo , t. , martini , p. , robertson , b. , springel , v. , 2005b , , 630 , 705 hopkins , p.f . , hernquist , l. , cox , t.j .
, di matteo , t. , robertson , b. , springel , v. , 2005c , , 630 , 716 hopkins , p.f . , hernquist , l. , cox , t.j .
, di matteo , t. , robertson , b. , springel , v. , 2005d , , 632 , 81 hopkins , p.f . , hernquist , l. , cox , t.j .
, di matteo , t. , robertson , b. , springel , v. , 2006a , , 163 , 1 hopkins , p.f . , hernquist , l. , cox , t.j . ,
robertson , b. , springel , v. , 2006b , , 163 , 50 hopkins , p.f . , hernquist , l. , cox , t.j . ,
robertson , b. , di matteo , t. , springel , v. , 2006c , , 639 , 700 hopkins , p.f . ,
somerville , r. , hernquist , l. , cox , t.j . ,
robertson , b. , li , y. , 2006d , , 652 , 864 hopkins , p.f . , hernquist , l. , cox , t.j , robertson , b. , krause , e. , 2007a , , in press [ astro - ph/0707.4005 ] hopkins , p.f . , hernquist , l. , cox , t.j , robertson , b. , krause , e. , 2007b , , in press [ astro - ph/0701351 ] hopkins , p.f . , hernquist , l. , cox , t.j , keres , d. , 2007c , , submitted [ astro - ph/0706.1243 ] hopkins , p.f . , cox , t.j , keres , d. hernquist , l. , 2007d , , submitted [ astro - ph/0706.1246 ] hopkins , p.f . , et al . , 2007e , apj , in preparation hopkins , p.f .
, et al . , 2007f , apj , in preparation hopkins , p.f . , et al . , 2007 g , apj , in preparation hopkins , p.f . ,
richards , g.t . , hernquist , l. , 2007h , , 654 , 731 hopkins , p.f . ,
hernquist , l. , 2006 , , 166 , 1 hoyle , f. , lyttleton , r.a . , 1939 , in proceedings of the cambridge philosophical society , 405 iono , d. , ho , p.t.p .
, yun , m. , matsushita , s. , peck , a. , sakamoto , k. , 2004 , , 616 , l63 iono , d. , yun , m.s . , ho , p.t.p . , 2005 , , 158 , 1 iono , d. , wilson , c.d . ,
takakuwa , s. , yun , m.s . ,
petitpas , g.r . , peck , a.b .
, ho , p.t.p . , matsushita , s. , pihlstrom , y. , wang , z. , 2007 , , 659 , 2831 kauffmann , g. , haehnelt , m.g . , 2000 , , 311 , 576 kennicutt , r. , 1998a , , 498 , 541 kennicutt , r. , 1998b , ara&a , 36 , 189 keres , d. , yun , m. , young , j. , 2003 , , 582 , 659 kim , d .-
, & sanders , d.b . , 1998 , , 119 , 41 kim , d .-
c . , veilleux , s. , sanders , d.b . , 2002 , apjs , 143 , 277 komossa , s. , burwitz , v. , hasinger , g. , predehl , p. , kaastra , j.s . ,
ikebe , y. , 2003 , , 582 , l15 kormendy , j. , fisher , d.b . ,
cornell , m.e .
, bender , r. , 2007 , apj , submitted kulesa , c.a . , 2002 , phd thesis , _ molecular hydrogen and its ions in dark interstellar clouds and star forming regions _ , university of arizona kulesa , c.a . ,
hungerford , a.l . ,
walker , c.k .
, zhang , x. , lane , a.p .
, 2005 , , 625 , 194 lee , h.h . ,
bettens , r.p.a . , herbst , e. , 1996 , a&as , 111 li , y. , hernquist , l. , robertson , b. , cox , t.j .
, hopkins p. , springel , v. , gao , l. , di matteo , t. , zentner , a. , jenkins , a. , yoshida , n. , 2007a , , in press [ astro - ph/0608190 ] li , y. , hernquist , l. , finkbeiner , d. , springel , v. , jiang , l. , fan , x. , 2007b , lidz , a. , mcquinn , m. , zaldarriaga , m. , hernquist , l. , dutta , s. , 2007 , apj , submitted [ astro - ph/0703667 ] magorrian , j. , tremaine , s. , richstone , d. , bender , r. , bower , g. , dressler , a. , faber , s.m . , gebhardt , k. , green , r. , grillmair , c. , kormendy , j. , lauer , t. , 1998 , aj , 115 , 2285 martin , c.l .
, 2005 , , 621 , 227 mihos , j.c . ,
hernquist , l.e . , 1994a , , 437 , l47 mihos , j.c . ,
hernquist , l.e . , 1994b , , 431 , l9 mihos , j.c . ,
hernquist , l.e . , 1996 , , 464 , 641 misawa , t. , eracleous , m. , charlton , j. , tajitsu , a. , 2005 , , 629 , 115 misawa , t. , charlton , j.c . ,
eracleous , m. , ganguly , r. , tytler , d. , kirkman , d. , suzuki , n. , lubin , d. , 2007a , , in press [ astro - ph/0702101 ] misawa , t. , eracleous , m. , charlton , j.c . ,
kashikawa , n. , 2007b , , 660 , 152 morganti , t. , tadhunter , c.n . ,
oosterloo , t.a . , 2005 ,
a&a , 444 , l9 narayanan , d. , hamann , f. , barlow , t. , burbidge , e.m . ,
cohen , r.d . ,
junkkarinen , v. , lyons , r. , 2004 , , 601 , 715 narayanan , d. , cox , t.j . ,
robertson , b. , dav , r. , di matteo , t. , hernquist , l. , hopkins , p. , kulesa , c. , walker , c.k .
, 2006a , , in press [ astro - ph/0604074 ] narayanan , d , kulesa , c , boss , a. , walker , c.k .
, 2006b , , 647 , 1426 narayanan , d. , li , y. , cox , t.j . ,
hernquist , l. , hopkins , p. , chakrabarti , s. , dav , r. , di matteo , t. , gao , l. , kulesa , c. , robertson , b. , walker , c.k . , 2007a ,
, in press [ arxiv:0707.3141 ] narayanan , d. , cox , t.j . , shirley , y. , dav , r. , hernquist , l. , walker , c.k . , 2007b , , in press [ arxiv:0711.1361 ] oppenheimer , b.d . ,
dav , r. , 2006 , mnras , 373 , 1265 pelupessy , f.i . , papadopoulos , p.p . , van der werf , p. , 2006
, , 645 , 1024 pettini , m. , rix , s.a . , steidel , c.c . , hunt , m.p . , shapley , a. , adelberger , k.l
, 2002 , ap&ss , 281 , 461 quinn , p.j .
, 1984 , apj , 279 , 596 richards , g.t . , et al . , 2006 , aj , 131 , 2766 robertson , b. , hernquist , l. , cox , t.j . ,
di matteo , t. , hopkins , p.f . , martini , p. , springel , v. , 2006a , , 641 , 90 robertson , b. , bullock , j. , cox , t.j . , di matteo , t. , hernquist , l. , springel , v. , yoshida , n. , 2006b , , 645 , 986 robertson , b. , cox , t.j . , hernquist , l. , franx , marijn , hopkins , p.f . , martini , p. , springel , v. , 2006c , , 641 , 21 robertson , b. , kravtsov , a. , 2007 , , in press [ arxiv:0710:2102 ] rosolowsky , e. , 2005 , , 117 , 1403 rosolowsky , e. , 2007 , , 654 , 240 rothberg , b. , joseph , r.d .
, 2004 , aj , 128 , 2098 rothberg , b. , joseph , r.d . , 2006 , aj , 131 , 185 rupke , d. , veilleux , s. , sanders , d.b . , 2005a , , 160 , 87 rupke , d. , veilleux , s. , sanders , d.b . , 2005b , , 160 , 115 rupke , d. , veilleux , s. , sanders , d.b . ,
2005c , , 632 , 751 rupke , d. , veilleux , s. , 2005 , , 631 , l37 sakamoto , k. , scoville , n.z .
, yun , m. , crosas , m , , genzel , r. , & tacconi , l.j . , 1999 , , 514 , 68 sakamoto , k. , matsushita , s. , peck , a.b . ,
wiedner , m.c . ,
iono , d. , 2004 , , 616 , l59 sakamoto , k. , ho , p.t.p . ,
peck , a. , 2006 , , 644 , 862 sanders , d.b . ,
soifer , b.t . ,
elias , j.h . ,
neugebauer , g. , matthews , k. , 1988a , , 328 , l35 sanders , d.b . , soifer , b.t . , elias , j.h . ,
madore , b.f . ,
matthews , k. , neugebauer , g. , scoville , n.z .
, 1988b , , 325 , 74 sanders , d.b . , scoville , n.z . , soifer , b.t . , 1991 , , 370 , 158 sanders , d.b . , mirabel , i.f . , 1996 , , 34 , 749 schmidt , m. , 1959 , , 129 , 243 schoier , f.l .
, van der tak , f.f.s . , van dishoeck , e.f . , black , j.h . , 2005 ,
a&a , 432 , 369 scoville , n. , young , j.s . , 1983 , , 265 , 148 scoville , n.z . , sanders , d.b . , sargent , a.i . , soifer , b.t . , scott , s.l . , lo , k.y . , 1986 , , 311 , l47 shapley , a.e . ,
steidel , c.c . ,
pettini , m. , adelberger , k.l . , 2003 , , 588 , 65 silk , j. , rees , m.j . , 1998 ,
a&a , 331 , l1 silk , j. , spaans , m. , 1997 , , 488 , l79 soifer , b.t . , sanders , d.b . ,
madore , b.f . ,
neugebauer , g. , danielson , g.e . , elias , j.h . ,
lonsdale , c.j . , rice , w.l .
, 1987 , , 320 , 238 solomon , p.m. , rivolo , a.r . ,
barrett , j. , yahil , a. , 1987 , , 319 , 730 solomon , p.m. , downes , d. , radford , s.j.e .
, 1992 , , 387 , l55 solomon , p.m. , vanden bout , p. , 2005
, ara&a , 43 , 677 springel , v. , 2005 , , 364 , 1105 springel , v. , hernquist , l. , 2002 , mnras , 333 , 649 springel , v. , hernquist , l. , 2003 , mnras , 339 , 312 springel , v. , hernquist , l. , 2005 , , 622 , l9 springel , v. , di matteo , t. , hernquist , l. , 2005a , , 361 , 776 springel , v. , di matteo , t. , hernquist , l. , 2005b , , 620l , 79 tacconi , l.j . , neri , r. , chapman , s.c . ,
genzel , r. , smail , i. , ivison , r.j . ,
bertoldi , f. , blain , a. , cox , p. , greve , t. , omont , a. , 2006 , , 640 , 228 tran , q.d . ,
lutz , d. , genzel , r. , rigopoulou , d. , spoon , h.w.w . , sturm , e. , gerin , m. , hines , d.c .
, moorwood , a.f.m . , sanders , d.b . , scoville , n. , taniguchi , y. , ward , m. , 2001 , , 552 , 527 tremonti , c.,a . ,
moustakas , j. , diamond - stanic , a.m. , 2007 , , in press [ arxiv0706.0527 ] trump , j.r . , hall , p .. b . , reichard , t.a . ,
richards , g.t . , schneider , d.p .
, vanden berk , d.e . ,
knapp , g.r . ,
anderson , s.f . , fan , x. , brinkman , j. , kleinman , s.j . ,
nitta , a. , 2006 , , 165 , 1 van zadelhoff , g - j .
, dullemond , c. , van der tak , f.f.s . ,
yates , j.a . , doty , s.d .
, ossenkopf , v. , hogerheijde , m.r . , juvela , m. , wiesemeyer , h. , schoier , f.l , 2002 , a&a , 395 , 373 veilleux , s. , kim , d .- c . , sanders , d.b . , 2002 ,
apjs , 143 , 315 veilleux , s. , cecil , g. , bland - hawthorn , j. , 2005 , , 43 , 769 wada , k. , tomisaka , k. , 2005 , , 619 , 93 walker , c.k . , adams , f.c . ,
lada , c.j .
, 1990 , , 349 , 515 walter , f. , wei , a. , scoville , n. , 2002 , , l21-l25 ward - thompson , d. , scott , p.f . , hills , r.e . , andre , p. , 1994 , , 268 , 276 weil , m. , hernquist , l. , 1992 , nature , 358 , 734 weil , m. , hernquist , l. , 1993 , apj , 405 , 142 wei , a. , downes , d. , walter , f. , henkel , c. , 2005a , a&a , 440,l45 wei , a. , walter , f. , scoville , n.z . , 2005b , a&a , 438 , 533 wei , a. , downes , d. , neri , r. , walter , f. , henkel , c. , wilner , d.j . ,
wagg , j. , wiklind , t. , 2007 , a&a , in press [ arxiv : astro - ph/0702669 ] wise , j.h . ,
eracleous , m. , charlton , j. , ganguly , r. , 2004 , , 613 , 129 yao , l. , seaquist , e.r . , kuno , n. , dunne , l. , 2003 , , 588 , 771 | galactic winds from starbursts and active galactic nuclei ( agn ) are thought to play an important role in driving galaxies along the starburst - agn sequence . here
, we assess the impact of these winds on the co emission from galaxy mergers , and , in particular , search for signatures of starburst and agn - feedback driven winds in the simulated co morphologies and emission line profiles .
we do so by combining a 3d non - lte molecular line radiative transfer code with smoothed particle hydrodynamics ( sph ) simulations of galaxy mergers that include prescriptions for star formation , black hole growth , a multiphase interstellar medium ( ism ) , and the winds associated with star formation and black hole growth .
our main results are : ( 1 ) galactic winds can drive outflows of masses @xmath010@xmath1 - 10@xmath2which may be imaged via co emission line mapping . ( 2 ) agn feedback - driven winds are able to drive detectable co outflows for longer periods of time than starburst - driven winds owing to the greater amount of energy imparted to the ism by agn feedback compared to star formation .
( 3 ) galactic winds can control the spatial extent of the co emission in post - merger galaxies , and may serve as a physical motivation for the sub - kiloparsec scale co emission radii observed in local advanced mergers .
( 4 ) secondary emission peaks at velocities greater than the circular velocity are seen in the co emission lines in all models , regardless of the associated wind model . in models with winds , however , these high velocity peaks are seen to preferentially correspond to outflowing gas entrained in winds , which is not the case in the model without winds .
the high velocity peaks seen in models without winds are typically confined to velocity offsets ( from the systemic ) @xmath31.7 times the circular velocity , whereas the models with agn feedback - driven winds can drive high velocity peaks to @xmath02.5 times the circular velocity . |
the last two decades since the inception of the iii - nitride based blue light emitting diodes ( leds ) @xcite , have witnessed the emergence of iii - nitrides as prominent material systems in state - of - the - art semiconductor technology @xcite . in particular , heterostructures based on al@xmath4ga@xmath5n / gan represent the building - blocks for a number of current ( opto)electronic devices , including blue and white leds @xcite , laser diodes and lasers @xcite , high - power- @xcite and high - electron - mobility - transistors ( hemts)@xcite . beside the tunability of the bandgap from the 3.4 ev of gan to the 6.2 ev in aln at room temperature , there is between gan and aln a 2.5% lattice mismatch , allowing the pseudomorphic growth of al@xmath4ga@xmath5n layers with a biaxial tensile strain on gan .
this leads to a compressive strain along the @xmath8-axis of the wurtzite ( wz ) structure , which induces a macroscopic electric field .
additionally , due to the lack of inversion symmetry , wz - iii - nitrides also have piezoelectric constants as large as @xmath10.73cm@xmath9 - 1.47cm@xmath9 that exceed by over two orders of magnitude those of relevant zinc - blende semiconductors , whose values are in the range @xmath10.01cm@xmath9 - 0.1cm@xmath9 [ ] . with _
ab initio _ calculations it was shown , that the macroscopic polarization in wurtzite nitrides is nonlinear@xcite .
while the nonlinearity of the spontaneous polarization depends on the microscopic structure of the alloy , the nonlinear piezoelectric polarization is a bulk effect .
the presence of this large spontaneous and induced polarization in al@xmath4ga@xmath5n / gan heterostructures leads to the formation of a 2-dimensional electron gas ( 2deg ) with electron densities exceeding 10@xmath10cm@xmath9 [ ] , even in the absence of nominal doping .
the polarization actually promotes the transfer of electrons from localized donor surface states at the barrier or even from the valence band to the quantum well at al@xmath4ga@xmath5n / gan / al@xmath4ga@xmath5n interfaces .
this property of iii - nitride heterostructures is extensively exploited for the fabrication of high power
high frequency devices like hemts@xcite .
the polarization charge sheet across al@xmath4ga@xmath5n / gan heterojunctions may be extended and spread into the al@xmath4ga@xmath5n by growing the latter with a compositional grading of al .
the spatially varying polarization @xmath12 will then have a fixed charge volume with a density given by @xmath13 . specifically , a compositional grading in the al@xmath4ga@xmath5n layer results in a non - vanishing divergence of the polarization , leading to the formation of a bulk three - dimensional ( 3d ) polarization charge background , @xmath14 to a polarization induced 3d electron slab ( 3des ) [ ] .
an analogous approach can also be exploited in a polarization doped hole system , where the grading may be designed to span from an al rich to an al depleted alloy@xcite .
one of the challenges in developing nitride based deep - uv leds and lasers is the doping of the structures with donor or acceptor impurities like si and mg [ ]
. the si activation energy in @xmath0-type al(ga)n can be as large as 280mev [ ] , while for the @xmath15-type case , the mg activation energy can be above 600mev for aln [ ] . taking these facts into account ,
polarization induced doping of wide bandgap iii - v nitride semiconductors may be viewed as an alternative and effective method to achieve high carrier concentration in otherwise insulating material systems , not only effective for the optimization of functional devices@xcite , but opening new avenues for studies of collective phenomena such as integral and fractional quantum hall effect@xcite , spin density waves , valley density wave states and wigner crystallization in bulk systems@xcite . in previous works on polarization doped @xmath0-al@xmath4ga@xmath5n and on 3des grown on a gan buffer by means of molecular beam epitaxy ( mbe ) [ ] bulk carrier densities up to @xmath1 10@xmath16 @xmath3 were reported .
@xcite studied the reservoir effect of various substrates on the carrier density of a 3des in graded al@xmath4ga@xmath5n ( g - al@xmath4ga@xmath5n ) layers grown by mbe . while jena _
@xcite reported electron mobilities as high as @xmath1 10@xmath17 @xmath18/vs in their 3dess , the values given by li _
[ ] are @xmath19@xmath18/vs in 3dess grown by mbe on semi - insulating and conducting gan templates . here , we report on 3des in g - al@xmath4ga@xmath5n layers grown on differently conducting buffer layers by metal organic vapor phase epitaxy ( movpe ) on @xmath8-sapphire substrates and reaching carrier concentrations up to @xmath1 10@xmath2 @xmath3 .
the samples are systematically characterized by atomic force microscopy ( afm ) , high resolution x - ray diffraction ( hrxrd ) and xrd reciprocal space mapping ( rsm ) , transmission electron microscopy ( tem ) also in high resolution ( hrtem ) and scanning mode ( stem ) in bright / dark field ( bf / df ) and high angle annular dark field ( haadf ) , electron energy loss spectroscopy ( eels ) and low temperature magnetotransport for hall effect and magnetoresistance ( mr ) . by combining the structural , chemical and magnetotransport studies we unravel the role of the buffer / reservoir on the carrier concentration in the 3dess and also the effects of dislocation density , interface and alloy scattering on the mobility observed in the samples .
we consider g - al@xmath4ga@xmath5n layers with 0.05 @xmath20 x @xmath20 0.37 and deposited on gan buffers ( reservoir ) having different electron concentration .
in particular , the gan - based reservoirs for the different samples are : a semi - insulating ( si ) gan : mn buffer , an unintentionally conducting ( u-)gan buffer with background electron concentration @xmath21 10@xmath22@xmath3 and a degenerate gan : si reservoir with @xmath21 10@xmath23@xmath3 . as reference , we take an al@xmath24ga@xmath25n : si / gan and a g - al@xmath4ga@xmath5n layer grown on a si gan : mn buffer . for all
graded al@xmath4ga@xmath5n the gradient of al concentration is @xmath6 : 5%@xmath730% over the layer thickness .
details on the s0-s5 studied samples structure and @xmath26 in the buffer / reservoir are provided in tablei .
the graded al@xmath5ga@xmath4n ( g - al@xmath5ga@xmath4n ) samples have been fabricated in an aixtron 200rf horizontal tube metal organic vapor phase epitaxy ( movpe ) reactor and deposited on a @xmath8-plane sapphire substrate using tmga , tmal , mncp@xmath27 , sih@xmath28 and nh@xmath29 as precursors for ga , al , mn , si and n respectively , with h@xmath27 as carrier gas . upon nitridation of the sapphire substrate , a low temperature gan nucleation layer ( nl ) is grown at 540@xmath30c and then annealed at 1040@xmath30c .
then , a 1@xmath31 m thick gan buffer is grown also at 1040@xmath32c for sample s0-s2 , while for samples s3-s5 , a 1@xmath31 m thick semi - insulating gan : mn buffer with @xmath10.6% mn cations concentration is deposited at 1040@xmath32c . on the buffer , graded al@xmath5ga@xmath4n layers have been grown by continuously increasing the flow of the tmal precursor from 0 to 35 standard cubic centimeters per minute ( sccm ) at a substrate temperature of 1000@xmath32c . in samples s3 and s4 , instead , a 500 nm thick intermediate layer of degenerate _ n_-gan : si is inserted between the si gan : mn buffer and the graded al@xmath5ga@xmath4n .
all steps of the growth process are monitored by _ in situ _ kinetic and spectroscopic ellipsometry .
the hrxrd and rsm measurements have been carried out with a panalytical s xpert pro materials research diffractometer ( mrd ) equipped with a hybrid monochromator with a 1/4@xmath32 divergence slit .
the diffracted beam is measured with a solid - state pixcel detector used as 256-channels detector with a 11.2 mm anti - scatter slit . in the radial @xmath33 scans for @xmath34 values between @xmath35 and @xmath36 ( not shown here ) only reflections from the @xmath8-al@xmath27o@xmath29 substrate , from the gan buffer / reservoir and from g - al@xmath4ga@xmath5n ( al@xmath4ga@xmath5n : si ) are detectable .
the radial hrxrd spectra of the symmetric ( 0004 ) reflection for the whole series of samples investigated in the present study are shown in fig.[fig : fig1](a ) .
while a sharp peak related to the al@xmath4ga@xmath5n ( 0004 ) reflection is detected for sample s0 , for samples s1-s5 containing a g - al@xmath4ga@xmath5n the multiple peaks at the al@xmath4ga@xmath5n ( 0004 ) reflection are fingerprints of the gradient in the al concentration . in order to gain a deeper insight into the distribution of al in the samples , rsm of the symmetric ( 0002 ) reflections
have been acquired and are reported in figs.[fig : fig1 ] ( b)-(d ) for samples s0 , s1 , and s4 .
the broadening in the direction q@xmath37 ( @xmath38-scan ) is determined by the tilt and finite lateral coherence length of mosaic blocks @xcite .
in figs.[fig : fig1 ] ( c)-(d ) , the elongation along the q@xmath39 direction points to a pseudomorphically grown al@xmath4ga@xmath5n layer with graded al concentration on gan . however , for sample s0 as evidenced in fig.[fig : fig1 ] ( b) this elongation of al@xmath4ga@xmath5n ( 0002 ) along q@xmath40 is not detected , confirming the presence of a non - graded al@xmath4ga@xmath5n : si layer with a constant @xmath6@xmath125% .
measurements of tem also in stem mode are performed in a fei titan cube 80 - 300 operating at 300kev and in a jeol 2010f working at 200kev .
bright / dark - field bf / df , hrtem and haadf are employed to analyse the structure of the sample , while chemical mapping is achieved through eels analysis .
cross - section tem specimens are prepared by mechanical polishing , dimpling and final ion milling in a gatan precision ion polishing system .
the tem micrograph of sample s1 reported in fig.[fig : fig2 ] ( a ) evidences the al composition changes in the 75 nm thick g - al@xmath4ga@xmath5n ( @xmath6 : 0%@xmath737% ) layer , and the eels data in fig.[fig : fig2 ] ( b ) confirm the increase of the al concentration in the g - al@xmath4ga@xmath5n corresponding to the nominal increment of al flow rate during the process . on the other hand , for the 30 nm thin g - al@xmath4ga@xmath5n ( @xmath6 : 0%@xmath737% ) layer in s4 , from the tem image in fig.[fig : fig2 ] ( c )
it is not possible to appreciate the contrast between the g - al@xmath4ga@xmath5n regions with different al content , but the monotonic increment of the al concentration with the thickness is confirmed by the eels data in fig.[fig : fig2 ] ( d ) .
this is in accord with the presence of a @xmath415 nm wide intermixing region around the interfaces between nominal variations in al detected @xmath42 haadf measurements as reported in the inset to fig.[fig : fig2 ] ( b ) for the interface al@xmath4ga@xmath5n / gan . for the hall effect measurements and magnetoresistance studies , a four - probe van der pauw geometry is employed .
the hall effect measurements are carried out in dc mode , while the mr studies are performed using a low frequency ac technique .
the ohmic contacts for all samples consist of a ti / au / al / ti / au metal stack deposited in an e - beam evaporation chamber and then annealed at 750@xmath30c in n@xmath27 atmosphere for 30 seconds .
the above results of structural and chemical analysis of the samples confirm the presence of a controlled concentration gradient of al in the wanted volume and the consequent non - vanishing divergence of the polarization is expected to determine the electrical properties of the samples .
the combination of spontaneous and piezoelectric polarization p(x ) for an al@xmath4ga@xmath5n layer coherently strained to gan is given by@xcite : @xmath43=[2x+1.1875x^{2}+3.25x]\times 10^{13}\end{aligned}\ ] ] for a fixed concentration @xmath6 of al .
however , in the presence of a gradient in the content @xmath6 of al over the al@xmath4ga@xmath5n layer thickness , the spatial change of the polarization leads to doping driven by dipole engineering and the induced charge n@xmath44 is given by@xcite : @xmath45 where x@xmath46 is the concentration of al at a position @xmath47 from the interface with the buffer and along the growth direction @xmath48 ( @xmath8 ) . by applying eq.2 and by implementing the al concentration as obtained from eels measurements , the value of n@xmath44 calculated at 30 nm from the interface with the buffer for sample s4 , is estimated to be ( 1.25@xmath4910@xmath23)@xmath3 .
a remarkable feature of this method of doping is that the doping level can be controlled by tuning the alloy composition and/or the thickness of the graded layer .
the de broglie wavelength @xmath50 is @xmath4117 nm for non - degenerate carrier densities , while the fermi - wavelength @xmath51 for degenerate systems ( @xmath110@xmath2@xmath3 ) is calculated to be @xmath130 nm .
any thickness of the graded layer which is smaller than these two characteristic lengths , leads to the formation of a quasi-2deg .
however , for wide slabs of graded regions with @xmath52 , the electron gas is 3d . in fig.[fig :
fig3 ] ( a ) the electron concentration @xmath26 and in fig.[fig : fig3 ] ( b ) the hall mobilities as a function of temperature in the range 40k@xmath20t@xmath20300k are reported for samples s0-s4 ( s5 is semi - insulating ) . according to the curves in fig.[fig : fig3 ] ( a ) , all the g - al@xmath4ga@xmath5n samples have carrier densities @xmath53@xmath3 , i.e. well above the critical concentration for the metal - to - insulator transition ( mit ) in _ n_-gan : si@xcite . by exploiting the mott model , where the critical carrier concentration @xmath54 is related to the bohr radius of the relevant dopant
a@xmath55 as @xmath56 @xcite , we calculate @xmath54 for _ n_-al@xmath57n , _ n_-al@xmath58ga@xmath59n and _ n_-al@xmath60ga@xmath61n as ( 2@xmath4910@xmath16)@xmath3 , ( 2.7@xmath4910@xmath16)@xmath3 and ( 4@xmath4910@xmath16)@xmath3 , respectively . by comparing these values with the hall data , we conclude that the reference al@xmath24ga@xmath25n : si is doped degenerately and is above the mit .
the polarization doped g - al@xmath4ga@xmath5n samples also exhibit carrier concentrations far above the mott limit for the corresponding al@xmath4ga@xmath5n systems doped with donor impurities like si .
these results indicate that polarization induced doping in the studied graded al@xmath4ga@xmath5n drives such systems to a degenerate regime with the formation of a 3des .
since all the samples discussed here have @xmath26 far above the mott mit limit , the graded al@xmath5ga@xmath4n layers are expected to be in a weak localization regime@xcite .
this is confirmed by the observation of negative mr ( positive magnetoconductance ) in the magnetoresistance data acquired as a function of the applied magnetic field and for various temperatures 2k@xmath20t@xmath2050k , as reported in fig.[fig : fig4 ] for sample s4 .
if the 3des is spread over a thickness @xmath62 , then the sheet carrier density of the 3des is calculated to be @xmath63 . for s4 ,
this thickness is then 30 nm , pointing to a 5 nm depletion of the 3des from the surface potential . the hall parameters at 50k and 300k for samples s0-s5 are listed in tablei .
[ cols= " < , < , < , < , < , < , < , < , < " , ] [ tab : tab1 ] * reservoir / buffer : layer on which g - al@xmath4ga@xmath5n and/or al@xmath4ga@xmath5n : si are grown .
sample s0 shows a carrier density of ( 2@xmath4910@xmath16)@xmath3 at 300k , which decreases with temperature .
this sample is at the boundary of the mit transition and for t@xmath2040k it can be driven to a non - degenerate electronic state . for samples
s1 and s2 , where the g - al@xmath4ga@xmath5n layer is grown on an u - gan buffer with @xmath26@xmath110@xmath22@xmath3 , the bulk carrier density at 300k is @xmath64(3@xmath4910@xmath16)@xmath3 , and it remains substantially constant over the measured temperature range .
however , for samples s3 and s4 , grown on a degenerate gan : si reservoir with @xmath65=(1.2@xmath4910@xmath23)@xmath3 , one can observe an enhanced carrier density , reaching ( 6@xmath4910@xmath23)@xmath3 for s3 and ( 1.4@xmath4910@xmath2)@xmath3 for s4 .
these values are greater than those observed for _ n_-gan : si [ ] and for 3des in mbe
g - al@xmath4ga@xmath5n / gan systems @xcite .
the difference between s3 and s4 resides solely in the thickness of the g - al@xmath4ga@xmath5n layer .
it is observed that sample s4 with a 35 nm thickness of the graded layer shows higher carrier concentration when compared to s3 , where the thickness of g - al@xmath4ga@xmath5n is 75 nm . also , from the observed carrier densities of the graded layers it can be concluded that the bulk concentration of carriers in the gan reservior / buffer plays a dominant role in determining the amount of charge carriers induced in a polarization doped g - al@xmath4ga@xmath5n system , where the graded layer grown on a degenerate gan : si shows a carrier concentration up to two orders of magnitude higher than when the g - al@xmath4ga@xmath5n layer is deposited on an u - gan buffer .
furthermore , the absence of significant free charge carriers in the buffer , like @xmath66 in s5 , keeps g - al@xmath4ga@xmath5n semi - insulating . t@xmath20300k . ] from fig.[fig : fig3 ] ( b ) we observe for samples s1-s4 carrier mobilities which are comparable@xcite or slightly lower than those observed in 3des fabricated by mbe@xcite .
however , the al concentration in the above mentioned works had a gradient in the range @xmath6:0%@xmath720% , while all the samples considered in this work have a al gradient @xmath6:0%@xmath737% .
a maximum mobility of 210 @xmath67/vs is achieved in s4 , while for s1 a value of 15.57 @xmath67/vs at 300k is obtained .
the samples s3 and s4 with a rich reservoir of electrons in the gan : si buffer have higher mobilities compared to s1 and s2 deposited on u - gan . from here it can be inferred , that the electronic properties ( carrier concentration ) of the reservoir / buffer directly affect the mobility in the 3des . the dislocation density and the dislocation - related scattering of the charge carriers in the 3des
play also a role in determining the carrier mobility of the 3des . from weak beam dark field image tem data on sample s1 , the screw- and edge - dislocation densities
are estimated in ( 1.0@xmath4910@xmath68)cm@xmath9 and ( 8.1@xmath4910@xmath69)cm@xmath9 , respectively , while for sample s4 , are of the order of ( 6.62@xmath4910@xmath68)cm@xmath9 and ( 1.26@xmath4910@xmath70)cm@xmath9 . in the calculation ,
the average specimen thickness is established by counting the dark fringes in the bf image under two beam conditions , and the extinction distance of gan used in the calculation is @xmath7156 nm [ ] .
+ dislocations affect the carrier transport mostly through two mechanisms , namely @xmath42 the local deformation they induce in the lattice and the accumulation of static charges along the dislocation line or the core effect scattering .
according to theoretical models , the main contribution to electron scattering due to dislocations in wz - iii - nitride systems , originates from @xmath72-edge dislocations and from the edge component of ( @xmath73 ) mixed dislocations . the contribution from @xmath8-screw dislocations
was shown to have a negligible effect on the electron scattering ( and consequently on the electron mobilities ) in al@xmath4ga@xmath5n / gan based structures@xcite . from tem measurements it is seen that the edge dislocation density in sample s4 is @xmath11.55 times greater than in s1 , but magnetotransport measurements reveal a hall mobility of @xmath1210 @xmath67/vs in s4 , much greater than the 15 @xmath67/vs of s1 .
this behaviour can be explained considering the carrier densities of the two samples .
while in s4 we observe a carrier density of @xmath1(1.3@xmath4910@xmath2)@xmath3 , the one measured in s1 is @xmath1(3.8@xmath74)@xmath3 . with an increase in carrier density ,
the dislocation band located below the conduction band becomes narrower , reducing the strength of the scattering potential .
thus , despite a higher edge dislocation density , we observe a greater electron mobility in sample s4 compared to s1 .
alloy disorder scattering and interface scattering additionaly affect the observed mobility in the 3des samples .
the negative mr is a signature of quantum corrections to conductivity due to weak localization confirming the degeneracy of the 3des . ]
in summary , g - al@xmath4ga@xmath5n layers on buffer / reservoir gan layers with different electron concentrations @xmath26 have been grown by movpe and thoroughly analysed .
the compositional grading of the al@xmath4ga@xmath5n layer leads to the formation of a 3des , whose hall concentration depends on the electron concentration in the buffer / reservoir .
the conductivity of the samples increases with increasing @xmath26 and this can be achieved either by augmenting the gradient range of the al content while not varying the thickness of the graded layer , or by decreasing the layer thickness for a fixed range of the gradient , or by using a buffer / reservoir with a carrier concentration orders of magnitude higher than the one of u - gan .
the observed magnetoresistance points to quantum corrections to the conductivity , due to weak localization@xcite . with the present work it is shown that controllable g - al@xmath4ga@xmath5n leading to a 3des can be fabricated in a movpe process .
the realization of such degenerate al@xmath4ga@xmath5n - based systems opens wide perspectives for the fabrication of efficient electrodes in nitride - based deep uv leds and lasers , active layers for diodes , transistors and spin devices like nitride - based spin field effect transistors .
this work was supported by the austrian science foundation fwf ( p22477 and p24471 and p26830 ) , by the nato science for peace programme ( project no .
984735 ) and by the eu @xmath76 framework programmes : capacities project regpot - ct-2013 - 316014 ( eagle ) , and fundms advanced grant of the european research council ( erc grant no.227690 ) .
42ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty \doibase http://dx.doi.org/10.1063/1.109374 [ * * , ( ) ] \doibase http://dx.doi.org/10.1116/1.579811 [ * * , ( ) ] http://stacks.iop.org/1882-0786/5/i=3/a=032101 [ * * , ( ) ] @noop _ _ , vol .
( , ) \doibase http://dx.doi.org/110.1038/nphoton.2008.135 [ * * , ( ) ] link:\doibase 10.1016/s0038 - 1101(98)00208 - 1 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4919866 [ * * , ( ) ] link:\doibase 10.1109/jproc.2002.1021567 [ * * , ( ) ] link:\doibase 10.1063/1.4907261 [ * * , ( ) ] link:\doibase 10.1103/physrevb.56.r10024 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.79.3958 [ * * , ( ) ] http://stacks.iop.org/0953-8984/14/i=13/a=302 [ * * , ( ) ] link:\doibase 10.1103/physrevb.64.085207 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.371396 [ * * , ( ) ]
link:\doibase 10.1103/physrevb.74.033302 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1339858 [ * * , ( ) ] link:\doibase 10.1109/led.2005.857701 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1526161
[ * * , ( ) ] link:\doibase 10.1103/physrevb.67.153306 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4792685 [ * * , ( ) ] link:\doibase 10.1126/science.1183226 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/s0921-4526(01)00417-3 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2061856 [ * * , ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4906449 [ * * , ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4833155 [ * * , ( ) ] link:\doibase 10.1109/led.2015.2394503 [ * * , ( ) ] @noop * * , ( ) http://iopscience.iop.org/1347-4065/26/s3-3/1913 [ * * , ( ) ] link:\doibase 10.1103/physrevb.45.13488 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.1062 [ * * , ( ) ] link:\doibase 10.1103/physrevb.67.195336 [ * * , ( ) ] link:\doibase 10.1103/physrevb.40.11634 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2168253 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4750039 [ * * , ( ) ] http://stacks.iop.org/0034-4885/72/i=3/a=036502 [ * * , ( ) ] and , eds . , link:\doibase 10.1007/978 - 0 - 387 - 68319 - 5 [ _ _ ] ( , ) link:\doibase 10.1103/physrevb.89.205201 [ * * , ( ) ] link:\doibase 10.1143/jpsj.77.031005 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1504876 [ * * , ( ) ] http://stacks.iop.org/0953-8984/20/i=32/a=325210 [ * * , ( ) ] | polarization induced degenerate @xmath0-type doping with electron concentrations up to @xmath110@xmath2@xmath3 is achieved in graded al@xmath4ga@xmath5n layers ( @xmath6 : 0%@xmath737% ) grown on unintentionally doped and on @xmath0-doped gan : si buffer / reservoir layers by metal organic vapor phase epitaxy .
high resolution x - ray diffraction , transmission electron microscopy and electron energy loss spectroscopy confirm the gradient in the composition of the al@xmath4ga@xmath5n layers , while magnetotransport studies reveal the formation of a three dimensional electron slab , whose conductivity can be adjusted through the gan(:si ) buffer / reservoir . |
a graph @xmath0 is word - representable if there exists a word @xmath1 over the alphabet @xmath12 such that letters @xmath3 and @xmath4 alternate in @xmath1 if and only if @xmath5 is an edge in @xmath13 .
it follows from definitions that word - representable graphs are a _ hereditary class _ of graphs . a comprehensive introduction to the theory of word - representable graphs
is given in @xcite .
a graph is word - representable if and only if it is @xmath7-word - representable for some @xmath7 , that is , if there exists a word containing @xmath7 copies of each letter that represents the graph ( see theorem [ repr - must - be - k - repr ] ) . by proposition [ k - implies - k - plus-1 ] , being @xmath7-word - representable implies being @xmath8-word - representable .
the minimum @xmath7 such that a word - representable graph is @xmath7-word - representable , is called _
graph s representation number_. we let @xmath14 denote the class of graphs having representation number @xmath7 . @xmath15 is easy to see to be the class of complete graphs ( also known as cliques ) , while @xmath16 is the class of circle graphs ( see theorem [ 2-repr - characterization ] ) . the only fact
known before this paper on @xmath9 was that the petersen graph ( to the right in figure [ fig : representable - graphs ] ) and the triangular prism ( to the left in figure [ prisms - pic ] ) belong to this class .
theorem [ known - r3 ] extends our knowledge on @xmath9 , in particular , showing that _ all _ prisms belong to this class ( also , see theorem [ prn - repr - num-3-thm ] ) . in section
[ glueing - connecting - two - graphs ] , we revise connecting two word - representable graphs , say @xmath17 and @xmath18 , by an edge and gluing these graphs in a vertex , originally studied in @xcite .
one can use a graph orientation argument involving theorem [ thm : rep - equals - semi - trans ] , or results in @xcite , to show that the resulting graph @xmath19 , in both cases , is word - representable .
however , these do not answer directly the following question : if @xmath17 is @xmath20-word - representable , @xmath18 is @xmath21-word - representable , and @xmath19 is @xmath7-word - representable ( such a @xmath7 must exist by theorem [ repr - must - be - k - repr ] ) then what can be said about @xmath7 ?
theorem [ connect - glue - thm ] answers the question .
further , in section [ modules - subsec ] we revise replacing a vertex in a graph with a module , briefly considered in @xcite .
theorem [ module - thm ] in that section is an extended version of an observation made in @xcite . in section [ sec - c - colorable - graphs ] , providing two arguments based on theorems [ connect - glue - thm ] and [ module - thm ] , we show that @xmath9 is not included in a class of @xmath10-colorable graphs for a constant @xmath10 . in section [ sec - ladder - graphs ]
we show that ladder graphs used in the study of prisms in @xcite are @xmath11-word - representable , and thus each ladder graph is a circle graph .
finally , in section [ crown - graphs - sec ] we discuss @xmath7-word - representing comparability graphs via consideration of crown graphs , where we state some problems for further research .
suppose that @xmath1 is a word and @xmath3 and @xmath4 are two distinct letters in @xmath1 .
we say that @xmath3 and @xmath4 _ alternate _ in @xmath1 if after deleting in @xmath1 all letters but the copies of @xmath3 and @xmath4 we either obtain a word @xmath22 ( of even or odd length ) or a word @xmath23 ( of even or odd length ) .
if @xmath3 and @xmath4 do not alternate in @xmath1 , we say that these letters are _ non - alternating _ in @xmath1 .
for example , if @xmath24 then the letters @xmath25 and @xmath26 are alternating in @xmath1 because removing all other letters we obtain @xmath27 , while @xmath26 and @xmath28 are non - alternating because removing all other letters from @xmath1 we have @xmath29 . a simple graph @xmath0 is _ word - representable _ if there exists a word @xmath1 over the alphabet @xmath2 such that letters @xmath3 and @xmath4 alternate in @xmath1 if and only if @xmath30 for each @xmath31 .
we say that @xmath1 _ represents _ @xmath19 , and @xmath1 is called a _ word - representant _ for @xmath19 .
the graphs in figure [ fig : representable - graphs ] are word - representable . indeed , for example , 1213423 is a word - representant for @xmath32 , 1234 is a word - representant for @xmath33 , and a word - representant for the petersen graph is @xmath34 ( left ) , the complete graph @xmath33 ( middle ) , and the petersen graph ( right ) . ] in what follows , we will need the following two propositions that are easy to see from definitions .
[ just - useful - observation](@xcite ) let @xmath35 be a word representing a graph @xmath19 , where @xmath36 , @xmath37 and @xmath38 are possibly empty words , and @xmath37 contains no @xmath3 . let @xmath39 be the set of all letters that appear only once in @xmath37 .
then possible candidates for @xmath3 to be adjacent to in @xmath19 are the letters in @xmath39 .
[ reverse - representant ] if @xmath1 represents @xmath19 , then the reverse @xmath40 , which is writing the letters in @xmath1 in the reverse order , also represents @xmath19 .
a word is _ @xmath7-uniform _ if each letter in it appears exactly @xmath7 times .
in particular , @xmath25-uniform words are permutations .
a graph is _ @xmath7-word - representable _ if there exists a @xmath7-uniform word representing it .
we say that the word _
@xmath7-represents _ the graph .
the following theorem is of special importance in the theory of word - representable graphs .
[ repr - must - be - k - repr](@xcite ) a graph @xmath19 is word - representable if and only if it is @xmath7-word - representable for some @xmath7 . in
what follows , we will also use the following two propositions .
[ k - implies - k - plus-1](@xcite ) a @xmath7-word - representable graph @xmath19 is also @xmath8-word - representable . in particular , each word - representable graph has infinitely many word - representants representing it since for every @xmath41 , a @xmath7-word - representable graph is also @xmath42-word - representable .
( @xcite)[cyclic - shift ] let @xmath43 be a @xmath7-uniform word representing a graph @xmath19 , where @xmath44 and @xmath45 are two , possibly empty , words
. then the word @xmath46 also represents @xmath19 .
a graph @xmath19 with the vertex set @xmath47 is _ permutationally representable _ if it can be represented by a word of the form @xmath48 , where @xmath49 is a permutation of @xmath2 for @xmath50 .
for example , a complete graph @xmath51 is permutationally representable for any @xmath52 .
indeed , take any permutation of @xmath53 and repeat it as many times ( maybe none ) as desired .
for another example , the path @xmath54 is also permutationally representable , and one such representation is 213231 , while some other such representations can be obtained from it by adjoining any number of permutations 213 and 231 to it .
if @xmath19 can be represented permutationally involving @xmath7 permutations , we say that @xmath19 is _ permutationally @xmath7-representable_. an orientation of graph s edges is called _ transitive _ if having ( directed ) edges @xmath55 and @xmath56 for some vertices @xmath57 implies having the edge @xmath58 .
a graph is a _ comparability graph _ if it accepts a transitive orientation .
( @xcite)[comparability - graphs ] a graph is permutationally representable if and only if it is a comparability graph .
comparability graphs correspond to partially ordered sets ( posets ) , and the question whether such a graph can be permutationally @xmath7-represented is equivalent to the question whether the respective poset can be represented as an intersection of @xmath7 linear orders .
the minimum such @xmath7 for a poset is called the _
poset dimension_. we call _ @xmath7-comparability graphs _ those graphs that correspond to posets having dimension @xmath7 , that is , those graphs that are permutationally @xmath7-representable but not permutationally @xmath59-representable .
a graph @xmath0 is _ semi - transitive _ if it admits an acyclic orientation such that for any directed path @xmath60 with @xmath61 for all @xmath62 , @xmath50 , either * there is no edge @xmath63 , or * the edge @xmath63 is present and there are edges @xmath64 for all @xmath65 .
that is , in this case , the ( acyclic ) subgraph induced by the vertices @xmath66 is transitive ( with the unique source @xmath67 and the unique sink @xmath68 ) .
we call such an orientation _ semi - transitive orientation_. for example , the orientation of the graph in figure [ fig : semi - transitive - orientation ] is semi - transitive , and thus the underlying ( non - directed ) graph is semi - transitive .
clearly , all transitive ( that is , comparability ) graphs are semi - transitive , and thus semi - transitive orientations are a generalization of transitive orientations .
semi - transitive orientations are defined in @xcite in terms of _ shortcuts _ as follows .
semi - cycle _ is the directed acyclic graph obtained by reversing the direction of one edge of a directed cycle .
an acyclic digraph is a shortcut if it is induced by the vertices of a semi - cycle and contains a pair of non - adjacent vertices .
thus , a digraph on the vertex set @xmath69 is a shortcut if it contains a directed path @xmath70 , the edge @xmath63 , and it is missing an edge @xmath64 for some @xmath71 ; in particular , we must have @xmath72 , so that any shortcut is on at least four vertices .
see figure [ fig : shortcut ] for an example of a shortcut ( there , the edges @xmath73 , @xmath74 , and @xmath75 are missing ) .
an orientation of a graph is semi - transitive , if it is acyclic and contains no shortcuts .
clearly , this definition is just another way to introduce the notion of semi - transitive orientations presented above .
the following theorem is a useful characterization of word - representable graphs that allows answering questions on word - representability in terms of graph orientations .
[ thm : rep - equals - semi - trans](@xcite ) a graph @xmath19 is word - representable if and only if it is semi - transitive ( that is , it accepts a semi - transitive orientation ) .
a direct corollary to the last theorem is the following statement .
( @xcite)[obs:3col ] @xmath26-colorable graphs are word - representable .
the following statement is easy to see .
[ 1-repr - characterization ] a graph @xmath19 is in @xmath15 if and only if @xmath76 , the complete graph on @xmath52 vertices , for some @xmath52 .
we also have a characterization of graphs in @xmath16 .
to state it we need the following definition : a _ circle graph _ is an undirected graph whose vertices can be associated with chords of a circle such that two vertices are adjacent if and only if the corresponding chords cross each other .
[ 2-repr - characterization](@xcite ) for a graph @xmath19 different from a complete graph , @xmath19 is in @xmath16 if and only if @xmath19 is a circle graph .
the following proposition is not difficult to see from definitions .
[ useful - remark - ind - subgraph ] if @xmath77 and @xmath78 is an induced subgraph of @xmath19 then @xmath79 . indeed , using the hereditary nature of word - representable graphs ,
if representing @xmath78 would require more than @xmath7 copies of each letter , then representing @xmath19 would obviously require more than @xmath7 copies of each letter .
( @xcite)[thm : di - to - string ] each word - representable graph on @xmath52 vertices is @xmath52-word - representable . . ]
it turns out that there are graphs on @xmath52 vertices with representation number of @xmath80 , matching the upper bound in theorem [ thm : di - to - string ] within a factor of 2 . a _ crown graph _
@xmath81 is a graph obtained from the complete bipartite graph @xmath82 by removing a perfect matching ( see figure [ crown - pic ] for examples of crown graphs ) .
then @xmath83 is the graph obtained from a crown graph @xmath81 by adding an all - adjacent vertex .
see figure [ fig : graph - g - k ] for graphs involved in defining @xmath84 , where note that all choices in defining @xmath85 lead to isomorphic graphs .
( @xcite)[thm : example ] the graph @xmath83 on @xmath86 vertices belongs to @xmath14 .
the following theorem gives a useful tool to construct @xmath26-word - representable graphs , that is , graphs with representation number at most @xmath26 .
[ add - path - to - graph](@xcite ) let @xmath0 be a @xmath26-word - representable graph and @xmath87 . denote by @xmath88 the graph obtained from @xmath19 by adding to it a path of length at least @xmath26 connecting @xmath3 and @xmath4 . then @xmath88 is also @xmath26-word - representable .
we also have the following proposition .
[ adding - leaf ] let @xmath89 , where @xmath90 , and @xmath91 . also , let @xmath78 be the graph obtained from @xmath19 by adding an edge @xmath5 , where @xmath92
. then @xmath93 .
suppose that @xmath19 is @xmath7-represented by a word @xmath94 , where for @xmath95 , @xmath96 is a word not containing @xmath3 .
then it is not difficult to check that the word @xmath97 @xmath7-represents @xmath78 ( in particular , the vertex @xmath3 is the only neighbor of @xmath4 ) . finally ,
if @xmath78 could be @xmath59-represented by some word , we would remove from that word the letter @xmath4 to obtain a @xmath59-representation of @xmath19 , which is impossible .
so , @xmath93 .
a _ prism _ @xmath98 is a graph consisting of two cycles @xmath99 and @xmath100 , where @xmath101 , connected by the edges @xmath102 for @xmath103 . in particular , the @xmath26-dimensional cube is a prism . examples of prisms are given in figure [ prisms - pic ] .
the leftmost prism there is called the _ triangular prism_.
the middle prism is the @xmath26-dimensional cube .
[ every - prism-3-repr](@xcite ) every prism @xmath98 is @xmath26-word - representable .
[ pr3-thm](@xcite ) the triangular prism @xmath104 is not @xmath11-word - representable , and thus , by theorem [ every - prism-3-repr ] , @xmath105 .
our next theorem extends theorem [ pr3-thm ] by showing that _ any _ prism belongs to @xmath9 .
[ prn - repr - num-3-thm ] for @xmath106 , @xmath98 is not @xmath11-word - representable , and thus , by theorem [ every - prism-3-repr ] , @xmath107 .
suppose that @xmath98 can be @xmath11-represented by a word @xmath1 for @xmath106 .
it is not difficult to see that there must exist a letter @xmath3 in @xmath1 such that no letter occurs twice between the two copies of @xmath3 . by proposition
[ just - useful - observation ] , there are exactly three letters between the copies of @xmath3 . using symmetry , proposition [ cyclic - shift ] and proposition [ reverse - representant ] , we only need to consider two cases ( the second one is unnecessary in the case of @xmath108 because of symmetry ) where we took into account that the vertices @xmath109 , @xmath11 and @xmath52 form an independent set : * @xmath1 is of the form @xmath110 . since @xmath111 and @xmath112 , we can refine the structure of @xmath1 as follows @xmath113 however , @xmath11 and @xmath114 alternate in @xmath1 contradicting to the fact that @xmath115 .
* @xmath1 is of the form @xmath116 . in this case
, we will refine the structure of @xmath1 in two different ways and then will merge these refinements : * * since @xmath117 , @xmath118 and @xmath119 , @xmath1 must be of the form @xmath120 * * since @xmath111 , @xmath112 and @xmath115 , @xmath1 must be of the form @xmath121 + merging the refinements , we see that @xmath1 must be of the form @xmath122 however , we see that the letters @xmath123 and @xmath114 alternate in @xmath1 contradicting to the fact that @xmath124 . applying proposition [ adding - leaf ] as many times as necessary , we see that if @xmath125 then a graph obtained from @xmath19 by attaching simple paths of any lengths to vertices in @xmath19 belongs to @xmath9 ; call this way to extend graphs `` operation 1 '' . also , by theorem [ add - path - to - graph ] , we can add simple paths of length at least @xmath26 connecting any pair of vertices in @xmath125 and still obtain a graph in @xmath9 ( if a @xmath11-word - representable graph would be obtained , we would have a contradiction with @xmath125 ) ; call this way to extend graphs `` operation 2 '' .
the only known to us for the moment graphs in @xmath9 are recorded in the following statement whose truth follows from considerations above .
[ known - r3 ] @xmath9 contains the petersen graph , prisms , and any other graph obtained from these by applying operations @xmath25 and @xmath11 arbitrary number of times in any order .
the operations of connecting two graphs , @xmath17 and @xmath18 , by an edge and gluing these graphs in a vertex are presented schematically in figure [ glueing - connecting ] .
it follows directly from theorem [ thm : rep - equals - semi - trans ] that if both @xmath17 and @xmath18 are word - representable then the resulting graphs will be word - representable too , while if at least one of @xmath17 or @xmath18 is non - word - representable then the resulting graphs will be non - word - representable . indeed ,
if @xmath17 and @xmath18 are oriented semi - transitively , then orienting the edge @xmath5 in either direction will not give a chance for the resulting graph to have a shortcut ( defined in subsection [ semi - transitive - orient - subsec ] ) thus resulting in a semi - transitively oriented graph ; similarly , no shortcut is possible when semi - transitively oriented @xmath17 and @xmath18 are glued in a vertex @xmath126 . while the arguments above involving orientations answer the question on word - representability of connecting graphs by an edge or gluing graphs in a vertex
, they do not allow to answer the following question : if @xmath17 is @xmath20-word - representable , @xmath18 is @xmath21-word - representable , and @xmath19 is @xmath7-word - representable ( such a @xmath7 must exist by theorem [ repr - must - be - k - repr ] ) then what can be said about @xmath7 ?
theorem [ connect - glue - thm ] below , that is based on theorems [ connect - by - edge ] and [ glue - in - vertex ] , answers this question . [ connect - by - edge](@xcite ) for @xmath90 , let @xmath36 and @xmath37 be @xmath7-uniform words representing graphs @xmath127 and @xmath128 , respectively , where @xmath129 and @xmath130 are disjoint .
suppose that @xmath131 and @xmath132 .
let @xmath133 be the graph @xmath134 .
then @xmath133 is @xmath7-word - representable .
[ glue - in - vertex](@xcite ) for @xmath90 , let @xmath36 and @xmath37 be @xmath7-uniform words representing graphs @xmath127 and @xmath128 , respectively , where @xmath129 and @xmath130 are disjoint .
suppose that @xmath131 and @xmath132 .
let @xmath135 be the graph obtained from @xmath17 and @xmath18 by identifying @xmath3 and @xmath4 into a new vertex @xmath126 .
then @xmath135 is @xmath7-word - representable .
[ connect - glue - thm ] suppose that for graphs @xmath127 and @xmath128 ,
@xmath136 and @xmath137 , @xmath131 , @xmath132 and @xmath138 .
also , let the graph @xmath78 be obtained by connecting @xmath17 and @xmath18 by the edge @xmath5 , and the graph @xmath139 be obtained from @xmath17 and @xmath18 by identifying the vertices @xmath3 and @xmath4 into a single vertex @xmath126 .
the following holds . 1 .
if @xmath140 then both @xmath78 and @xmath139 are cliques and thus @xmath141 . in this case , @xmath142 .
2 . if @xmath143 but @xmath144 then @xmath145 and @xmath146 .
3 . if @xmath147 then @xmath148 .
the first part of the statement is easy to see since both a single vertex ( @xmath139 ) and the one edge graph ( @xmath78 ) are @xmath25-word - representable by theorem [ 1-repr - characterization ] . for part 2 , without loss of generality , @xmath149 ( that is , @xmath150 ) and
thus @xmath151 leading to @xmath145 . on the other hand , @xmath78 is not a clique and thus @xmath152 . if @xmath153 then @xmath18 can be represented by the permutation @xmath154 for @xmath155 and thus @xmath78 can be represented by @xmath156 leading to @xmath157
. however , if @xmath158 , so that @xmath159 , we can take any @xmath7-word - representation of @xmath18 and replace in it every other occurrence of the letter @xmath4 by @xmath160 to obtain a @xmath7-word - representation of @xmath78 .
thus , @xmath161 because if it would be less than @xmath7 , we would have @xmath162 by proposition [ useful - remark - ind - subgraph ] , a contradiction .
for part 3 , neither @xmath78 nor @xmath139 is a clique and thus @xmath163 . by proposition [ k - implies - k - plus-1 ] ,
both @xmath17 and @xmath18 are @xmath7-word - representable .
if @xmath90 then by theorems [ connect - by - edge ] and [ glue - in - vertex ] both @xmath78 and @xmath139 are @xmath7-word - representable leading to @xmath164 since if @xmath165 or @xmath166 we would obtain a contradiction either with @xmath136 or with @xmath137 by proposition [ useful - remark - ind - subgraph ] . finally ,
if @xmath167 then @xmath17 and @xmath18 must be cliques that can be represented by permutations @xmath168 and @xmath154 , respectively , for @xmath169 and @xmath155 .
then the words @xmath170 and @xmath171 2-word - represent the graphs @xmath78 and @xmath139 , respectively , and thus @xmath172 .
a subset @xmath39 of the set of vetrices @xmath2 of a graph @xmath19 is a _ module _ if all members of @xmath39 have the same set of neighbors among vertices not in @xmath39 ( that is , among vertices in @xmath173 ) .
for example , figure [ modules - business ] shows replacing the vertex @xmath25 in the triangular prism by the module @xmath174 formed by the vertices @xmath175 , @xmath176 and @xmath10 .
the following theorem is an extended version of an observation made in @xcite .
[ module - thm ] suppose that @xmath19 is a word - representable graph and @xmath91 .
let @xmath78 be obtained from @xmath19 by replacing @xmath3 with a module @xmath32 , where @xmath32 is any comparability graph ( in particular , any clique ) .
then @xmath78 is also word - representable .
moreover , if @xmath177 and @xmath178 then @xmath161 , where @xmath179 . by theorem
[ comparability - graphs ] , @xmath32 can be represented by a word @xmath180 , where @xmath49 is a permutation ( of length equal to the number of vertices in @xmath32 ) for @xmath181 .
if @xmath182 , we can adjoin to the representation of @xmath32 any number of copies of permutations in the set @xmath183 to obtain a @xmath7-representation @xmath184 of @xmath32 , where @xmath185 ( indeed , no alternation properties will be changed while adjoining such extra permutations ) . using proposition [ k - implies - k - plus-1 ] , if necessary ( in case when @xmath186 ) , we can assume that @xmath19 can be @xmath7-represented by a word @xmath187 for some words @xmath96 not containing @xmath3 for all @xmath95 .
but then the word @xmath188 @xmath7-represents
@xmath78 . indeed , it is easy to see that the pairs of letters in @xmath189 from @xmath190 have the right alternation properties , as do the pairs of letters from @xmath191 in @xmath189 . on the other hand ,
it is not difficult to see that if @xmath192 and @xmath193 then @xmath194 if and only if @xmath195 . if @xmath78 would be @xmath59-word - representable
, we would either obtain a contradiction with @xmath177 ( after replacing each @xmath49 in @xmath189 with @xmath3 ) or with @xmath178 ( after removing all letters in @xmath189 that are not in @xmath32 ) .
thus , @xmath161 .
theorems [ 1-repr - characterization ] and [ 2-repr - characterization ] show that there exists no constant @xmath10 such that all @xmath25- or @xmath11-word - representable graphs are @xmath10-colorable .
indeed , @xmath51 is @xmath52-colorable , while circle graphs formed by @xmath196 and an isolated vertex ( which are not @xmath25-word - representable ) are @xmath197-colorable . on the other hand , known to us graphs @xmath83 ( considered in theorem [ thm : example ] )
that require many copies of each letter to be represented , are @xmath26-colorable for any @xmath90 ( because @xmath83 is a bipartite graph with an all - adjacent vertex ) .
thus , @xmath26-word - representable graphs do not contain a class of @xmath10-colorable graphs for some constant @xmath198 ( this claim follows from the fact that @xmath83 , being @xmath26-colorable , is not 3-word - representable for @xmath72 ) .
a natural question to ask here is : is @xmath9 properly included in a class of @xmath10-colorable graphs for a constant @xmath10 ?
the following theorem shows that this is not the case .
[ thm - r3-c - colorable ] the class @xmath9 is not included in a class of @xmath10-colorable graphs for some constant @xmath10 .
suppose that any graph in class @xmath9 is @xmath10-colorable for some constant @xmath10 .
we can assume that @xmath198 since @xmath84 being @xmath26-colorable belongs to @xmath9 by theorem [ thm : example ] .
consider the triangular prism @xmath199 to the left in figure [ modules - business ] , which is also @xmath26-colorable and , by theorem [ pr3-thm ] , belongs to @xmath9 .
replace the vertex @xmath25 in @xmath199 with a module @xmath200 , the complete graph on @xmath201 vertices ( see section [ modules - subsec ] for the notion of a module ) as shown for the case of @xmath202 in figure [ modules - business ] .
denote the obtained graph by @xmath203 . since @xmath204 , by theorem [ module - thm ] , @xmath205 .
however , @xmath203 is not @xmath10-colorable since it contains a clique of size @xmath201 ( @xmath203 is @xmath206-colorable ) .
we obtain a contradiction with our assumption .
we note that an alternative way to obtain a contradiction in the proof of theorem [ thm - r3-c - colorable ] is to consider the triangular prism @xmath199 ( which is in @xmath9 ) , the complete graph @xmath200 ( which is in @xmath15 ) , and either to connect these graphs by an edge , or glue these graphs in a vertex .
then by theorem [ connect - glue - thm ] , the obtained graph will be in @xmath9 , but it is @xmath206-colorable . from considerations in this section
it follows that , in particular , the classes @xmath9 and @xmath26-colorable graphs ( that are word - representable by theorem [ obs:3col ] ) are not comparable in the sense that none of these classes is included in the other one . however , this section does not answer the following question .
[ problem0 ] can each bipartite ( that is , @xmath11-colorable ) graph be @xmath26-word - represented ?
namely , does each bipartite graph belong to the union of the sets @xmath207 ?
we suspect the answer to the question in problem [ problem0 ] to be negative . a good candidate for a counterexample should be the crown graph @xmath81 for @xmath208 ;
see section [ crown - graphs - sec ] and , in particular , problem [ problem2 ] .
the _ ladder graph _ @xmath209 with @xmath210 vertices and @xmath211 edges is presented in figure [ ladder - pic ] .
it follows from the proof of theorem [ every - prism-3-repr ] in @xcite that @xmath209 is 3-word - representable . in the following theorem
, we will show that @xmath209 is actually 2-word - representable for @xmath212 ( it is clearly @xmath25-word - representable for @xmath213 ) .
we prove the statement by induction on @xmath52 .
@xmath215 can be 2-represented by the word @xmath216 that has the factor @xmath217 .
substituting the factor @xmath217 in @xmath36 by @xmath218 and reversing the entire word , we obtain the word @xmath219 that contains the factor @xmath220 .
it is straightforward to check that @xmath37 represents @xmath221 since one only needs to check the alternation properties of the just added letters @xmath11 and @xmath123 . more generally , given a @xmath11-representation @xmath96 of @xmath222 containing the factor @xmath223
, we substitute @xmath223 in @xmath96 by @xmath224 and reverse the entire word to obtain the word @xmath225 containing the factor @xmath226 .
it is straightforward to check that @xmath225 represents @xmath227 since the only thing that needs to be checked is the right alternation properties of the just added letters @xmath228 and @xmath229 .
we are done .
consider the cycle graph @xmath232 on five edges .
@xmath232 is in @xmath16 ( it is 2-word - representable , but not 1-word - representable ) , but it is not a comparability graph and thus , by theorem [ comparability - graphs ] , @xmath232 is not permutationally representable . on the other hand , the crown graphs @xmath233 and @xmath234 , being @xmath11-comparability graphs , belong to @xmath16 .
also , @xmath85 , being a @xmath26-comparability graph , belongs to @xmath16 , since @xmath85 is the cycle graph @xmath235 .
moreover , @xmath236 , being a 4-comparability graph , belongs to @xmath237 , which follows from the fact that @xmath236 is the prism @xmath238 ( the @xmath26-dimensional cube ) and theorem [ prn - repr - num-3-thm ] can be applied .
crown graphs , being bipartite graphs , and thus comparability graphs , provide an interesting case study of relations between @xmath7-comparability graphs and @xmath7-word - representable graphs . while each @xmath7-comparability graph is necessarily @xmath7-word - representable , in some cases such
a graph is also @xmath59-word - representable , and , in fact , it is in @xmath239 in the known to us situations .
thus , it seems like giving up permutational representability , we should be able to come up with a shorter representation of a given comparability graph . however , we do not know whether this is essentially always the case ( except for some particular cases like the graphs @xmath233 and @xmath234 ) .
thus we state the following open problem .
[ problem1 ] characterize @xmath7-comparability graphs that belong to @xmath240 for a fixed @xmath42 .
in particular , characterize those @xmath7-comparability graphs that belong to @xmath239 . is the set of @xmath7-comparability graphs that belong to @xmath240 ( non-)empty for a fixed @xmath241 ? [ problem2 ] characterize those @xmath81 that belong to @xmath240 for a fixed @xmath42 .
in particular , characterize those @xmath81 that belong to @xmath239 . is the set of crown graphs @xmath81 that belong to @xmath240 ( non-)empty for a fixed @xmath241 ? we suspect that each @xmath81 belongs to @xmath239 for @xmath242 .
even if we would manage to prove this statement for a single @xmath208 , we would answer at once ( negatively ) the question in problem [ problem0 ] .
a step towards solving problem [ problem2 ] can be the following observation : take the concatenation of the permutations representing @xmath81 discussed above and remove the leftmost @xmath243 letters @xmath244 and the rightmost @xmath243 letters @xmath245 .
the remaining word will still represent @xmath81 , which is not difficult to see .
one would then need to remove from the word two more letters , one copy of @xmath7 and one copy of @xmath109 , then possibly do some rearrangements of the remaining letters with a hope to obtain a word that would @xmath59-word - represent @xmath81 .
for example , for @xmath246 , we begin with the word representing @xmath85 in table [ ex - k - repr - hkk ] , and remove its first two , and its last two letters to obtain the following representation of @xmath85 : @xmath247 now , remove the leftmost 3 and the leftmost @xmath109 , make the last letter 1 to be the first letter 1 ( that is , make the cyclic one - position shift to the right ) , and , finally , replace the six rightmost letters in the obtained word by the word obtained by listing these letters in the reverse order to obtain a @xmath11-word - representation of @xmath85 ( this is the same representation as one would obtain for the cycle @xmath235 if to follow the strategy described in @xcite ) : @xmath248 .
unfortunately , similar steps do not work for @xmath26-word - representing @xmath236 , and it is not so clear which copies of unwanted letters 4 and @xmath109 one should remove before conducting further rearrangements . in either case , even if one could demonstrate how to obtain a @xmath59-word - representation of @xmath81 for some @xmath208 , it is not so clear what generic argument would show that no @xmath249-word - representation exists , if this would be the case .
m. halldrsson , s. kitaev , a. pyatkin : graphs capturing alternations in words , _ lecture notes in computer science _ * 6224 * ( 2010 ) 436437 .
_ proc . _
14_th conf . on developments in language theory _ , _ dlt _ 2010 , london , ontario , canada , august 1720 , 2010 .
m. halldrsson , s. kitaev , a. pyatkin : alternation graphs , _ lecture notes in computer science _ * 6986 * ( 2011 ) 191202 .
proceedings of the _ _ 37_th international workshop on graph - theoretic concepts in computer science _ , _ wg _ 2011 , tepl monastery , czech republic , june 2124 , 2011 . | a graph @xmath0 is word - representable if there exists a word @xmath1 over the alphabet @xmath2 such that letters @xmath3 and @xmath4 alternate in @xmath1 if and only if @xmath5 is an edge in @xmath6 .
a graph is word - representable if and only if it is @xmath7-word - representable for some @xmath7 , that is , if there exists a word containing @xmath7 copies of each letter that represents the graph .
also , being @xmath7-word - representable implies being @xmath8-word - representable .
the minimum @xmath7 such that a word - representable graph is @xmath7-word - representable , is called graph s representation number . graphs with representation number 1 are complete graphs , while graphs with representation number 2 are circle graphs . the only fact known before
this paper on the class of graphs with representation number 3 , denoted by @xmath9 , is that the petersen graph and triangular prism belong to this class . in this paper , we show that any prism belongs to @xmath9 , and that two particular operations of extending graphs preserve the property of being in @xmath9 .
further , we show that @xmath9 is not included in a class of @xmath10-colorable graphs for a constant @xmath10 . to this end , we extend three known results related to operations on graphs . we also show that ladder graphs used in the study of prisms are @xmath11-word - representable , and thus each ladder graph is a circle graph .
finally , we discuss @xmath7-word - representing comparability graphs via consideration of crown graphs , where we state some problems for further research . + * keywords : * word - representable graph , representation number , prism , ladder graph , circle graph , crown graph , comparability graph |
the phenomenology of compact stars is intimately connected to the eos of matter at densities well beyond the nuclear saturation density , @xmath4 @xmath5 .
compact stars are therefore natural laboratories for the exploration of baryonic matter under extreme conditions , complementary to those created in terrestrial experiments with atomic nuclei and heavy - ion collisions .
recent results derived from observations of compact stars provide serious constraints on the nuclear eos , see @xcite and references therein .
a stiff eos at high density is needed to explain the high compact - star masses , @xmath6 m@xmath3 , reported for some low - mass x - ray binaries ( lmxbs ) , _
e.g. _ , 4u 1636 - 536 @xcite , and the large radius , @xmath7 km , of the isolated neutron star rx j1856.5 - 3754 ( shorthand : rx j1856 ) @xcite .
another example is exo 0748 - 676 , an lmxb for which the compact - star mass _ and _ radius have been constrained to @xmath8 m@xmath3 and @xmath9 km @xcite .
however , the status of the results for the latter object is unclear , because the gravitational redshift @xmath10 observed in the x - ray burst spectra @xcite has not been confirmed , despite numerous attempts .
while compact - star phenomenology apparently points towards a stiff eos at high density , heavy - ion collision data for kaon production @xcite and elliptic flow @xcite set an upper limit on the stiffness of the eos @xcite .
a key question regarding the structure of matter at high density is whether a phase transition to quark matter occurs inside compact stars , and whether it is accompanied by unambiguous observable signatures .
it has been argued that the observation of a compact star with high mass and large radius , likewise reported for exo 0748 - 676 , would be incompatible with a quark core @xcite , because the softening of the eos due to quark deconfinement lowers the maximum mass and the radius in comparison to the nuclear matter case .
however , alford _
_ @xcite have demonstrated with a few counter examples that quark matter can not be excluded by this argument .
in particular , for a recently developed hybrid star eos @xcite , based on the dbhf approach in the nuclear sector and a three - flavor chiral quark model @xcite , stable hybrid stars in the mass range from 1.2 m@xmath3 up to 2.1 m@xmath3 have been obtained , in accordance with modern mass - radius constraints , see also @xcite . in that model , a sufficiently low critical density for quark deconfinement was achieved with a strong diquark coupling , while a sufficient stiffness for a high maximum mass of the compact star sequence was obtained with a repulsive vector meanfield in the quark matter sector .
the corresponding hybrid eos for symmetric matter was shown to fulfill the constraints from elliptic flow data in heavy - ion collisions . in the present work
we want to discuss a new scenario of quark deconfinement , which could play an important role in asymmetric matter , in particular for the phenomenology of compact stars .
chiral quark models of the njl type with dynamical chiral symmetry breaking have the property that the restoration of this symmetry ( and the related quark deconfinement ) at zero temperature is flavor specific .
when solving the gap and charge - neutrality equations selfconsistently one finds that the chiral symmetry restoration for a given flavor occurs when the chemical potential of that flavor reaches a critical value that is approximately equal to the dynamically generated quark mass , @xmath11 , where @xmath12 .
in asymmetric matter the quark chemical potentials are different .
consequently , the njl model behavior suggests that the critical density of deconfinement is flavor dependent , see fig .
[ f : phases ] . [ cols= " < ,
< " , ] one of the main arguments for strange matter is the fact that superconducting phases , like the cfl phase , can suppress fast neutrino emission processes of all quark flavors and are able to fulfill the fusion ignition condition .
however , as we have shown above , in the single flavor csl - phase the fast direct urca process is not possible at all , whereas slow neutrino cooling processes like bremsstrahlung of electrons and d - quarks occur . as one can see from fig .
[ f : profile - release ] the energy release from the conversion of dbhf nuclear matter to dbhf - csl hybrid matter gives 12 ( 8) mev for a compact star core with mass 1.4 ( 2.0 ) @xmath13 , which is in the range 1 100 mev and could , in principle , explain burst ignition at appropriate depths for a suitable value of @xmath14 .
therefore , the compact star does not necessarily need to be made of strange matter , but could be a hybrid star with quark matter in the d - csl phase and a thin crust .
_ @xcite show that deep crustal heating mechanisms at the crust - core boundary like the conversion of baryonic matter to strange quark matter , which can fulfill the constraints of the superburst scenario do provide a consistent explanation of the cooling of soft x - ray transients too .
along the lines of this argument we claim that the d - quark drip effect at the crust core boundary , which leads to a mixture of nuclear matter with single - flavor quark matter in the csl phase can serve as a deep crustal heating mechanism @xcite .
superbursts and the cooling of x - ray transients are not only consistent with quark matter in compact stars but may qualify as a signature !
in this contribution we have suggested a new quark - nuclear hybrid eos for compact star applications that fulfills modern observational constraints from compact stars . due to isospin asymmetry
, down - quarks may `` drip out '' from nucleons and form a single - flavor color superconducting ( csl ) phase that is mixed with nuclear matter already at the crust - core boundary in compact stars .
the csl phase has interesting cooling and transport properties that are in accordance with constraints from the thermal and rotational evolution of compact stars .
it remains to be investigated whether this new compact star composition could lead to unambiguous observational consequences .
we conjecture that the d - quark drip may serve as an effective deep crustal heating mechanism for the explanation of the puzzling superburst phenomenon and the cooling of x - ray transients .
we thank the organizers for bringing together colleagues from different communities and for creating the stimulating atmosphere at the workshop in spa , belgium .
d.b . is supported in part by the polish ministry of science and higher education , grant no .
n n 202 0953 33 ; t.k .
is grateful for partial support from the department of energy , office of nuclear physics , contract no .
de - ac02 - 06ch11357 .
the work of f.s . was supported by the belgian fund for scientific research ( fnrs ) .
d.b . and f.s .
acknowledge the european science foundation for supporting their participation at the workshop of the research networking programme `` the physics of compact stars ( compstar ) '' in ladek zdrj , poland , where part of this work has been done .
j. e. trmper , v. burwitz , f. haberl and v. e. zavlin , _ nucl .
_ * 132 * , 560 ( 2004 ) .
f. zel , _ nature _ * 441 * , 1115 ( 2006 ) .
j. cottam , f. paerels and m. mendez , _ nature _ * 420 * , 51 ( 2002 ) .
d. blaschke , s. fredriksson , h. grigorian , a. m. ztas and f. sandin , _ phys .
d _ * 72 * , 065020 ( 2005 ) .
d. b. blaschke , d. gomez dumm , a. g. grunfeld , t. klhn and n. n. scoccola , _ phys .
c _ * 75 * , 065804 ( 2007 ) . c. j. horowitz , e. j. moniz and j. w. negele , _ phys .
d _ * 31 * , 1689 ( 1985 ) .
g. rpke , d. blaschke and h. schulz " , _ phys .
d _ * 34 * , 3499 ( 1986 ) .
e. n. e. van dalen , c. fuchs and a. faessler , _ nucl .
a _ * 744 * , 227 ( 2004 ) ; _ phys .
c _ * 72 * , 065803 ( 2005 )
. e. n. e. van dalen , c. fuchs and a. faessler , _ phys .
lett . _ * 95 * , 022302 ( 2005 ) .
s. b. rster , v. werth , m. buballa , i. a. shovkovy and d. h. rischke , _ phys .
d _ * 72 * , 034004 ( 2005 ) .
h. abuki and t. kunihiro , _ nucl .
phys . a _ * 768 * , 118 ( 2006 ) . h. j. warringa , d. boer and j. o. andersen , _ phys . rev . d _ * 72 * , 014015 ( 2005 ) .
h. grigorian , _ phys .
_ * 4 * , 382 ( 2007 ) .
m. buballa , _ phys .
_ * 407 * , 205 ( 2005 ) .
t. schfer , _ phys .
d _ * 62 * ( 2000 ) 094007 .
m. g. alford , j. a. bowers , j. m. cheyne and g. a. cowan , _ phys .
d _ * 67 * ( 2003 ) 054018 .
a. schmitt , _ phys .
d _ * 71 * ( 2005 ) 054016 .
d. n. aguilera , d. blaschke , m. buballa , and v. l. yudichev , _ phys .
d _ * 72 * , 034008 ( 2005 ) .
d. n. aguilera , d. blaschke , h. grigorian and n. n. scoccola , _ phys .
d _ * 74 * , 114005 ( 2006 ) .
f. marhauser , d. nickel , m. buballa and j. wambach , _ phys .
d _ * 75 * , 054022 ( 2007 ) .
m. alford , m. braby , m. w. paris and s. reddy , _ astrophys .
j. _ * 629 * , 969 ( 2005 ) .
d. blaschke and h. grigorian , _ prog .
_ * 59 * , 139 ( 2007 ) .
s. popov , h. grigorian , r. turolla and d. blaschke , _ astron .
astrophys . _ * 448 * , 327 ( 2006 ) .
s. popov , h. grigorian and d. blaschke , _ phys .
c _ * 74 * , 025803 ( 2006 ) .
h. grigorian , d. blaschke and t. klhn , _ neutron stars and pulsars _ , becker w and huang h h ( eds . ) , _ mpe report _ * 291 * , 193 ( 2006 )
. j. madsen , _ phys .
_ * 85 * , 10 ( 1999 ) .
a. drago , g. pagliara and i. parenti , _ astrophys .
j. _ * 678 * , l117 ( 2008 ) .
m. stejner and j. madsen , _ astron .
astrophys . _ * 458 * , 523 ( 2006 ) . d. page and a. cumming , _ astrophys .
j. _ * 635 * , l157 ( 2005 ) .
d. blaschke , t. klhn and d. n. voskresensky , _ astrophys .
j. _ * 533 * , 406 ( 2000 ) .
d. page , m. prakash , j. m. lattimer and a. steiner , _ phys .
lett . _ * 85 * , 2048 ( 2000 ) .
d. blaschke , h. grigorian and d. n. voskresensky , _ astron .
_ * 368 * , 561 ( 2001 ) .
a. cumming and l. bildsten , _ astrophys .
j _ * 559 * , l127 ( 2001 ) .
g. ushomirsky and r. e. rutledge , _ mon . not .
_ * 325 * , 1157 ( 2001 ) .
s. shternin , d. g. yakovlev , p. haensel and a. y. potekhin , _ mon . not .
r. astron .
soc . _ * 382 * , l43 ( 2007 ) .
d. blaschke , f. sandin and t. klahn , arxiv:0808.0181 [ nucl - th ] .
d. blaschke , f. sandin , t. klahn and j. berdermann , arxiv:0807.0414 [ nucl - th ] . | we suggest a scenario where the three light quark flavors are sequentially deconfined under increasing pressure in cold asymmetric nuclear matter as , _ e.g. _ , in neutron stars .
the basis for our analysis is a chiral quark matter model of nambu jona - lasinio ( njl ) type with diquark pairing in the spin-1 single flavor ( csl ) , spin-0 two flavor ( 2sc ) and three flavor ( cfl ) channels .
we find that nucleon dissociation sets in at about the saturation density , @xmath0 , when the down - quark fermi sea is populated ( d - quark dripline ) due to the flavor asymmetry induced by @xmath1-equilibrium and charge neutrality . at about @xmath2 u - quarks appear and a two - flavor color superconducting ( 2sc ) phase is formed .
the s - quark fermi sea is populated only at still higher baryon density , when the quark chemical potential is of the order of the dynamically generated strange quark mass .
we construct two different hybrid equations of state ( eos ) using the dirac - brueckner hartree - fock ( dbhf ) approach and the eos by shen _
et al .
_ in the nuclear matter sector .
the corresponding hybrid star sequences have maximum masses of , respectively , 2.1 and 2.0 m@xmath3 .
two- and three - flavor quark - matter phases exist only in gravitationally unstable hybrid star solutions in the dbhf case , while the shen - based eos produce stable configurations with a 2sc phase component in the core of massive stars .
nucleon dissociation due to d - quark drip at the crust - core boundary fulfills basic criteria for a deep crustal heating process which is required to explain superbusts as well as cooling of x - ray transients .
address = instytut fizyki teoretycznej , uniwersytet wrocawski , 50 - 204 wrocaw , poland , altaddress = bogoliubov laboratory for theoretical physics , jinr , 141980 dubna , russia address = ifpa , dpartement ago , universit de lige , 4000 lige , belgium , altaddress = division of physics , lule university of technology , se-97187 lule , sweden address = theory division , argonne national laboratory , argonne il , usa address = desy zeuthen , platanenallee 6 , d-15738 berlin , germany + e - mail : [email protected] , [email protected] , [email protected] , [email protected] |
the primary concern of this paper is the erasure channel , which is a common digital communication channel model that plays a fundamental role in coding and information theory . throughout the paper
, we assume that time is discrete and indexed by the integers . at time @xmath1
, the erasure channel of interest can be described by the following equation : @xmath2 where the channel input @xmath3 , supported on an irreducible finite - type constraint @xmath4 , is a stationary process taking values from the input alphabet @xmath5 , and the erasure process @xmath6 , independent of @xmath3 , is a binary stationary and ergodic process with _ erasure rate _ @xmath7 , and @xmath8 is the channel output process over the output alphabet @xmath9 .
the word `` erasure '' as in the name of our channel naturally arises if a `` @xmath10 '' is interpreted as an erasure at the receiving end of the channel ; so , at time @xmath1 , the channel output @xmath11 is nothing but the channel input @xmath12 if @xmath13 , but an erasure if @xmath14
. let @xmath15 denote the set of all the finite length words over @xmath16 .
let @xmath17 be a finite subset of @xmath15 , and let @xmath4 be the _ finite - type constraint _ with respect to @xmath17 , which is a subset of @xmath15 consisting of all the finite length words over @xmath16 , each of which does not contain any element in @xmath17 as a contiguous subsequence ( or , roughly , elements in @xmath17 are `` forbidden '' in @xmath4 ) .
the most well known example is the @xmath18-run - length - limited ( rll ) constraint over the alphabet @xmath19 , which forbids any sequence with fewer than @xmath20 or more than @xmath21 consecutive @xmath22 s in between two successive @xmath23 s ; in particular , a prominent example is the @xmath24-rll constraint , a widely used constraint in magnetic recording and data storage ; see @xcite . for the @xmath25-rll constraint with @xmath26 ,
a forbidden set @xmath17 is @xmath27 when @xmath28 , one can choose @xmath17 to be @xmath29 in particular when @xmath30 , @xmath17 can be chosen to be @xmath31 .
the _ length _ of @xmath17 is defined to be that of the longest words in @xmath17 .
generally speaking , there may be many such @xmath17 s with different lengths that give rise to the same constraint @xmath4 ; the length of the shortest such @xmath17 s minus @xmath22 gives the _ topological order _ of @xmath4 .
for example , the topological order of the @xmath0-rll constraint , whose shortest @xmath17 proves to be @xmath31 , is @xmath22 .
a finite - type constraint @xmath4 is said to be _ irreducible _
if for any @xmath32 , there is a @xmath33 such that @xmath34 . as mentioned before , the input process @xmath35 of our channel ( [ mec ] )
is assumed to be _ supported _ on an irreducible finite - type constraint @xmath4 , namely , @xmath36 , where @xmath37 the capacity of the channel ( [ mec ] ) , denoted by @xmath38 , can be computed by @xmath39 where the supremum is taken over all stationary processes @xmath35 supported on @xmath4 . here , we note that input - constraints @xcite are widely used in various real - life applications such as magnetic and optical recording @xcite and communications over band - limited channels with inter - symbol interference @xcite .
particularly , we will pay special attention in this paper to a binary erasure channel with erasure rate @xmath40 ( bec(@xmath40 ) ) with the input supported on the @xmath0-rll constraint , denoted by @xmath41 throughout the paper .
when there is no constraint imposed on the input process @xmath35 , that is , @xmath42 , it is well known that @xmath43 ; see theorem [ fbnot ] . when @xmath44 , that is , when the channel is perfect with no erasures , @xmath45 proves to be the _ noiseless capacity _ of the constraint @xmath4 , which can be achieved by a unique @xmath46-th order markov chain @xmath47 with @xmath48 @xcite . on the other hand , other than these two above - mentioned `` degenerated '' cases , `` explicit '' analytic formulas of capacity for `` non - degenerated '' cases have remained evasive , and the problem of analytically characterizing the noisy constrained capacity is widely believed to be intractable . the problem of numerically computing the capacity @xmath45 seems to be as challenging : the computation of the capacity of a general channel with memory or input constraints is notoriously difficult and has been open for decades ; and the fact that our erasure channel is only a special class of such ones does not appear to make the problem easier . here , we note that for a discrete memoryless channel , shannon gave a closed - form formula of the capacity in his celebrated paper @xcite , and blahut @xcite and arimoto @xcite , independently proposed an algorithm which can efficiently compute the capacity and the capacity - achieving distribution simultaneously .
however , unlike the discrete memoryless channels , the capacity of a channel with memory or input constraints in general admits no single - letter characterization and very little is known about the efficient computation of the channel capacity . to date , most known results in this regard have been in the forms of numerically computed bounds : for instance , numerically computed lower bounds by arnold and loeliger @xcite , a. kavcic @xcite , pfister , soriaga and siegel @xcite , vontobel and arnold @xcite .
one of the most effective strategies to compute the capacity of channels with memory or input constraints is the so - called _ markov approximation _ scheme .
the idea is that instead of maximizing the mutual information rate over all stationary processes , one can maximize the mutual information rate over all @xmath46-th order markov processes to obtain the @xmath46-th order markov capacity . under suitable assumptions ( see , e.g. , @xcite ) , when @xmath46 tends to infinity , the corresponding sequence of markov capacities will tend to the channel capacity . for our erasure channel , the _
@xmath46-th order markov capacity _ is defined as @xmath49 where the supremum is taken over all @xmath46-th order markov chains supported on @xmath4 .
the main contributions of this work are the characterization of the asymptotics of the above - mentioned input - constrained erasure channel capacity . of great relevance to this work
are results by han and marcus @xcite , jacquet and szpankowski @xcite , which have characterized asymptotics of the capacity of the a binary symmetric channel with crossover probability @xmath40 ( bsc(@xmath40 ) ) with the input supported on the @xmath0-rll constraint .
the approach in the above - mentioned work is to obtain the asymptotics of the mutual information rate first , and then apply some bounding argument to obtain that of the capacity .
the approach in this work roughly follows the same strategy , however , as elaborated below , our approach differs from theirs to a great extent in terms of technical implementations . throughout the paper
, we use the logarithm with base @xmath50 in the proofs and we use the logarithms with base @xmath23 in the numerical computations of the channel capacity . below is a brief account of our results and methodology employed in this work .
the starting point of our approach is lemma [ nformula ] in section [ mutual - information - rate - section ] , a key lemma that expresses the conditional entropy @xmath51 in a form that is particularly effective for analyzing asymptotics of @xmath45 when @xmath40 is close to @xmath10 .
as elaborated in theorem [ wolfsconjecture ] , lemma [ nformula ] naturally gives a lower and upper bound on @xmath45 , where the lower bound gives a counterpart result of wolf s conjecture for a bec(@xmath40 ) .
moreover , when applied to the case when @xmath35 is a markov chain , lemma [ nformula ] yields some explicit series expansion type formulas in theorem [ entropyformula ] and corollary [ memorylessec ] , which aptly pave the way for characterizing the asymptotics of the input - constrained erasure channel capacity . here
we remark that the method in @xcite have been further developed for more general families of memory channels in @xcite via examining the contractiveness of an associated random dynamical system @xcite .
however , the methodology to derive asymptotics of the mutual information rate in this work capitalizes on certain characteristics that are in a sense unique to erasure channels . in section [ general - asymptotics ]
, we consider a memoryless erasure channel with the input supported on an irreducible finite - type constraint , and in theorem [ asyerasurec ] , we derive partial asymptotics of its capacity @xmath45 in the vicinity of @xmath44 where @xmath45 is written as the sum of a constant term , a linear term in @xmath40 and an @xmath52-term .
the lower bound part in the proof of this theorem follows from an easy application of theorem [ entropyformula ] , and the upper bound part hings on an adapted argument in @xcite . in section [ binary - asymptotics ]
, we consider a bec(@xmath40 ) with the input being a first - order markov process supported on the @xmath0-rll constraint @xmath41 .
within this special setup , we show in theorem [ concavitybec ] that the @xmath53 is strictly concave with respect to some parameterization of @xmath35 . and in section [ sub-2 ] , we numerically evaluate @xmath54 and the corresponding capacity - achieving distribution using the randomized algorithm proposed in @xcite which proves to be convergent given the concavity of @xmath53
. moreover , the concavity of @xmath53 guarantees the uniqueness of the capacity achieving distribution , based on which we derive full asymptotics of the above input - constrained bec(@xmath40 ) around @xmath44 in theorem [ foc ] , where @xmath55 is expressed as an infinite sum of all @xmath56-terms . in section [ feedback - section ]
, we turn to the scenarios when there might be feedback in our erasure channel .
we first prove in theorem [ fbnot ] that when there is no input constraint , the feedback does not increase the capacity of the erasure channel even with the presence of the channel memory .
when the input constraint is not trivial , however , we show in theorem [ yonglong - feedback ] that feedback does increase the capacity using the example of a bec(@xmath40 ) with the ( @xmath57)-rll input constraint , and so feedback may increase the capacity of input - constrained erasure channels even if there is no channel memory .
the results obtained in this section suggest the intricacy of the interplay between feedback , memory and input constraints .
in this section , we focus on the mutual information of the erasure channel ( [ mec ] ) introduced in section [ introduction - section ] .
the starting point of our approach is the following key lemma , which is particularly effective for analysis of input - constrained erasure channels .
[ nformula ] for any @xmath58 , we have @xmath59}h(x_{0}|x_{d})p(e_0=1,e_{d}=1,e_{d^c}=0),\ ] ] where @xmath60 \triangleq \{-n,\cdots,-1\}$ ] .
note that @xmath61 where @xmath62 from the independence of @xmath3 and @xmath6 , it follows that @xmath63 here and throughout the paper , let @xmath64 be the set of all finite length words over @xmath65 and we define , for any @xmath66 , @xmath67 and @xmath68 for @xmath69 , @xmath70 where @xmath71 follows from the fact that @xmath72 .
similarly , for @xmath73 , @xmath74 where @xmath71 follows from the fact that @xmath75 .
therefore , @xmath76}\sum_{y_{-n}^0:\mathcal{i}(y_{-n}^{-1})=d , y_{0}=0}p(y_{-n}^{0})\log p(e_0=0|e_{d})=1,e_{d^c}=0)\notag\\ & \stackrel{(a)}{=}-\sum_{d \subseteq [ -n,-1]}p(e_0=0,e_{d}=1,e_{d^c}=0)\log p(e_0=0|e_{d}=1,e_{d^c}=0)\label{0erasure},\end{aligned}\ ] ] where @xmath71 follows from the fact that for any given @xmath77 $ ] , @xmath78 also , we have @xmath79}p(e_0=1,e_{d}=1,e_{d^c}=0)\log p(e_0=1|e_{d}=1,e_{d^c}=0)\label{1erasure},\end{aligned}\ ] ] where @xmath71 follows from a similar argument as in the proof of ( [ 0erasure ] ) and @xmath80 from ( [ pformula ] ) , it then follows that @xmath81}\sum_{y_{-n}^{0 } : \mathcal{i}(y_{-n}^{0})=d\cup\{0\}}p_{x}(y_{d},y_0)p(e_0=1,e_{d}=1,e_{d^c}=0)\log p_{x}(y_{0}|y_{d})\notag\\ & = \sum_{d \subseteq [ -n , -1]}h(x_0|x_{d})p(e_0=1,e_{d}=1,e_{d^c}=0).\label{eerasure}\end{aligned}\ ] ] the desired formula for @xmath51 then follows from ( [ 0erasure ] ) , ( [ 1erasure ] ) and ( [ eerasure ] ) .
one of the immediate applications of lemma [ nformula ] is the following lower and upper bounds on @xmath45 .
[ wolfsconjecture ] @xmath82 for the upper bound , it follows from lemma [ nformula ] that @xmath83}h(x_{0}|x_{d})p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & \stackrel{(a)}{\leq } \lim_{n\to \infty}\sum_{d \subseteq [ -n,-1]}h(x_{0 } ) p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & \leq \lim_{n\to \infty}\sum_{d \subseteq [ -n,-1 ] } p(e_0=1,e_{d}=1,e_{d^c}=0 ) \log k \notag\\ & = p(e_0=1 ) \log k \notag\\ & = ( 1-{\varepsilon } ) \log k , \notag\end{aligned}\ ] ] where we have used the fact that conditioning reduces entropy for @xmath71 .
assume @xmath4 is of topological order @xmath46 , and let @xmath47 be the @xmath46-order markov chain that achieves the noiseless capacity @xmath84 of the constraint @xmath85 .
again , it follows from lemma [ nformula ] that @xmath86}h(\hat{x}_{0}|\hat{x}_{d})p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & \ge \lim_{n\to \infty}\sum_{d \subseteq [ -n , -1]}h(\hat{x}_{0}|\hat{x}_{-m}^{-1},\hat{x}_{d})p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & \stackrel{(a)}{= } \lim_{n\to \infty}\sum_{d \subseteq [ -n , -1]}h(\hat{x}_{0}|\hat{x}_{-m}^{-1})p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & = p(e_0=1 ) h(\hat{x}_{0}|\hat{x}_{-m}^{-1 } ) \notag\\ & = ( 1-{\varepsilon } ) c(\mathcal{s } , 0 ) , \notag\end{aligned}\ ] ] where we have used the fact that @xmath87 is an @xmath46-th order markov chain for @xmath71 .
the upper bound part of theorem [ wolfsconjecture ] also follows from the well - known fact that ( see theorem [ fbnot ] ) @xmath88 and for any @xmath4 , @xmath89 which is obviously true .
let @xmath90 denote the capacity of a bsc(@xmath40 ) with the @xmath18-rll constraint . in @xcite
wolf posed the following conjecture on @xmath91 : @xmath92 where @xmath93 .
a weaker form of this bound has been established in @xcite by counting the possible subcodes satisfying the @xmath18-rll constraint in some linear coding scheme , but the conjecture for the general case still remains open .
it is well known that @xmath94 is the capacity of a bsc(@xmath40 ) without any input constraint , and @xmath95 is the capacity of a bec(@xmath40 ) without any input constraint .
so , for an input - constrained bec(@xmath40 ) , the lower bound part of theorem [ wolfsconjecture ] gives a counterpart result of wolf s conjecture . when applied to the channel with a markovian input , lemma [ nformula ]
gives a relatively explicit series expansion type formula for the mutual information rate of ( [ mec ] ) .
[ entropyformula ] assume @xmath3 is an @xmath46-th order input markov chain .
then , @xmath96 where @xmath97 and @xmath98 : \mbox{for all $ j=1,\cdots , u$ } , \{i_{j},i_{j}+1,\cdots , i_{j}+m\ } \not\subseteq \ { i_1,\cdots , i_{u}\}\}$ ] and @xmath99 , here @xmath100 denotes the remainder of @xmath101 divided by @xmath46 .
note that @xmath102 where @xmath71 follows from the independence of @xmath3 and @xmath103 . from lemma [ nformula ]
, it then follows that @xmath104}h(x_0|x_{d})p(e_0=1,e_{d}=1,e_{d^c}=0).\end{aligned}\ ] ] now , letting @xmath105:|d|=u\ } \mbox { and } b_1(n , u)=b(n , u)-b_{2}(n , u),\ ] ] we deduce that , for @xmath106 @xmath107}p(e_{a(k , i_1^t)}=1,e_d=1,e_{\bar{a}(k , i_1^t)}=0,e_{[-n ,- k - m-1]-d}=0)&=p(e_{a(k , i_1^t)}=1,e_{\bar{a}(k , i_1^t)}=0).\end{aligned}\ ] ] and @xmath108}\left\{h(x_0|x_{a(k , i_1^t)},x_d)\right.\\ & \hspace{4.5mm}\times \left.p(e_{a(k , i_1^t)}=1,e_d=1,e_{\bar{a}(k , i_1^t)}=0,e_{[-n ,- k - m-1]-d}=0)\right\}\\ & \stackrel{(a)}{=}\sum_{k=0}^{n - m+1}\sum_{t=0}^{b(k-1,m)}\sum_{\{i_1^t\}\in b(k-1,t)}\sum_{d \subseteq [ -n ,- k - m-1]}\left\{h(x_0|x_{a(k , i_1^t)})\right.\\ & \hspace{4.5mm}\times\left.p(e_{a(k , i_1^t)}=1,e_d=1,e_{\bar{a}(k , i_1^t)}=0,e_{[-n ,- k - m-1]-d}=0)\right\}\\ & = \sum_{k=0}^{n - m+1}\sum_{t=0}^{b(k-1,m)}\sum_{\{i_1^t\}\in b(k-1,t)}h(x_0|x_{a(k , i_1^t)})p(e_{a(k , i_1^t)}=1,e_{\bar{a}(k , i_1^t)}=0),\end{aligned}\ ] ] where @xmath71 follows from the fact that @xmath3 is an @xmath46-th order markov chain .
then it follows that @xmath109}h(x_0|x_{d})p(e_0=1,e_{d}=1,e_{d^c}=0)\notag\\ & = \sum_{k=0}^{n}\ \sum_{\{i_1,\dots , i_k\}\in b(n , k)}h(x_0|x_{i_1^k})p(e_0=1,e_{i_1^k}=1,e_{\bar{i}_1^k}=0)\notag\\ & = \left(\sum_{k = m}^{n}\ \sum_{\{i_1,\dots , i_k\}\in b_{1}(n , k)}+\sum_{k=0}^{b(n , m)}\sum_{\{i_1,\dots , i_k\}\in b_{2}(n , k)}\right)h(x_0|x_{i_1^k})p(e_0=1,e_{i_1^k}=1,e_{\bar{i}_1^k}=0)\notag\\ & = \sum_{k=0}^{n - m+1}\sum_{t=0}^{b(k-1,m)}\sum_{\{i_1^t\}\in b(k-1,t)}h(x_0|x_{a(k , i_1^t)})p(e_{a(k , i_1^t)}=1,e_{\bar{a}(k , i_1^t)}=0)+t(n),\label{part}\end{aligned}\ ] ] where @xmath110 it follows from @xmath111 that @xmath112 where @xmath113 is the event that `` there is no @xmath46 consecutive @xmath22 s in @xmath114 '' .
now , let @xmath115 for @xmath116 .
then it follows from the assumption that @xmath117 is also a stationary and ergodic process with @xmath118 . using poincare s recurrence theorem @xcite
, we have that @xmath119 , which implies that @xmath120 , where @xmath121 denotes the event that `` there is no @xmath46 consecutive @xmath22 s in @xmath122 '' .
this , together with the fact that @xmath123 , implies that @xmath124 , and therefore the proof of the theorem is complete .
the following corollary can be readily deduced from theorem [ entropyformula ] .
[ memorylessec ] assume that @xmath6 is i.i.d .
and @xmath125 is an @xmath46-th order markov chain .
then @xmath126 where @xmath127 in particular , if @xmath125 is a first - order markov chain , @xmath128 [ verduerasure ] a series expansion type formula for @xmath129 different from ( [ first ] ) is given in theorem @xmath130 of @xcite for a discrete memoryless erasure channel with a first - order input markov chain .
it can be verified that these two formulas are `` equivalent '' in the sense that either one can be deduced from the other one via simple derivations .
the form that our formula takes however makes it particularly effective for the capacity analysis of an input - constrained erasure channel .
in this section , we will focus on the case when @xmath6 is i.i.d .
and @xmath4 is an irreducible finite - type constraint of topological order @xmath46 . with lemma [ nformula ] and corollary [ memorylessec ] established ,
we are ready to characterize the asymptotics of the capacity of this type of erasure channels . as mentioned in section [ introduction - section ] , when @xmath44 , it is well known @xcite that there exists an @xmath46th - order markov chain @xmath47 with @xmath48 such that @xmath131 where the maximization is over all stationary processes supported on @xmath4 .
the following theorem characterizes the asymptotics of @xmath45 near @xmath44 .
[ asyerasurec ] assume that @xmath6 is i.i.d .
then , @xmath132 moreover , for any @xmath133 , @xmath134 is of the same asymptotic form as in ( [ asyerasurec ] ) , namely , @xmath135 to establish ( [ highsnrasy ] ) , we prove that @xmath45 is lower and upper bounded by the same asymptotic form as in ( [ highsnrasy ] ) . for the lower bound part
, we consider the channel ( [ mec ] ) with @xmath47 as its input .
note that @xmath136 and furthermore , for @xmath137 @xmath138 where we have defined @xmath139 it then follows that @xmath140 where @xmath71 follows from @xmath141 and the constant in @xmath52 depends only on @xmath46 and @xmath142 . then , from corollary [ memorylessec ] , it follows that @xmath143 where @xmath144 follows from ( [ tailestimate ] ) and @xmath145 this , together with ( [ parry ] ) , establishes that @xmath45 is lower bounded by the asymptotic form in ( [ highsnrasy ] ) .
for the upper bound part , we will adapt the argument in @xcite .
let @xmath146 and @xmath147 where @xmath148 . in this proof , we define @xmath149 it then follows from lemma [ nformula ] that @xmath150 where @xmath151,|d|=k } h(x_0|x_{d})(1-{\varepsilon})^{k+1}{\varepsilon}^{n - k}.\ ] ] let @xmath152 maximize @xmath153 . as @xmath153 is continuous in @xmath154 and is maximized at @xmath155 when @xmath44 , there exists some @xmath156 ( depends on @xmath1 ) and @xmath157 such that for all @xmath158 , @xmath159 .
then for @xmath160 , there exists some constant @xmath161 ( depends on @xmath1 ) such that @xmath162 from now on , we write @xmath163 letting @xmath164 be the hessian of @xmath165 , we now expand @xmath165 and @xmath166 around @xmath167 to obtain @xmath168 and @xmath169 where @xmath170 contains all @xmath171 as its coordinates . since @xmath172 is negative definite ( see lemma 3.1 @xcite ) , we deduce that , for @xmath173 sufficiently small , @xmath174 without loss of generality , we henceforth assume @xmath172 is a diagonal matrix with all diagonal entries denoted @xmath175 ( since otherwise we can diagonalize @xmath172 ) . now , let @xmath176 and let @xmath177 denote the complement of @xmath178 in @xmath179 : @xmath180 then , we have @xmath181 where @xmath71 follows from the easily verifiable fact that for any @xmath182 , @xmath183 since @xmath184 is an @xmath46-th order markov chain , we deduce that @xmath185 we now ready to deduce that , for some positive constant @xmath186 , @xmath187 which further implies that @xmath188 the proof of ( [ highsnrasy ] ) is then complete . with @xmath45 replaced with @xmath189 , the proof of ( [ highsnrasy ] ) also establishes ( [ highsnrasy-1 ] ) . in a fairly general setting ( where input constraints are not considered ) , a similar asymptotic formula with a constant term , a term linear in @xmath40 and
a residual @xmath190-term has been derived in theorem @xmath191 of @xcite . as an immediate corollary of theorem [ asyerasurec ]
, the following result gives asymptotics of the capacity of a bec(@xmath40 ) with the input supported on the @xmath0-rll constraint @xmath41 .
[ asmpt ] assume @xmath192 and @xmath6 is i.i.d .
then , we have @xmath193 and for any @xmath58 , @xmath194 is of the same asymptotic form , namely , @xmath195 let @xmath196 .
it is well known @xcite that the noiseless capacity @xmath197 and the first - order markov chain @xmath87 with the following transition probability matrix @xmath198 achieves the noiseless capacity , that is , @xmath199 .
furthermore , via straightforward computations , we deduce that @xmath200 which , together with the fact that @xmath201 implies @xmath193 as desired . and the asymptotic form of @xmath194 follows from similar computations .
the asymptotic form in ( [ first - order - partial ] ) only gives partial asymptotics of @xmath194 for any @xmath58 .
for the case @xmath202 , we will derive later on the full asymptotics of @xmath54 ; see ( [ first - order - full ] ) in section [ binary - asymptotics ] .
in this section , we will focus on a bec(@xmath40 ) with the input being a first - order markov process supported on the @xmath0-rll constraint @xmath41 . to be more precise , we assume that @xmath192 , @xmath6 is i.i.d . and
@xmath125 is a first - order markov chain , taking values in @xmath203 and having the following transition probability matrix : @xmath204 in section [ sub-1 ] , we will show that @xmath53 is concave with respect to @xmath205 , and in section [ sub-2 ] , we apply the algorithm in @xcite to numerically evaluate @xmath54 , whose convergence is guaranteed by the above - mentioned concavity result .
finally , in section [ sub-3 ] , we characterize the full asymptotics of @xmath54 around @xmath44 . the concavity of the mutual information rate of special families of finite - state machine channels ( fsmcs ) has been considered in @xcite and @xcite .
the results therein actually imply that the concavity of @xmath53 with respect to some parameterization of the markov chain @xmath35 when @xmath40 is small enough . in this section , however , we will show that @xmath53 is concave with respect to @xmath205 , irrespective of the values of @xmath40 .
below is the main theorem of this section .
[ concavitybec ] for all @xmath206 , @xmath207 is strictly concave with respect to @xmath205 , @xmath208 . from corollary [ memorylessec ] , it follows that to prove the theorem , it suffices to show that for any @xmath58 , @xmath209 is strictly concave with respect to @xmath205 , @xmath208 . to prove this
, we will deal with the following several cases : * * straightforward computations give @xmath210 and @xmath211 one checks that the function within the brace is negative and it takes the maximum at @xmath212 . therefore @xmath213 is strictly concave in @xmath205 . * * by definition , @xmath214 the following facts can be verified easily : * @xmath215 ; * the @xmath1-step transition probability matrix of the markov chain @xmath3 is @xmath216 where @xmath217 * @xmath218 is strictly concave with respect to @xmath219 for @xmath220 . with the above notation
, we have @xmath221 and @xmath222 where @xmath223 and @xmath224 .
it follows from ( i ) and ( iii ) that the term ( [ term1 ] ) is strictly negative .
so , to prove the theorem , it suffices to show that @xmath225 . by the mean value theorem , @xmath226 where @xmath227 lies between @xmath228 and @xmath229 . as a function of @xmath227
, @xmath230 takes the maximum at @xmath228 .
it then follows that @xmath231\}}{(1+\theta)^{4}}\\ & { } & + \frac{c_{n+1}\{4-(-\theta)^{n-1}[(n^2 - 3n)\theta^2 + 2(n^2-n-2)+n^2+n]\}}{(1+\theta)^{4}}\\ & = & \frac{c_{n}[2\theta-2+q(n,\theta)(-\theta)^{n-1}]}{(1+\theta)^{4}}+\frac{c_{n+1}[4-q(n,\theta)(-\theta)^{n-1}]}{(1+\theta)^{4}},\end{aligned}\ ] ] where @xmath232 and @xmath233 we then consider the following several cases : * * we first consider the case that @xmath234 . for this case , obviously we have @xmath235 , @xmath236 and @xmath237 , which further implies that @xmath238}{(1+\theta)^{4}}+\frac{c_{n+1}[4+q(n,\theta)\theta^{n-1}]}{(1+\theta)^{4}}<0.\end{aligned}\ ] ] now , for the case that @xmath239 , again obviously we have @xmath240 and furthermore , @xmath241 where we have used the fact that @xmath242 for a positive even number @xmath1 .
now , we are ready to deduce that @xmath243}{(1+\theta)^{4}}+\frac{c_{n+1}[4+q(n,\theta)\theta^{n-1}]}{(1+\theta)^{4}}\\ & = \frac{(c_{n+1}-c_{n})(4+q(n,\theta)\theta^{n-1})}{(1+\theta)^{4}}+\frac{c_{n}(2\theta+2)}{(1+\theta)^{4}}\\ & < 0.\end{aligned}\ ] ] * * in this case , we have @xmath244}{(1+\theta)^{4}}+\frac{c_{n+1}[4-q(n,\theta)\theta^{n-1}]}{(1+\theta)^{4}}\\ & = \frac{(c_{n+1}-c_{n})[4-q(n,\theta)\theta^{n-1}]}{(1+\theta)^{4}}+\frac{c_{n}(2\theta+2)}{(1+\theta)^{4}}\\ & \le\frac{1}{(1+\theta)^{4}}\left\{\frac{[4-q(n,\theta)\theta^{n-1}]\theta^{n}}{(1-y_{2})y_{2}}+(2\theta+2)(1/g_{n}-2)\right\},\end{aligned}\ ] ] where the last inequality follows from the mean value theorem , the inequality @xmath245 for @xmath246 and the fact that @xmath247 lies between @xmath228 and @xmath229 .
let @xmath248 and @xmath249 , where @xmath250 $ ] . then @xmath251\theta^{n}}{(1-g_{2})g_{2}}+(2\theta+2)(1/g_{n}-2)\right\}\\ & \le&\frac{(2\theta-2 - 4\theta^n)b_{n}c_{n}+(1+\theta)(1+\theta^n)\theta^n(4-q(n,\theta)\theta^{n-1})}{b_{n}c_{n}(1+\theta)^{3}(1+\theta^n)}. \vspace{-3mm}\end{aligned}\ ] ] note that the above numerator , as a function of @xmath252 , takes the maximum at @xmath253
. denote this maximum by @xmath254 , where @xmath255 to complete the proof , it suffices to prove @xmath256 .
substituting @xmath257 into @xmath258 , we have @xmath259\\ & \ge&h(n,\theta),\end{aligned}\ ] ] where @xmath260 the following facts can be easily verified : * @xmath261 takes the minimum at some @xmath262 , where @xmath262 satisfies the following equation @xmath263 * @xmath264 for @xmath265 $ ] and @xmath266
. it then follows from ( a ) and ( b ) that @xmath267 for @xmath268 . solving ( [ theta0 ] ) in @xmath269
, we have @xmath270 for @xmath266 , substituting @xmath271 into @xmath261 , we have @xmath272\\ & \ge\theta_{0}^{n-1 } v(n),\end{aligned}\ ] ] where @xmath273 one checks that @xmath274 for @xmath275 . now , with the fact that @xmath276 for @xmath277 ( this can be verified via tedious yet straightforward computations since @xmath258 is an elementary function ) , we conclude that for @xmath278 for all @xmath279 and @xmath280 $ ] .
when @xmath44 , that is , when the channel is `` perfect '' with no erasures , both @xmath45 and @xmath189 boil down to the noiseless capacity of the constraint @xmath4 , which can be explicitly computed @xcite ; however , little progress has been made for the case when @xmath281 due to the lack of simple and explicit characterization for @xmath38 and @xmath282 . in terms of numerically computing @xmath45 and @xmath189 , relevant work can be found in the subject of fsmcs , as input - constrained memoryless erasure channels can be regarded as special cases of fsmcs .
unfortunately , the capacity of an fsmc is still largely unknown and the fact that our channel is only a special fsmc does not seem to make the problem easier .
recently , vontobel _ et al . _
@xcite proposed a generalized blahut - arimoto algoritm ( gbaa ) to compute the capacity of an fsmc ; and in @xcite , han also proposed a randomized algorithm to compute the capacity of an fsmc . for both algorithms ,
the concavity of the mutual information rate is a desired property for the convergence ( the convergence of the gbaa requires , in addition , the concavity of certain conditional entropy rate ) . on the other hand , as elaborated in @xcite , such a desired property , albeit established for a few special cases @xcite , is not true in general . the concavity established in the previous section allows us to numerically compute @xmath54 using the algorithm in @xcite .
the randomized algorithm proposed in @xcite iteratively compute @xmath283 in the following way : @xmath284,\\ \theta_{n}+a_{n}g_{n^b}(\theta_{n } ) , & \mbox{otherwise } , \end{cases}\ ] ] where @xmath285 is a simulator for @xmath286 ( for details , see @xcite ) .
the author shows that @xmath283 converges to the first - order capacity - achieving distribution if @xmath207 is concave with respect to @xmath205 , which has been proven in theorem [ concavitybec ] .
therefore , with proven convergence , this algorithm can be used to compute the first - order capacity - achieving distribution @xmath287 and the first - order capacity @xmath54 ( in bits ) , which are shown in fig .
[ capacity - achievingdistribution ] and fig .
[ capacity1 ] , respectively . as in section [ sub-1 ] , the noiseless capacity of @xmath24-rll constraint @xmath41
is achieved by the first - order markov chain with the transition probability matrix ( [ achievingmatrix ] ) .
so , we have @xmath288 where @xmath196 . in this section ,
we give a full asymptotic formula for @xmath289 around @xmath44 , which further leads to a full asymptotic formula for @xmath54 around @xmath44 .
the following theorem gives the taylor series of @xmath290 in @xmath40 around @xmath44 , which leads to an explicit formula for the @xmath1-th derivative of @xmath54 at @xmath44 , whose coefficients can be explicitly computed . [ foc ]
a ) @xmath291 is analytic in @xmath40 for @xmath206 and @xmath292 where @xmath293 is taken over all nonnegative intergers @xmath294 satisfying the constraint @xmath295 and @xmath296 \b ) @xmath54 is analytic in @xmath40 for @xmath206 with the following taylor series expansion around @xmath44 : @xmath297 where @xmath298 .
\a ) for @xmath299 , @xmath300 with theorem [ concavitybec ] establishing the concavity of @xmath53 , @xmath290 should be the unique zero point of the derivative of the mutual information rate .
so , @xmath301 and satisfies @xmath302 according to the analytic implicit function theorem @xcite , @xmath291 is analytic in @xmath40 for @xmath303 . in the following the @xmath304-th order derivative of @xmath291 at @xmath44
is computed .
it follows from the leibniz formula and the faa di bruno formula @xcite that @xmath305 which immediately implies a ) .
\b ) note that @xmath306 it then follows from the leibniz formula that @xmath307 therefore , @xmath308 which immediately implies b ) . despite their convoluted looks , ( [ theta - max ] ) and ( [ first - order - full ] ) are explicit and computable .
below , we list the coefficients of @xmath54 ( in bits ) and @xmath291 up to the third order , which are numerically computed according to ( [ theta - max ] ) and ( [ first - order - full ] ) and rounded off to the ten thousandths decimal digit : [ cols="^,^,^,^,^",options="header " , ]
in this section , we consider the input - constrained erasure channel ( [ mec ] ) as in section [ introduction - section ] however with possible feedback , and we are interested in comparing its feedback capacity @xmath309 and its non - feedback capacity @xmath45 .
the following theorem states that for the erasure channel without any input - constraint , feedback does not increase the capacity and both of them can be computed explicitly .
this result is in fact implied by theorem @xmath130 in @xcite , where a random coding argument has been employed in the proof ; we nonetheless give an alternative proof in appendix [ fb - nfb ] for completeness .
[ fbnot ] for the erasure channel ( [ mec ] ) without any input constraints , feedback does not increase the capacity , and we have @xmath310 on the other hand , we will show in the following that feedback may increase the capacity when the input constraint in the erasure channel is non - trivial . as elaborated below , this
is achieved by comparing the asymptotics of the feedback capacity and the non - feedback capacity for a special input - constrained erasure channel . in @xcite ,
sabag _ et al .
_ computed an explicit formula of feedback capacity for bec with @xmath24-rll input constraint @xmath41 .
[ fbc ] @xcite the feedback capacity of the @xmath24-rll input - constrained erasure channel is @xmath311 where the unique maximizer @xmath312 satisfies @xmath313 clearly , the explicit formula in theorem [ fbc ] readily gives the asymptotics of the feedback capacity .
to see this , note that @xmath314 and @xmath315 .
straightforward computations yield @xmath316 hence , @xmath317 it then follows from straightforward computations that for the case when @xmath40 is close to @xmath10 , @xmath318 .
so , we have proven the following theorem : [ yonglong - feedback ] for a bec(@xmath40 ) with the @xmath0-rll input constraint , feedback increases the channel capacity when @xmath40 is small enough .
an independent work in @xcite also found that feedback does increase the capacity of a bec(@xmath40 ) with the same input constraint @xmath41 , by comparing a tighter bound of non - feedback capacity @xmath319 , obtained via a dual capacity approach , with the feedback capacity @xmath320 .
recently , sabag _ et al . _
@xcite also computed an explicit asymptotic formula for the feedback capacity of a bsc(@xmath40 ) with the input supported on the @xmath24-rll constraint . by comparing the asymptotics of the feedback capacity with the that of non - feedback capacity
@xcite , they showed that feedback does increase the channel capacity in the high snr regime .
it is well known that for any memoryless channel without any input constraint , feedback does not increase the channel capacity .
theorem [ fbnot ] states that when there is no input constraint , the feedback does not increase the capacity of the erasure channel even with the presence of the channel memory .
theorem [ yonglong - feedback ] says that feedback may increase the capacity of input - constrained erasure channels even if there is no channel memory .
these two theorems , together with the results in @xcite , suggest the intricacy of the interplay between feedback , memory and input constraints .
we first prove that @xmath321 a similar argument using the independence of @xmath322 and @xmath323 as in the proof of ( [ pformula ] ) yields that @xmath324 it then follows that @xmath325}\sum_{y_1^n:\mathcal{i}(y_1^n)=d}p(e_{d}=1 , e_{d^c}=0)p(x_{d}=y_{d})\log p(e_{d}=1 , e_{d^c}=0)\nonumber\\ & { } \hspace{4.5mm}-\sum_{d \subseteq [ 1 , n]}\sum_{y_1^n:\mathcal{i}(y_1^n)=d}p(e_{d}=1 , e_{d^c}=0)p(x_{d}=y_{d } ) \log p(x_{d}=y_{d})\nonumber\\ & = -\sum_{d \subseteq [ 1 , n]}p(e_{d}=1 , e_{d^c}=0)\log p(e_{d}=1 , e_{d^c}=0)+\sum_{d\subseteq [ 1 , n]}p(e_{d}=1 , e_{d^c}=0)h(x_d)\nonumber\\ & = \sum_{d \subseteq [ 1 , n]}p(e_{d}=1 , e_{d^c}=0)h(x_d)+h(e_1^n)\nonumber\\ & \le h(e_1^n)+\sum_{d \subseteq [ 1 , n]}p(e_{d}=1 ,
e_{d^c}=0)|d| \log k\nonumber\\ & = h(e_1^n)+\ee[e_1+\cdots+e_n ] \log k \label{paomo},\end{aligned}\ ] ] where the only inequality becomes equality if @xmath3 is i.i.d . with the uniform distribution .
it then further follows that @xmath326 \log k\\ & \stackrel{(b)}{= } p(e_1=1 ) \log k\\ & = ( 1-{\varepsilon } ) \log k , \end{aligned}\ ] ] where @xmath71 follows from ( [ paomo ] ) and @xmath144 follows from the ergodicity of @xmath6 . the desired ( [ c - nfb ] ) then follows from the fact that the only inequality @xmath71 becomes equality if @xmath3 is i.i.d . with the uniform distribution
. let @xmath328 , independent of @xmath323 , be the message to be sent and @xmath329 denote the encoding function .
as shown in @xcite , @xmath330 using the chain rule for entropy , we have @xmath331 where ( a ) follows from the fact that @xmath332 is a function of @xmath328 and @xmath333 and @xmath334 if and only if @xmath335 , ( b ) follows from the independence of @xmath328 and @xmath323 . note that for @xmath336 , @xmath337 and for @xmath338 , @xmath339 where @xmath71 follows from the independence of @xmath328 and @xmath323 .
it then follows that @xmath340 where @xmath71 follows from ( [ not0p ] ) and @xmath144 follows from ( [ 0p ] ) . since for any @xmath341 $ ] , @xmath342 which implies that @xmath343 is an @xmath344-dimensional probability mass function
therefore , through a similar argument as before , we have @xmath345}\sum_{y_1^n:\mathcal{i}(y_1^n)=d}p(e_{d}=1 , e_{d^c}=0)q(y_1^n ) \log q(y_1^n)\nonumber\\ & \le h(e_1^n)+\sum_{d \subseteq [ 1 , n]}p(e_{d}=1 , e_{d^c}=0)|d| \log k\nonumber\\ & = h(e_1^n)+\ee[e_1+\cdots+e_n ] \log k , \label{hz}\end{aligned}\ ] ] where the inequality follows from the fact that @xmath346 is an @xmath347-dimensional probability mass function . * acknowledgement .
* we would like to thank navin kashyap , haim permuter , oron sabag and wenyi zhang for insightful discussions and suggestions and for pointing out relevant references that result in great improvements in many aspects of this work .
d. m. arnold and h. a. loeliger , `` on the information rate of binary - input channels with memory , '' in _ proceedings of ieee international conference on communications _
, vol . 9 , pp . 26922695 , jun .
d. m. arnold , h. a. loeliger , p. o. vontobel , a. kavcic and w. zeng , simulation - based computation of information rates for channels with memory , " _ ieee .
inf . theory _ , vol.52 , no.8 , pp . 34983508 ,
2006 .
b. marcus , r. roth , and p. siegel , `` constrained systems and coding for recording channels , '' _ handbook of coding theory _ ,
i , ii.1em plus 0.5em minus 0.4emamsterdam : north - holland , pp . 16351764 , 1998 .
h. d. pfister , `` the capacity of finite - state channels in the high - noise regime , '' _ entropy of hidden markov processes and connections to dynamical systems _ , london math .
lecture note series .
cambridge : cambridge univ . press , 2011 , vol .
179222 .
h. pfister , j. b. soriaga , and p. siegel , `` on the achievable information rates of finite state isi channels , '' in _ proceedings of ieee global telecommunications conference _
, vol . 5 , pp . 29922996 , nov .
2001 .
yang , shaohua and kavcic , a. and tatikonda , s.,``feedback capacity of stationary sources over gaussian intersymbol interference channels , '' _ global telecommunications conference , 2006 .
globecom 06 .
, no . , pp.16 , nov . 27 2006-dec . 1 2006 . p. o. vontobel , a. kavi , d. m. arnold , and h. a. loeliger , `` a generalization of the blahut - arimoto algorithm to finite - state channels , '' _ ieee trans .
inf . theory _
54 , no .
18871918 , may 2008 . | in this paper , we examine an input - constrained erasure channel and we characterize the asymptotics of its capacity when the erasure rate is low . more specifically , for a general memoryless erasure channel with its input supported on an irreducible finite - type constraint , we derive partial asymptotics of its capacity , using some series expansion type formulas of its mutual information rate ; and for a binary erasure channel with its first - order markovian input supported on the @xmath0-rll constraint , based on the concavity of its mutual information rate with respect to some parameterization of the input , we numerically evaluate its first - order markov capacity and further derive its full asymptotics .
the asymptotics obtained in this paper , when compared with the recently derived feedback capacity for a binary erasure channel with the same input constraint , enable us to draw the conclusion that feedback may increase the capacity of an input - constrained channel , even if the channel is memoryless .
_ index terms : _ erasure channel , input constraint , capacity , feedback . |
after the discovery of the _ attractor mechanism _ in the mid 90 s @xcite-@xcite in the context of bps black holes ( bhs ) , recently _ extremal bh attractors _ have been object of intensive study @xcite@xcite .
this is mainly due to the ( re)discovery of new classes of scalar attractor configurations , which do not saturate the bps bound and , when considering a supergravity theory , break all supersymmetries at the bh event horizon .
the various classes of bps and non - bps attractors are strictly related to the geometry of the scalar manifold . in @xmath10 , @xmath11 ungauged supergravity coupled to @xmath12 abelian vector multiplets , the scalar manifold is endowed with the so - called special khler ( sk ) geometry ( see _ e.g. _ @xcite and refs . therein ) , in which the holomorphic prepotential function @xmath13 plays a key role . in general , sk geometry admits three classes of extremal bh attractors : @xmath14-bps ( preserving four supersymmetries out of the eight pertaining to asymptotical @xmath10 , @xmath11 superpoincar algebra , and two non - supersymmetric typologies , discriminated by the eventual vanising of the @xmath10 central charge function @xmath15 : non - bps @xmath16 and non - bps @xmath17 ( see _ e.g. _
@xcite for a detailed analysis in the case of symmetric sk geometries ) . when considering @xmath10 , @xmath11 supergravity coming from a string compactification , the perturbative quantum corrections to the holomorphic prepotential of sk geometries can be polynomials or some non - polynomial ( usually polylogarithmic ) functions of the moduli ( see _ e.g. _ @xcite and refs .
therein ) . in the case of cubic classical sk geometries , such as the ones obtained in the large volume limit of @xmath18-compactifications of type iia , the sub - leading nature of the quantum corrections constrains the most general polynomial correction to be at most of degree two in the moduli , with a priori complex coefficients .
in @xcite it has been shown that the only polynomial quantum perturbative correction to the prepotential of classical cubic sk geometries which is consistent with the perturbative ( continuous ) axion - shift symmetry @xcite is the constant purely imaginary term ( @xmath19 throughout ) : @xmath20 where @xmath21 is the real , constant , completely symmetric tensor defining the cubic geometry .
indeed , it can be easily shown that all other polynomial perturbative corrections ( quadratic , linear and real constant terms in the moduli ) do not modify the classical cubic geometry , since they do not contribute to the khler potential at all @xcite .
let us start by considering the holomorphic prepotential of a certain class of compactifications of the heterotic @xmath22 superstring over @xmath23 , having the form ( see _ e.g. _ @xcite and refs . therein ) @xmath24 since in this case the dilaton @xmath25 belong to a vector multiplet , there exist ( @xmath26-symmetric ) quantum perturbative string - loop corrections and non - perturbative corrections , as well . the tree - level , classical term @xmath27 is the prepotential of the generic cubic sequence @xmath28 ( @xmath29 ) of homogeneous symmetric sk manifolds ( see _ e.g. _ @xcite , and @xcite and refs . therein ) , in the symplectic gauge exhibiting the largest possible amount of explicit symmetry @xmath30 .
due to non - renormalization theorems , all the quantum perturbative string - loop corrections are encoded in the 1-loop contribution @xmath31 , which contains a constant term , a purely cubic polynomial term and a polylogarithmic part ( see _ e.g. _ @xcite and refs . therein ) .
finally , @xmath32 encodes the non - perturbative corrections , exponentially suppressed in the limit @xmath33 . by exploiting the type iia / heterotic duality and correspondingly by suitably identifying the relevant moduli fields ,
the heterotic prepotential ( [ f - heterotic ] ) becomes structurally identical to the one arising from type iia compactifications over calabi - yau threefolds ( @xmath18s ) .
in such a case , the prepotential of the resulting low - energy @xmath10 , @xmath11 supergravity is purely of classical origin .
indeed , there are only khler structure moduli , and the dilaton @xmath25 belongs to an hypermultiplet ; thus , there are no string - loop corrections , and all corrections to the large - volume limit cubic prepotential come from the world - sheet sigma - model @xcite . as shown in @xcite * * , * * there are no 1- , 2- and 3-loop contributions . moreover , the non - perturbative , world - sheet instanton corrections destroy continuous axion - shift symmetry , by making it discrete @xcite . by disregarding such non - perturbative world - sheet instanton corrections ,
the type iia prepotential can be written as follows ( @xmath34 ) @xcite : @xmath35 the @xmath36 are the real classical intersection numbers , determining the large volume limit cubic sk geometry .
the perturbative contributions from @xmath7-dimensional cft on the world - sheet are encoded in a linear and in a constant term .
the linear term is determined by the @xmath37 , which are the real expansion coefficients of the second chern class @xmath38 of @xmath18 with respect to the basis @xmath39 of the cohomology group @xmath40 , dual to the basis of the ( 1,1)-forms @xmath41 of the cohomology @xmath42 .
it has been shown that the linear term @xmath43 can be reabsorbed by a suitable symplectic transformation of the period vector , and thus in the heterotic picture it has just the effect of a constant shift in @xmath44 ( @xcite * * ; * * see also @xcite ) . from the general analysis of @xcite ,
the constant term is the only relevant one .
it is determined by the euler characters @xmath45 , and thus @xmath46 is of order @xmath6 .
notice that @xmath47 for _ self - mirror _ @xmath18s , and thus such a constant term vanishes .
moreover , for some particular _ self - mirror _ models , such as the so - called fhsv one @xcite , also the non - perturbative , world - sheet instanton corrections vanish ; thus , in such models , up to suitable symplectic transformations of the period vector , the classical cubic prepotential does not receive any perturbative and non - perturbative correction . ]
@xmath48 of @xmath18 ( @xmath49 is the riemann zeta - function ) , and it has a 4-loop origin in the non - linear sigma - model @xcite .
it is worth pointing out here that @xmath18-compactifications of type iib do not admit a large volume limit ; moreover , the attractor eqs .
only depend on the complex structure moduli ( which are the scalars of the @xmath10 vector multiplets ) .
the solutions to @xmath10 , @xmath11 attractor eqs . for the resulting sk geometries were studied in @xcite for the particular class of fermat @xmath18s with @xmath50 , and in @xcite for a particular @xmath18 with @xmath51 .
aim of the present paper is to study the solutions to the @xmath10 , @xmath11 attractor eqs . in a dyonic background in the simplest case of perturbative
quantum corrected cubic sk geometry , namely in the @xmath6-modulus sk geometry described ( in a suitable special symplectic coordinate ) by the holomorphic khler gauge - invariant prepotential @xmath0 , with @xmath1 ( see eq .
( [ t^3+ic ] ) below ) . by doing so
, we will extend the bps analysis of @xcite .
the plan of the paper is as follows . in sect .
[ classical ] we shortly review the classical so - called @xmath52 model , focussing on the attractors supported by the `` magnetic '' bh charge configuration , in which the charge vector @xmath53 .
thence , in sect .
[ quantum ] we will study the solutions to @xmath10 , @xmath11 attractor eqs . for the @xmath54 model , dividing our analysis in the classical bps ( subsect .
[ quantum - bps ] ) and non - bps ( subsect .
[ quantum - non - bps ] ) charge domains .
despite the simplicity of the correction added to the classical prepotential @xmath55 ( see eq . ( [ t^3 ] ) below ) , we will find evidence for interesting phenomena , such as the _ `` separation '' _ and the _ `` transmutation '' of extremal bh attractors_. finally , in subsect .
[ d0-d6 ] we briefly consider the so - called @xmath56 bh charge configuration @xmath57 , pointing out that it does not support admissible bps attractors , both at the classical and quantum level . an appendix , containing some technical details , concludes the paper .
the classical so - called @xmath52 model is a @xmath6-modulus model based on a cubic holomorphic prepotential which ( in the local special symplectic coordinate @xmath58 ) reads @xmath59 constrained by the condition @xmath60 the corresponding manifold is the rank-@xmath6 symmetric special khler ( sk ) space @xmath61 .
this is an isolated case in the classification of the symmetric sk manifolds ( see _ e.g. _ @xcite , and @xcite and refs .
therein ) .
furthermore , it is worth pointing out that @xmath61 can be endowed with a khler gauge - invariant quadratic prepotential @xmath62 , as well ; in such a case , it corresponds to the @xmath63 element of the sequence of symmetric irreducible quadratic sk manifolds @xmath64 ( @xmath65 , see _ e.g. _
@xcite , and @xcite and refs .
therein ) , with geometric properties completely different from the cubic case ( see _ e.g. _ the discussion in @xcite ) .
the reason for such a difference lies in the fact that the four charges @xmath66 sit in the spin @xmath67 real representation of @xmath68 in the former case , and in the spin @xmath14 complex representation of @xmath69 in the latter case . at the present time
, the @xmath52 model is the only model whose @xmath70 , @xmath11 attractor eqs .
have been solved for a completely generic bh charge configuration @xmath71 .
its @xmath14-bps solutions were known after @xcite , whereas in @xciteits non - bps solutions with non - vanishing @xmath10 central charge @xmath15 were determined for @xmath72 . recently , in @xcite such a result was extended to the general case @xmath73 .
since such a model does not admit non - bps @xmath17 attractors , in the following `` non - bps '' will be understood for `` non - bps @xmath16 '' .
furthermore , we will always consider the simple case corresponding to the so - called `` magnetic '' charge configuration , in which the only non - vanishing bh charges are @xmath74 and @xmath75 ( in a stringy interpretation , this corresponds to consider only @xmath76 and @xmath77 branes ) . in the remaining of this section we will report
the main known results about @xmath52 model in `` magnetic '' bh charge configuration , in order to make the comparison with the quantum - perturbed case easier . the khler potential and metric function
are given by @xmath78 , \qquad g\equiv \partial _ { t}\overline{\partial } _ { \overline{t}}k=\frac{3}{4({{\tt im\,}}t)^{2 } } ; \label{fkgclass}\ ] ] thus the condition ( [ classkcons ] ) is nothing but the reality condition for @xmath79 .
notice that @xmath80 for @xmath81 , and in particular in the lower half of the argand - gauss plane @xmath82 determined by ( [ classkcons ] ) .
the superpotential , its covariant derivative and the effective black hole ( bh ) potential are respectively given by @xmath83 the bps and non - bps solutions to the attractor eq .
@xmath84 are always stable , and they are supported by bh charges satisfying @xmath85 and @xmath86 , respectively and @xmath86 respectively as the `` magnetic '' branches of the 2 homogeneous symmetric bh charge bps and non - bps @xmath16 orbits of the symplectic vector representation space of the @xmath87-duality group @xmath68 of the @xmath52 model ( see appendix ii of @xcite ) . ] @xcite .
they are both axion - free : @xmath88{0.2pt}{2em}\frac{q_{0}}{p^{1}}\rule[-0.75em]{0.2pt}{2em}\ , } , \label{classical - solutions}\ ] ] and the corresponding bh entropy reads @xmath89{0.2pt}{1.4em}\,q_{0}\left ( p^{1}\right ) ^{3}\rule[-0.4em]{0.2pt}{1.4em}\,}. \label{classical - sbh}\ ] ]
from the considerations made in the introduction , it follows that the most general form of @xmath6-modulus cubic geometry with polynomial quantum perturbative corrections consistent with axion - shift symmetry is based on the holomorphic khler gauge invariant prepotential function @xmath91 let us start by noticing that the sk manifold based on the prepotential ( [ t^3+ic ] ) is no more symmetric nor homogeneous .
its khler potential and metric function have the following form : @xmath92 , \qquad g=3\frac{\left ( \strut ( { { \tt im\,}}t)^{3}-\lambda \right ) { { \tt im\,}}t}{\left ( \lambda + 2({{\tt im\,}}t)^{3}\right ) ^{2}}. \label{geom - t^3+ic}\ ] ] in the `` magnetic '' bh charge configuration , the superpotential , its covariant derivative and the effective bh potential respectively reads @xmath93 .
\label{wk}\ ] ] when switching @xmath94 , the reality condition ( [ classkcons ] ) on the khler potential gets modified as follows : @xmath95{\frac{\lambda } { 2}}.\ ] ] furthermore , the positivity of the metric function yields another restriction on allowed values of @xmath96 : @xmath97 @xmath98 thus , @xmath96 belong to the shaded regions represented in fig .
[ regions ] . if @xmath99 , the allowed region consists of a single interval , otherwise it is the union of two non - overlapping intervals : @xmath100{\frac{\lambda } { 2}}\;\right ) \\
\displaystyle\lambda < 0 : & \qquad { { \tt im\,}}y\in \left ( -\infty , \sqrt[3]{\strut \lambda } \;\right ) \,\bigcup \,\left ( 0,-\sqrt[3]{\frac{\lambda } { 2}}\;\right ) .
\end{array}\ ] ] it is instructive to consider the @xmath2-dependence of the ricci scalar curvature @xmath101 . whereas in the classical case @xmath102 , in the considered quantum case @xmath101 is not constant nor necessarily negative anymore : @xmath103 for @xmath99 the @xmath104 limit gives @xmath105 ; on the other hand , @xmath101 decreases approaching the boundary value @xmath106{\lambda /2}$ ] , where it equals @xmath107 . for @xmath108 , @xmath101 increases starting from its value at @xmath109 , and it diverges in the limit @xmath110{\strut \lambda } ^{-}$ ] . in the other interval @xmath111{\frac{\lambda } { 2}}\;\right ) $ ] , @xmath101 diverges in the limit @xmath112 , whereas it reaches its minimal value @xmath107 at the boundary @xmath106{\frac{\lambda } { 2}}$ ] .
concerning the effective bh potential , it is a real function of one complex variable @xmath58 , and it contains three parameters : the charges @xmath113 and @xmath114 and the quantum parameter @xmath2 .
it enjoys the following properties with respect to reflection of its arguments : @xmath115 in order to decrease the number of independent parameters and simplify the analysis , in the following treatment we will redefine @xmath58 and @xmath2 in such a way to factorize the dependence of @xmath116 on the charges .
the second reflection property shows that @xmath116 is somehow sensitive to the sign of @xmath117 .
this fact , in light of the form of the `` magnetic '' supporting bh charge orbits in the classical case , leads us to divide the treatment of the quantum model in two parts , respectively corresponding to : 1 .
@xmath85 and refer to this range of charges as _ classical bps charge domain _ ( or shortly _ bps domain _ ; for @xmath118 it supports only bps attractors ) ; 2 .
@xmath86 and refer to this range of charges as _ classical non - bps charge domain _ ( or shortly _ non - bps domain _ ; for @xmath119 it supports only non - bps attractors ) .
let us switch to a new coordinate @xmath120 and to a new quantum parameter @xmath121 : @xmath122 in terms of which the dependence of all the considered quantities on the charges is factorized instead of @xmath123 .
the same holds for @xmath124 .
] : @xmath125 \displaystyle g=\frac{3p^{1}}{q_{0}}\,\frac{\left ( ( { { \tt im\,}}y)^{3}-\alpha \right ) { { \tt im\,}}y}{\left ( 2({{\tt im\,}}y)^{3}+\alpha \right ) ^{2}},\qquad v_{bh}=v(y,\bar{y},\alpha ) \frac{q_{0}}{\displaystyle\sqrt{\frac{q_{0}}{(p^{1})^{3 } } } } ; \\[0.5em ] \displaystyle v(y,\bar{y},\alpha ) \equiv \frac{1}{4{{\tt im\,}}y\left ( \alpha ^{2}+\alpha { { \tt im\,}}y^{3}-2{{\tt im\,}}y^{6}\right ) } \left [ \rule[-1em]{0pt}{2.5em}12\alpha ^{2}\left ( { { \tt im\,}}y^{2}+{{\tt re\,}}y^{2}\right ) \right . \\[0.5em ] \displaystyle\phantom{v_{eff}=4{{\tt im\,}}y\left ( \alpha^2 \alpha { { \tt im\,}}y^3 \right)}\left .
+ 4{{\tt im\,}}y^{4}\left ( 3{{\tt im\,}}y^{4}+12{{\tt im\,}}y^{2}{{\tt re\,}}y^{2}+\left ( 1 - 3{{\tt re\,}}y^{2}\right )
^{2}\right ) \right .
\\ \displaystyle\phantom{v_{eff}=4{{\tt im\,}}y\left ( \alpha^2 \alpha { { \tt im\,}}y^3 \right)}\left
. + \alpha { { \tt im\,}}y\left ( 3{{\tt im\,}}y^{4}-\left ( 1 - 3{{\tt re\,}}y^{2}\right ) ^{2}-6{{\tt im\,}}y^{2}\left ( 3+{{\tt re\,}}y^{2}\right ) \right ) \rule[-1em]{0pt}{2.5em}\right ] .
\end{array } \label{wkgbpsd}\ ] ] as stated above , the topology of the allowed regions of @xmath96 depends on the sign of @xmath2 , as given by fig .
[ regions ] . in terms of @xmath120 and @xmath126 defined in the bps domain by eq .
( [ def - bps ] ) , the corresponding allowed regions are represented in fig .
[ bpsregions ] .
@xmath127 @xmath128 for example , if one considers @xmath129 then for @xmath130 there are two disconnected allowed intervals for @xmath131 : @xmath132{\strut \alpha } \;\right ) \,\bigcup \,\left ( 0,-\sqrt[3]{\frac{\alpha } { 2}}\;\right ) , \ ] ] while for @xmath133 it can range in only one interval : @xmath134{\strut \frac{\alpha } { 2}},\infty ) .\ ] ] in the classical case there is a one - to - one correspondence between the symplectic vector of bh charges inserted as input in the attractor eq .
( [ ae ] ) and the solutions to this equation .
thus , for example in the bps domain @xmath135 , when inserting two arbitrary values for @xmath114 and @xmath113 of the same sign , in the classical case there exists one and only one solution ( in particular , of bps type ) to the classical attractor eq .
( [ ae ] ) .
this is no more true in the considered quantum case : in the bps domain @xmath135 , when inserting two arbitrary values for @xmath114 and @xmath113 of the same sign , now there exist two stable critical points may exist ; however , we will not deal with them , because they do not determine _ attractors _ in strict sense . ] of @xmath116 , one bps and the other non - bps .
it is worth remarking that in the considered bps domain the non - bps critical points of @xmath116 do not admit a classical limit @xmath136 .
furthermore , while in the classical case both bps and non - bps critical points of @xmath116 supported by a `` magnetic '' bh charge configuration are axion - free , in the quantum case also non - bps critical points of @xmath116 with non - vanishing axion may arise out ( and they can also be stable ) . as in the classical case , also in the
considered quantum case non - bps critical points of @xmath116 with @xmath17 do not exist at all . a pictorial view of the more complicated situation typical of the quantum case is given by fig . [ bpssolutions ] .
as mentioned in the appendix , in both domains ( bps and non - bps ) of the quantum case the bps solutions are known analytically , whereas the non - bps solutions can be investigated only numerically .
[ cols="<,^ , > " , ] concerning the bps minima , for @xmath129 they exist in the range @xmath137 , whereas for @xmath138 they exist for @xmath139 .
the same holds for @xmath133 , but with opposite sign of @xmath140 . as mentioned , in the considered non - bps domain there are two kinds of non - bps minima , both axion - free ( see fig .
[ nonbpssolutions ] ) .
the corresponding relevant critical value of the quantum parameter is @xmath141 . as mentioned above , since the bps critical points
can be computed analytically , one can calculate the @xmath126-dependent expression of the bps bh entropy in the non - bps domain to be @xmath142 where the solution of the cubic equation must be chosen inside the allowed region(s ) of the moduli space , and the @xmath143 branches of @xmath144 must be chosen in order to obtain @xmath145 .
as it has to be in the non - bps domain , in the classical limit the cubic equation has no admissible solutions .
let us now analyze the evolution @xmath116 with respect to the quantum parameter @xmath126 , choosing @xmath129 without loss of generality . the classical limit of the function @xmath146 defined in eq .
( [ wkgnonbpsd ] ) has the form of a `` scoop '' , with a minimum at the point @xmath147 , @xmath148 ( see eq .
( [ classical - solutions ] ) and fig .
[ nonbpsplot ] ) . in the classical limit ( @xmath149 , left ) and for @xmath150 ( right),title="fig:",scaledwidth=45.0% ] in the classical limit ( @xmath149 , left ) and for @xmath150 ( right),title="fig:",scaledwidth=45.0% ]
as done in the bps domain , let us now slightly increase the value of @xmath121 from @xmath151 to @xmath152 . as one can see from fig .
[ nonbpsplot ] , @xmath153 is in the range supporting the _ `` separation '' of attractors _ , _ _ _ _ in this case corresponding to the coexistence of a bps attractors with a non - bps one , both axion - free .
such stable critical points of @xmath116 respectively have coordinates @xmath154 $ bps:$ & $ { { \tt re\,}}y=0 $ , & $ { { \tt im\,}}y\approx 0.02 $ , & $ v_{bh , bps}\approx 24.98\sqrt{-q_{0}(p^{1})^{3}}.$\end{tabular } \label{comparison-2}\ ] ] as one can see from the plot of @xmath116 and analogously to what happens in the bps domain , in contrast to the classical case there appears a new chump , which contains a stable bps critical point ( not existing in the classical limit ) .
the width of such a disconnected branch of @xmath116 ( which vanishes in the classical limit ) and its separation from the other branch admitting a consistent classical limit are both proportional to @xmath155{\alpha } $ ] . as in the bps domain ,
such two branches of @xmath116 are separated by a region in which @xmath156 . as yielded by eq .
( [ comparison-2 ] ) , for @xmath157 it holds that @xmath158 ( see discussion below ) .
however in the non - bps domain , once again differently from the bps domain , in the range @xmath159 supporting the _ `` separation '' of attractors _ one can observe also the phenomenon of the _ `` transmutation '' _ of the supersymmetry - preserving features of the such minima of @xmath116 .
this can be seen by increasing @xmath126 .
of the two branches of @xmath116 : scoop ( admitting a classical limit , left ) and hump ( vanishing in the classical limit , right),title="fig:",scaledwidth=50.0% ] of the two branches of @xmath116 : scoop ( admitting a classical limit , left ) and hump ( vanishing in the classical limit , right),title="fig:",scaledwidth=50.0% ] as evident by looking at fig .
[ vbhnonbpsdomevol ] , the scoop chump of @xmath116 ( _ i.e. _ the branch of @xmath116 containing an axion - free non - bps minimum and admitting a classical limit ) , does not undergo any qualitative change when increasing @xmath126 . on the other hand ,
the hump chump of @xmath116 ( _ i.e. _ the one vanishing for @xmath136 ) exhibits a _
`` transmutation '' _ of the minimum : for @xmath137 it constains a bps minimum . for @xmath160 such a critical point of @xmath116
sits on the boundary ( see fig . [ nonbpsregions ] ) where @xmath161 diverges , so it is not an allowed critical point of @xmath116 . for
@xmath162 the minimum is still present , it is not bps any more , but rather it is non - bps ( and axion - free ) . thence , for @xmath163 the hump chump of @xmath116 does not contain any critical point . also the scoop chump of @xmath116 exhibits a _
`` transmutation '' _ of the minimum , but for negative values of @xmath126 . indeed , when @xmath126 gets smaller than @xmath164 the axion - free non - bps minimum _
`` transmutes '' _ into a bps one ( axion - free , as all the admissible bps critical points found in our analysis ) .
the shape of @xmath116 for various values of @xmath126 is plotted in fig .
[ vbhnonbpsdomevol ] .
one might see that for small values of @xmath126 ( blue line ) the hump chump contains a bps minimum . at the critical value @xmath160 ( green line ) the minimum disappears and reappears , but non - bps , when @xmath139 ( red line ) .
for @xmath163 , no critical points exist at all ( black line ) .
the @xmath126-dependence of the bh entropy is shown in fig .
[ entropynonbps ] .
consistently with the colors used in fig . [ nonbpssolutions ] , the red plot corresponds to bps bh entropy , whereas the blue and green plots to non - bps bh entropy . as done in the bps domain
, it is worth pointing out that the definitions ( [ def - non - bps ] ) are dimensionless in charges , and thus all considered phenomena hold for any magnitude of bh charges @xmath74 and @xmath75 .
thus , the non - bps domain exhibits various differences with respect to the bps domain , which can be summarized as follows : _ i ) _ all non - bps minima in such a domain are axion - free ; _ ii ) _ the _ `` transmutation '' _ of the supersymmetry - preserving features of the minimum of @xmath116 happens not only in the branch of @xmath165 admitting a classical limit , but also in the one vanishing in such a limit ; _ iii ) _ the _ `` transmutation '' _ inside the branch of @xmath116 vanishing in the classical limit actually determines a _
`` transmutation '' _ in the region of _ `` separation '' _ of attractors : for @xmath137 a bps and a non - bps minimum coexist ; when @xmath126 gets bigger than @xmath166 the bps minimum
_ `` transmutes '' _ into a non - bps one , who then coexists with the pre - existing non - bps ( axion - free ) minimum until @xmath126 reaches the critical value @xmath167 ; _ iv ) _ as evident by looking at fig .
[ entropynonbps ] , in the range @xmath137 supporting the coexistence of a bps and a non - bps minimum , the sign of @xmath168 ( for fixed bh charges @xmath114 and @xmath113 satisfying @xmath169 ) changes .
indeed , there exists a particular value @xmath170 for which remarkably @xmath171 .
this means that for @xmath172 two _ different _ purely charge - dependent attractor configurations of the scalar @xmath58 exist ( one preserving one half of the @xmath173 supersymmetries pertaining to the asymptotical @xmath10 , @xmath11 superpoincar algebra and the other not preserving any of them ) such that they determine the _ same _ bh entropy .
for @xmath174 it holds @xmath175 , whereas for @xmath176 it holds @xmath177 ; _ v ) _ for @xmath162 it holds @xmath178 , where @xmath179 is the bh entropy determined by the non - bps minimum originated from the bps one by _
`` transmutation '' _ ( see the green and blue plots in fig . [ entropynonbps ] ) ; _ vi _ ) for @xmath163 , the branch of @xmath116 vanishing in the classical limit actually still contains critical points of @xmath116 , but they are not admissible and/or they are not stable .
-dependence of the bh entropy @xmath144 for @xmath180,scaledwidth=60.0% ] it is worth pointing out that all the obtained results are expressed in terms of coordinates and quantum parameters _ dimensionless _ in the bh charges ( _ i.e. _ depending on _ ratios _ of bh charges , as given by eqs .
( [ def - bps ] ) and ( [ def - non - bps ] ) ) .
thus , _ apriori _
they hold for every range of values of the bh charges .
actually , we disregard the actual quantization of the electric and magnetic charges , considering @xmath181 , rather than the 4-dimensional charge lattice @xmath182 , to be the space spanned by the bh charge vector @xmath183 . in other words ,
all our analysis and the obtained results hold in the semiclassical regime of large , continuous , real bh charges . from the above analysis of the extremal bh attractors supported by the `` magnetic '' charge configuration @xmath184 in the @xmath54 model , one might be lead to argue that starting from a bh charge configuration supporting only one class of attractors at the classical level , at the quantum level one obtains that such a charge configuration generally supports more than one class of attractors ( at least for a certain range of the quantum parameter @xmath2 ) .
this is actually not true , and a counterexample is provided by the bh charge configuration @xmath185 , usually named kaluza - klein bh ( in @xmath186-theory language ) @xcite or @xmath56 system ( in type iia calabi - yau compactifications in the language of superstring theory ) @xcite , for which the superpotential acquires the form @xmath187 it is easy to check that at the classical level such a charge configuration only supports non - bps attractors @xcite .
differently from what happens for the `` magnetic '' charge configuration , this holds also at the quantum level , @xmath188 ; in other words , the @xmath54 model does not have admissible bps critical points of @xmath116 supported by the @xmath56 charge configuration .
this can be easily realized by looking at the real and imaginary parts of the bps attractor equation for such a charge configuration , respectively of the following form ( having a smooth classical limit ) : @xmath189 \displaystyle\frac{6{{\tt im\,}}t(2\lambda p^{0}{{\tt re\,}}t+p^{0}{{\tt im\,}}t^{3}{{\tt re\,}}t+{{\tt im\,}}t(q_{0}+p^{0}{{\tt re\,}}t^{3}))}{\lambda + 2{{\tt im\,}}t^{3}}=0 . & \end{array}\ ] ] by recalling the expression of the metric function @xmath190 given by eq .
( [ geom - t^3+ic ] ) , it is clear that all bps critical points of @xmath116 supported by the @xmath56 charge configuration are points where @xmath191 , and therefore they are not admissible . also in view of the analysis of @xcite , such a result can be directly extended to the axion - free bps critical points of @xmath116 in a @xmath192__-corrected _ _ sk cubic geometry , based on the khler gauge - invariant holomorphic prepotential ( in a suitable basis of special symplectic coordinates @xmath193 , @xmath19 ) @xmath194 which is nothing but the @xmath12 moduli generalization of eq .
( [ t^3+ic ] ) .
the @xmath192__-corrected _ _ real metric @xmath195 can be easily computed from eq .
( [ d - skg+ic ] ) to be @xmath196 , \label{cern1}\ ] ] where @xmath197 , @xmath198 and @xmath199 .
notice that the limit @xmath200 consistently yields the expression of the real metric of a sk cubic geometry , given by eq .
( 2.4 ) of @xcite . in @xcite the axion - free bps critical points of @xmath116 in the sk geometry based on @xmath13 given by eq .
( [ d - skg+ic ] ) have been computed to be @xmath201 where the charge - dependent quantity @xmath202 satisfies the third order algebraic equation @xmath203 it is then clear that the @xmath56 bh charge configuration does not support axion - free bps critical points of @xmath116 , because for @xmath204 @xmath205 eqs .
( [ cern1 ] ) and ( [ cern2 ] ) yield @xmath206 .
notice that the non - admissibility of the axion - free bps critical points of @xmath116 in the @xmath56 bh charge configuration would not be evident by only looking at the bps bh entropy determined by axion - free solutions , which reads @xmath207 whose limit ( independent on @xmath12 ) in the @xmath56 bh charge configuration is finite ( and constrains @xmath208 ) : @xmath209
the work of s.b . , s.f . and a.s .
has been supported in part by the european community human potential program under contract mrtn - ct-2004 - 005104 _ `` constituents , fundamental forces and symmetries of the universe''_. the work of s.f . has been supported in part by european community human potential program under contract mrtn - ct-2004 - 503369 _ `` the quest for unification : theory confronts experiments '' _ , in association with infn frascati national laboratories and by d.o.e .
grant de - fg03 - 91er40662 , task c. the work of a.m. has been supported by a junior grant of the _ `` enrico fermi '' _ center , rome , in association with infn frascati national laboratories .
the complex bps attractor eq . for the quantum @xmath6-modulus model based on the prepotential
@xmath0 has the form @xmath210 by taking its real and imaginary parts , one obtains the following two real eqs . :
@xmath211 whose solution reads @xmath212 obviously , the number of real roots of the obtained algebraic cubic eq . depends on the relations between its coefficients . on the other hand , the real part of the complex non - bps attractor eq . for the model based on @xmath0
reads @xmath213}{4{{\tt im\,}}t^{7}-2{{\tt
im\,}}t^{4}\lambda -2{{\tt im\,}}t\lambda ^{2}}=0 ; \label{re - non - bps - aes}\ ] ] its imaginary part is rather cumbersome , and it is not worth presenting it here . the above eqs . get modified when passing to the special coordinate and quantum parameter suitably defined in the bps and non - bps domains ( see definitions ( [ def - bps ] ) and ( [ def - non - bps ] ) , respectively ) . in the following treatment
we will analyze them case - by - case . in the bps domain the bps attractor equations ( aes ) ( [ bps - aes ] )
acquire the form @xmath214 from eq .
( [ wkgbpsd ] ) it follows that , in order to have a regular solution , the real part of the modulus @xmath120 has to vanish , while its imaginary part has to satisfy a cubic eq . :
@xmath215 depending on @xmath126 , the obtained cubic eq . has up to three real roots . if @xmath216 there are three real roots but , due to the consistency conditions on the metric and khler potential , only one of them turns out to determine an admissible critical point of @xmath116 . on the other hand , for @xmath217 the situation
is described by fig .
[ bpssolutions ] .
notice that , as it has to be in the bps domain , in the classical limit @xmath136 one reobtains the classical solutions given by eq .
( [ classical - solutions ] ) . in the bps domain
the real part ( [ re - non - bps - aes ] ) of the non - bps aes acquire the form @xmath218 once again , the imaginary part is cumbersome , and it is not worth writing it here .
( [ recritbpsd ] ) can be solved with respect to @xmath124 , obtaining two solutions : @xmath219 we are now going to substitute each of these solutions in the imaginary part of the non - bps aes . in this case
the imaginary part of the non - bps aes acquires form @xmath220 = 0 .
\label{12-oct-1}\ ] ] it is worth remarking that the cubic term @xmath221 which gets factorized determines the bps solution ( [ bpssolbpsd ] ) , and thus here it has to be disregarded .
the other factor in eq .
( [ 12-oct-1 ] ) is the one yielding the non - bps solutions .
as it has to be since we are considering the bps domain , in the classical limit @xmath136 this factor reduces to a polynomial of the form @xmath222 , which has no admissible solutions . since the solution of an algebraic eq . of ninth order can not be found analytically , one can analyze it numerically , and determine the number of real solutions depending on the value of @xmath126 .
it turns out that @xmath126 has two critical values : one of them is @xmath223 , while the other only numerically , @xmath224 . for @xmath225 and @xmath226 , only one real root exists , while for @xmath227 there are tree real roots . by choosing @xmath129 ( without any loss of generality ) and requiring the stability of the solutions and the belonging to the allowed regions of the moduli space given by fig .
[ bpsregions ] , one gets that in the bps domain stable and admissible non - bps axion - free critical points of @xmath116 exist for @xmath228 ( see the green line in the plot on the left of fig .
[ bpssolutions ] ) .
the other possibility is to substitute the more complicated expression for @xmath124 given by eq .
( [ reybpsd ] ) into the imaginary part of the non - bps aes . by doing so
, the imaginary part of the non - bps aes reduces to an algebraic eq . of fifteenth
order : @xmath229 \displaystyle\phantom{2\alpha^5 + } -92\alpha ^{2}{{\tt im\,}}y^{9}+320\alpha { { \tt im\,}}y^{10}-32\alpha { { \tt im\,}}y^{12}-128{{\tt im\,}}y^{13}-32{{\tt im\,}}y^{15}=0 , \end{array } \label{oct-12-evening-2}\ ] ] which clearly can not be solved analytically . as it has to be since we are considering the bps domain , in the classical limit @xmath136 eq .
( [ oct-12-evening-2 ] ) reduces to a polynomial of the form @xmath230 , which has no admissible solutions . by inspecting the number of real roots of such an eq . with respect to the value of @xmath126 , one obtains two critical values of @xmath126 .
one of them is known just numerically : @xmath231 , while the other one is @xmath166 .
when @xmath232 , five real roots exist but only one of them is admissible , whereas for @xmath233 and @xmath139 three and one real root exist respectively , but none of them is admissible .
thus ( by choosing @xmath129 , once again without loss of generality ) , in the bps domain stable and admissible non - bps critical points of @xmath116 with non - vanishing axion exist for @xmath232 ( see the blue line in the plot on the left of fig .
[ bpssolutions ] ) . in the non - bps domain the bps aes ( [ bps - aes ] )
acquire the form @xmath234 from eq .
( [ wkgnonbpsd ] ) it follows that , in order to have a regular solution , the real part of the modulus @xmath120 has to vanish , while its imaginary part has to satisfy a cubic eq .
: @xmath235 such a cubic eq .
has only one real root @xmath236 : @xmath237{\alpha -\sqrt{\alpha ^{2}+\frac{1}{27}}}+\sqrt[3]{\alpha + \sqrt{\alpha ^{2}+\frac{1}{27}}}. \label{12-oct-2}\ ] ] even if such a solution exists @xmath238 , the admissible bps minima exist for @xmath129 in the range @xmath137 , and for @xmath138 in the range @xmath139 .
the same holds for @xmath133 , but with opposite sign of @xmath74 ( see the red lines in fig .
[ nonbpssolutions ] ) .
notice that , as it has to be in the non - bps domain , in the classical limit @xmath136 the cubic eq .
has not admissible solutions . in the non - bps domain
the real part ( [ re - non - bps - aes ] ) of the non - bps aes acquire the form @xmath239 once again , the imaginary part is cumbersome , and it is not worth writing it here .
( [ recritnbpsd ] ) can be solved with respect to @xmath124 , obtaining two solutions : @xmath240 notice that the non - axion - free branch of solutions given by the second expression of eq .
( [ 12-oct - evening-1 ] ) is not consistent with the classical admissible and stable non - bps solutions , which are axion - free ( see eq .
( [ classical - solutions ] ) ; on the other hand , as stated below , admissible and stable non - axion - free non - bps solutions do not exist in the considered quantum case , too ) . in this case
the imaginary part of the non - bps aes acquires form @xmath241 = 0 .
\label{12-oct-3}\ ] ] it is worth remarking that the cubic term @xmath242 which gets factorized determines the bps solution ( [ 12-oct-2 ] ) , and thus here it has to be disregarded .
the other factor in eq .
( [ 12-oct-3 ] ) is the one yielding the axion - free non - bps solutions , and , as it has to be in the non - bps domain , in the classical limit @xmath136 it gives the classical solutions ( see eq .
( [ classical - solutions ] ) ) . since the solution of an algebraic eq . of ninth order can not be found analytically , as done in the bps domain one can analyze it numerically , and determine the number of real solutions depending on the value of @xmath126 .
it turns out that @xmath126 has two relevant critical values , one is @xmath166 and the other is known only numerically : @xmath141 . for @xmath163 and @xmath133 , only one real root exists , while for @xmath159 there are tree real roots . by choosing @xmath129 ( without any loss of generality ) and requiring the stability of the solutions and the belonging to the allowed regions of the moduli space given by fig . [ nonbpsregions ] , one gets that in the non - bps domain stable and admissible non - bps axion - free critical points of @xmath116 exist for @xmath243 and that they double in the range @xmath162 ( see the blue and green lines in the plot on the left of fig .
[ nonbpssolutions ] ) .
the other possibility is to substitute the more complicated expression for @xmath124 given by eq .
( [ 12-oct - evening-1 ] ) into the imaginary part of the non - bps aes . by doing so , as in the bps domain
the imaginary part of the non - bps aes reduces to an algebraic eq . of fifteenth
order : @xmath244 \displaystyle\phantom{2\alpha^5 + } + 92\alpha ^{2}{{\tt im\,}}y^{9}-320\alpha { { \tt im\,}}y^{10}-32\alpha { { \tt im\,}}y^{12}-128{{\tt im\,}}y^{13}+32{{\tt im\,}}y^{15}=0 , \end{array}\ ] ] which can not be solved analytically .
by performing a numerical inspection of the number of real roots depending on the values of @xmath126 , and imposing the conditions of stability and belonging to the allowed regions of the moduli space given by fig .
[ nonbpsregions ] , in this case one finds that there are no admissible solutions . in other words , in the non - bps domain admissible and stable non - bps critical points of @xmath116 with non - vanishing axion do not exist @xmath236 .
s. bellucci , s. ferrara and a. marrani , _ attractor horizon geometries of extremal black holes _ , contribution to the proceedings of the xvii sigrav conference,47
september 2006 , turin , italy , ` hep - th/0702019 ` .
a. misra and p. shukla , _
area codes , large volume ( non-)perturbative alpha - prime and instanton : corrected non - supersymmetric ( a)ds minimum , the inverse problem and fake superpotentials for multiple - singular - loci - two - parameter calabi - yau s _
, ` arxiv:0707.0105 ` . s. ferrara and a. marrani , _ black hole attractors in extended supergravity _ ,
to appear in the proceedings of _
13th international symposium on particles , strings and cosmology ( pascos 07 ) _ , london , england , 2 - 7 jul 2007 , ` arxiv:0708.1268 ` .
a. ceresole , r. dauria and s. ferrara , _ the symplectic structure of _ @xmath245 _ supergravity and its central extension _ , talk given at ictp trieste conference on physical and mathematical implications of mirror symmetry in string theory , trieste , italy , 5 - 9 june 1995 , nucl .
* 46 * ( 1996 ) , ` hep - th/9509160 ` .
r. d. peccei and h. r. quinn , _ constraints imposed by cp conservation in the presence of instantons _ , phys . rev . *
d16 * , 1791 ( 1977 ) @xmath251 r. d. peccei and h. r. quinn , _ cp conservation in the presence of instantons _ , phys .
lett . * 38 * , 1440 ( 1977 ) @xmath251 r. d. peccei and h. r. quinn , _ some aspects of instantons _ , nuovo cim .
* a41 * , 309 ( 1977 ) .
m. t. grisaru , a. van de ven and d. zanon , _ four loop beta function for the _ @xmath248 _ _ and _ _ @xmath245 _ _ supersymmetric nonlinear sigma model in two dimensions _ _ , phys . lett . * b173 * , 423 ( 1986 ) @xmath251 m. t. grisaru , a. van de ven and d. zanon , _ two dimensional supersymmetric sigma models on ricci flat khler manifolds are not finite _ , nucl . phys . *
b277 * , 388 ( 1986 ) @xmath251 m. t. grisaru , a. van de ven and d. zanon , _ four loop divergences for the _ @xmath248 _ _ supersymmetric nonlinear sigma model in two dimensions _
_ , nucl . phys . *
b277 * , 409 ( 1986 ) .
s. ferrara , j. a. harvey , a. strominger and c. vafa , _ second quantized mirror symmetry _ ,
b361 * , 59 ( 1995 ) , ` hep - th/9505162 ` @xmath251 p. s. aspinwall , _ an _
@xmath245 _ _ dual pair and a phase transition _ _ , nucl .
b460 * , 57 ( 1996 ) , ` hep - th/9510142 ` @xmath251 d. r. morrison and c. vafa , _ compactifications of f theory on calabi - yau threefolds . 1 _ , nucl
b473 * , 74 ( 1996 ) , ` hep - th/9602114 ` @xmath251 d. r. morrison and c. vafa , _ compactifications of f theory on calabi - yau threefolds
. 2 _ , nucl .
b476 * , 437 ( 1996 ) , ` hep - th/9603161 ` @xmath251 j. a. harvey and g. w. moore , _ exact gravitational threshold correction in the fhsv model _ ,
phys . rev .
* d57 * , 2329 ( 1998 ) , ` hep - th/9611176 ` @xmath251 a. klemm and m. marino , _ counting bps states on the enriques calabi - yau _ , ` hep - th/0512227 ` @xmath251 j. r. david , _ on the dyon partition function in _
@xmath245 _ _ theories _ _ , ` arxiv:0711.1971 ` .
a. chou , r. kallosh , j. rahmfeld , soo - jong rey , m. shmakova and wing kai wong , _ critical points and phase transitions in 5-d compactifications of m - theory _ , nucl .
b508 * , 147 ( 1997 ) , ` hep - th/9704142 ` .
moore , _ attractors and arithmetic _ , ` hep - th/9807056 ` @xmath251 g.w .
moore , _ arithmetic and attractors _ , ` hep - th/9807087 ` @xmath251 g.w .
moore , _ les houches lectures on strings and arithmetic _ , ` hep - th/0401049 ` .
h. ooguri , _ the entropic principle _ , lectures given at the _ international school of subnuclear physics _
, 45th course : _ searching for the `` totally unexpected '' in the lhc era _ , emfcsc , 29 aug .
2007 , erice , italy , to appear in the proceedings . | we study the attractor equations for a quantum corrected prepotential @xmath0 , with @xmath1,which is the only correction which preserves the axion shift symmetry and modifies the geometry . by performing computations in the `` magnetic '' charge configuration
, we find evidence for interesting phenomena ( absent in the classical limit of vanishing @xmath2 ) .
for a certain range of the quantum parameter @xmath2 we find a _
`` separation '' _ of attractors , _
i.e. _ the existence of multiple solutions to the attractor equations for fixed supporting charge configuration .
furthermore , we find that , away from the classical limit , a _
`` transmutation '' _ of the supersymmetry - preserving features of the attractors takes place when @xmath2 reaches a particular critical value .
+ + * s.bellucci@xmath3 , s.ferrara@xmath4 , a.marrani@xmath5 and a.shcherbakov@xmath3 * @xmath6 _ infn - laboratori nazionali di frascati , + via enrico fermi 40,00044 frascati , italy + ` bellucci , marrani , [email protected]`_ @xmath7 _ physics department , theory unit , cern , + ch 1211 , geneva 23 , switzerland + ` [email protected] ` _
@xmath8 _ department of physics and astronomy , + university of california , los angeles , ca usa + ` [email protected] ` _
@xmath9 _ museo storico della fisica e + centro studi e ricerche
enrico fermi " + via panisperna 89a , 00184 roma , italy _ |
it has been realized that quantum entanglement is a key ingredient in quantum computation , quantum communication , and quantum cryptography@xcite .
many two - level quantum systems , or qubits , have been widely used in quantum information@xcite .
recently , the bipartite qutrit systems have drawn people s attention and exhibited varieties of advantages .
it enables powerful computation@xcite , establishes secure communication@xcite and cryptography@xcite , and reduces the communication complexity@xcite .
additionally , the high energy quantum teleportation using neutral kaons has been investigated@xcite .
these hint the possibility of connecting quantum information and particle physics , this connection may reveal novel and interesting feature .
considering these two aspects , we take the mixed light pseudoscalar meson states , a type of bipartite qutrit system , into account . in particle physics ,
the study on mixed meson states@xcite plays an important role and many works are devoted to this field@xcite , especially the study on @xmath5 mixed state , since it provides a unique opportunity for testing qcd @xcite which is widely used in describing strong interaction . in the last decades , people have studied the mixing angle@xcite , the hadronic @xmath2-@xmath4 decay@xcite , @xmath2-@xmath4 mixing in radiative @xmath6-meson decay@xcite , and so on .
however , much work of particle physics has been done for testing the theory and experiment while hardly concerned with quantum information and the related algebra configuration , such as yangians . as is known ,
the light pseudoscalar meson states in quark - flavor basis have @xmath7 symmetry@xcite .
it is important to expand the research on the symmetry of pseudoscalar meson states to yangian @xmath0 .
yangian , as an algebra beyond the lie algebra , is a powerful mathematical method for investigating the new symmetry of quantum systems which are nonlinear and integrable . demonstrating and investigating whether simple physical systems possess yangian is important and helpful for exploring physical systems via quantum algebra .
people have found the yangian symmetry in many physical models , such as calogero - sutherland model@xcite , the hubbard model@xcite and the heisenberg model@xcite , etc .
the realizations of @xmath8@xcite as the simplest one in yangian algebra have been gained much attention .
however , the yangian related to the lie algebra @xmath7 , @xmath0 , which is closed to the light pseudoscalar meson states in particle physics , need to be investigated in more detail .
thus , much attention is going to be payed in this paper to the application of @xmath0 algebra in the meson systems here .
in the low mass hadron region , the violations of isospin symmetry for pseudoscalar meson states within qcd are generated by the admixtures of @xmath2-@xmath3-@xmath4 . @xmath2 and @xmath4 are linear combination of @xmath7 singlet @xmath9 and octet @xmath10 , @xmath3 is another @xmath7 octet . because of the important application and significance of @xmath2-@xmath3-@xmath4 mixing , we choose their superposition states as the initial state @xmath11 where @xmath12 , @xmath13 and @xmath14 are the normalized real amplitudes and they satisfy @xmath15 .
@xmath16 , @xmath17 , @xmath18 .
the entanglement degree of the genuine n - particle qutrit pure state@xcite is measured by its mean entropy @xmath19 where @xmath20 is the reduced partial von neumann entropy for the @xmath21th particle only , with the other @xmath22 particles traced out , and @xmath23 is the corresponding reduced density matrix . since the system in eq .
( [ eq1 ] ) is bipartite qutrit , i.e. @xmath24 , the corresponding entanglement degree of this initial state can be calculated via eq.([2 ] ) , thus we have @xmath25 the behavior of @xmath26 depending on @xmath12 and @xmath13 ( due to the normalization condition , @xmath14 is not an independent parameter . )
is given in fig .
[ fig1 ] . for different @xmath12 and @xmath13 .
] before constructing transition operators , let us first give a brief introduction of yangian @xmath0 .
@xmath0 algebra@xcite is generated by the generators @xmath27 which are usually defined as follows @xmath28 here @xmath29 form a @xmath7 algebra characterized by @xmath30 , @xmath31 form a local @xmath7 on the @xmath21 site , and are equal to half of the corresponding gell - mann matrices , @xmath32 , @xmath33 , @xmath34 are parameters or casimir operators and @xmath35 which satisfies @xmath36 a more practical expression is expressed as @xmath37 here @xmath38 are the generators of @xmath0 .
due to the transition effect of yangian generators , transition operators @xmath39 can be constructed as compositions of the generators in eq .
( [ a ] ) .
@xmath39 can be looked upon as a function of @xmath40 and @xmath41 , namely , @xmath42 $ ] . when acting the transition operator on @xmath43 in eq .
( [ eq1 ] ) , a final state @xmath44 can be gotten and its entanglement degree @xmath45 can also be calculated out via eq.([2 ] ) . different final states with desired entanglement degrees thus can be gotten by acting corresponding transition operators on initial states . in order to illustrate this issue clearly , several simple examples are going to be discussed in more detail .
as we discussed before , the entanglement degree of final state can be tuned by changing the parameters of the transition operator . as an example , let us act the transition operator of @xmath46 on the initial state in eq.([eq1 ] ) , the corresponding final state is @xmath47.\end{aligned}\ ] ] the normalizing condition leads to @xmath48 and @xmath49=1 $ ] . as discussed in the preceding section , the entanglement degree of the final state @xmath50 can be calculated by the use of eq.([2 ] ) as @xmath51 as we can see , by changing the value of @xmath33 and @xmath34 , the entanglement degree @xmath52 can be tuned .
this behavior is shown in fig .
[ fig3 ] for the case of @xmath53
. varies with @xmath33 and @xmath34 when @xmath53 . ]
another more simpler case is @xmath54 for which the final state reads latexmath:[\[\begin{aligned } with normalization condition @xmath48 and @xmath56=3 $ ] .
similarly , @xmath57 which also depends on @xmath33 and @xmath34 , as shown in fig . [ fig5 ]
. varies with @xmath33 and @xmath34 when @xmath53 . ]
these two examples clearly shows that by tuning the parameters @xmath33 and @xmath34 of transition operators , the entanglement degree of final state varies between 0 and 1
. however , there are exceptional cases in which the entanglement degree of final state is found to be the same as the initial state or be zero , independent with @xmath33 and @xmath34 , as is shown in the next section .
although generally , transition operators with different @xmath33 and @xmath34 provide final states with different entanglement degrees , there do exist special cases in which the final state possesses the same entanglement degree with the initial state or the final state are totally disentangled .
both cases are very important in quantum information and quantum computation . for the transition operator @xmath58 , the final state is @xmath59 with normalizing condition @xmath60 and @xmath48 .
it is very easy to verify that the entanglement degree of the final state @xmath50 is equal to the one of the initial state , namely , @xmath61 .
same thing happens on transition operator @xmath62 , corresponding to whom the final state reads @xmath63 with normalizing condition being @xmath60 and @xmath48 .
again , calculation shows that @xmath61 . without any difficulty
, it can be verified that transition operators @xmath64 , @xmath65 , and @xmath66 lead to total disentangled final states , independent of the choice of @xmath12 and @xmath13 in the initial state .
in conclusion , we have investigated the entanglement degrees of pseudoscalar meson states via quantum algebra @xmath0 . by making use of transition effect of generators @xmath1 of @xmath0
, we have constructed various transition operators in terms of @xmath1 , and have acted them on @xmath2-@xmath3-@xmath4 mixing meson state .
the entanglement degrees of both the initial state and final state have been calculated with the help of entropy theory .
our result shows that a state with desired entanglement degree can be achieved by acting proper chosen transition operator on an initial state .
this is very helpful to control the degree of entanglement in quantum communication .
entanglement has been considered as an essential resource in most applications of quantum information@xcite . although the quantum channels used in quantum teleportation are usually represented by a maximally entangled pair@xcite
, partially entangled quantum channel becomes a hot topic@xcite nowadays due to the noise factors in the realistic world .
moreover , in quantum computing , a new important development is the probabilistic implementation of a nonlocal gate by using a single non - maximally entangled state@xcite .
the success fidelity maintains a high level though the state shared is partially entangled .
99 a. galindo and m. a. m. delgado , _ rev .
* 74 * ( 2002 ) 347 .
f. g. deng , g. l. long and x. s. liu , _ phys .
rev . a _ * 68 * ( 2003 ) 042317 . h. lu and g. c. guo , _ phys .
a _ * 276 * ( 2000 ) 209 . a. m. childs and i. l. chuang , _ phys .
a _ * 63 * ( 2000 ) 012306 .
n. k. langford , r. b. dalton , m. d. harvey , j. l. brien , g. j. pryde , a. gilchrist , s. d. bartlett and a. g. white , _ phys .
* 93 * ( 2004 ) 053601 .
h. b. pasquinucci and a. peres , _ phys .
* 85 * ( 2000 ) 3313 .
brukner , m. ukowski and a. zeilinger , _ phys .
_ * 89 * ( 2002 ) 197901 .
y. shi , _ phys .
b _ * 641 * ( 2006 ) 75 . m.gell-mann and a. pais , phys .
97 ( 1955 ) 1387 .
t. feldmann and p. kroll , phys .
d 58 ( 1998 ) 114006 a. magiera and h. machner , nucl . phys . a 674 ( 2000 )
515 p. kroll , _ modern phys .
a _ * 20 * ( 2005 ) 2667 .
v. dmitrasinovic , _ phys .
rev.d _ * 56 * ( 1997 ) 247 .
j. cao , f. g. cao , t. huang and b. q. ma , _ phys .
* 58 * ( 1998 ) 113006 .
b. borasoy and r. nilbler , _ phys .
j. a _ * 26 * ( 2005 ) 383 . f. d. fazio and m. r. pennington , _ j. high energy phys . _ * 7 * ( 2000 ) 51
. m. y. han and y. nambu , _ phys .
* 139 * ( 1965 ) b1006 .
d. uglov , _ commun .
* 191 * ( 1998 ) 663 .
a. kundu , _ phys .
a _ * 249 * ( 1998 ) 126 .
d. bernard , _ inter .
j. modern phys .
b _ * 7 * ( 1993 ) 3517 . f. d. m. haldane , _ phys . rev .
* 60 * ( 1988 ) 635 .
b. s. shastry , _ phys .
lett . _ * 60 * ( 1988 ) 639 .
f. pan , d. liu , g. y. lu and j. p. draayer , _ phys .
a _ * 336 * ( 2005 ) 384 . m. l. ge . and
yang@xmath67baxter equation _ ( shanghai : shanghai scientific and technical publishers ) ( 1999 ) p511 . c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres and w. k. wootters , _ phys .
* 70 * ( 1993 ) 1895 .
j. x. fang , y. s. lin , s. q. zhu and x. f. chen , _ phys .
a _ * 67 * ( 2004 ) 014305 .
l. chen and y. x. chen , _ phys .
a _ * 71 * ( 2005 ) 054302 . | in this paper , we investigate the entanglement degrees of pseudoscalar meson states via quantum algebra @xmath0 . by making use of transition effect of generators @xmath1 of @xmath0
, we construct various transition operators in terms of @xmath1 of , and act them on @xmath2-@xmath3-@xmath4 mixing meson state .
the entanglement degrees of both the initial state and final state are calculated with the help of entropy theory .
the diagrams of entanglement degrees are presented .
our result shows that a state with desired entanglement degree can be achieved by acting proper chosen transition operator on an initial state .
this sheds new light on the connect among quantum information , particle physics and yangian algebra .
+ + keywords : yangian @xmath0 ; @xmath2-@xmath3-@xmath4 mixing ; the entanglement degree |
assume a cluster and a moving perturbing object ( gmc , spiral arm , disc , etc . ) with relative separation and velocity @xmath1 and @xmath2 , respectively .
the acceleration due to the perturber is referred to as @xmath3 .
the _ tidal _ acceleration ( @xmath4 ) on a star at a distance @xmath5 from the cluster centre can be obtained by linearising @xmath3 : @xmath6 the total velocity increase ( @xmath7 ) of the star after the perturbation can be obtained by substituting @xmath8 and integrating eq .
if the time it takes the perturber to pass the cluster is much shorter than the crossing time of stars in the cluster , @xmath5 can be assumed constant .
this is referred to as the _ impulsive approximation_. the resulting then simply scales with @xmath5 , @xmath2 and the maximum acceleration ( @xmath9 ) as : @xmath10 .
the total kinetic energy gain per unit of cluster mass ( @xmath11 ) can then be written as @xmath12 where we used that the mean - square position of stars ( ) is @xmath13 . _ eq .
[ eq:2 ] shows that @xmath11 is large for slow perturbations on large clusters_. for passing point masses @xmath14 , with @xmath15 the gravitational constant , @xmath16 the mass of the perturber and @xmath17 the distance of closest approach ( see @xcite for the exact definition of eq .
[ eq:2 ] , taking into account an extended density profile for a gmc ) . for a one - dimensional density perturbation , such as the perpendicular crossing of a galactic disc and , to lesser extend , the crossing of a spiral arm , @xmath18 , with @xmath19 the scale height / length and @xmath20 the central density . for more details
see @xcite and references therein .
a single tidal perturbation can accelerate stars to velocities much higher than the escape velocity .
this _ overkill _ energy can not be used to unbind more stars and is lost with the escaping stars . for this reason , the fractional mass loss ( , with @xmath21 the cluster mass ) is lower than the fractional energy gain ( , with @xmath11 from eq .
[ eq:2 ] and @xmath22 the cluster energy per unit mass ) , by a factor @xmath23 , with @xmath24 @xcite . with this and the definition of ( @xmath25 , with the cluster half - mass radius and @xmath26 , depending on the cluster density profile ) , we can write @xmath27 as @xmath28 where we used eq . [ eq:2 ] and @xmath29 , with @xmath30 , depending on the cluster density profile .
the statistical mean mass loss per unit time ( @xmath31 ) is defined as @xmath32 , where @xmath33 for gmc encounters is the time between successive encounters and depends on the gmc density ( @xmath34 ) as @xmath35 .
_ note that @xmath36 and @xmath31 due to external perturbations are independent of cluster mass .
defining the cluster disruption time ( @xmath37 ) as @xmath38 , we can write ( for gmc encounters ) from observations of clusters in different galaxies , @xcite found that @xmath40 , with @xmath41 .
the same value for @xmath42 was found in @xmath0-body simulations of clusters dissolving in a galactic tidal field @xcite , however , their values are much longer than the observed ones .
recently found that @xmath43 , which implies @xmath44 and hence @xmath45 ( eq . [ eq:4 ] ) .
the fact that due to the galactic tidal field _ and
_ due to the tidal perturbations depend on in almost the same way can explain why the values of @xmath42 derived from observations of clusters in different galaxies are so similar @xcite , while varies by almost two orders of magnitude . , h. & makino , j. 2003 , , 340 , 227 , s. g. & lamers , h. j. g. l. m. 2003 , , 338 , 717 , m. , athanassoula , e. , & portegies zwart , s. f. 2006 , mnras , submitted , m. , portegies zwart , s. f. , baumgardt , h. , et al .
2006 , , 371 , 793 , h. j. g. l. m. , gieles , m. , bastian , n. , et al .
2005 , , 441 , 117 , h. j. g. l. m. , gieles , m. , & portegies zwart , s. f. 2005 , , 429 , 173 , s. s. 2004 , , 416 , 537 , l. 1987 , dynamical evolution of globular clusters ( princeton , nj , princeton university press , 1987 ) | star clusters are subject to density irregularities in their host galaxy , such as giant molecular clouds ( gmcs ) , the galactic disc and spiral arms , which are largely ignored in present day ( @xmath0-body ) simulations of cluster evolution .
time dependent external potentials give rise to tidal forces that accelerate stars leading to an expansion and more rapid dissolution of the cluster .
i explain the basic principles of this tidal heating in the _ impulse approximation _ and show how related disruption time - scales depend on properties of the cluster . |
superfluidity plays a key role in many fields of physics , such as liquid helium @xcite , ultracold atomic gases @xcite , quantum magnets @xcite , and astrophysics @xcite .
in particular , superfluid bose - fermi mixtures ( sbfms ) in ultracold atomic gases have attracted considerable interest recently . it is well known that for liquid helium the superfluidity of bosonic @xmath0he and fermionic @xmath1he can be achieved separately .
however , the simultaneous superfluidity for the two isotopes is remarkably prevented by the strong interactions between the two species in spite of such a superfluid @xmath0he-@xmath1he mixture being a long - sought object @xcite . by contrast , a superfluid bose - fermi mixture ( sbfm ) in ultracold atomic gases can be realized by using the combination of the feshbach resonance and radio - frequency techniques .
most recently , a sbfm consisting of condensed @xmath2li bosons and ultracold @xmath3li fermions in two - spin states has been produced by salomon s group ferrier - barbut .
the experimental breakthrough provides new opportunities to study the intriguing properties of sbfms inaccessible in a bose - einstein condensate ( bec ) with arbitrary spin @xcite , pure bosonic superfluid mixtures , and pure fermionic superfluid mixtures . as a matter of fact
, many novel physical characteristics have been predicted theoretically in sbfms , including the saturation effect of nonlinear interaction in the strong - coupling unitarity limit @xcite , localized bose - fermi bright soliton @xcite , possibility of simulating dense quantum chromodynamics matter @xcite , phase transition overlapping with phase separation @xcite , and the faraday pattern generation abdullaev .
all existing studies of the sbfms refer to the nonrotating case . considering that one of the most striking hallmarks of a superfluid is its response to rotation , in this paper we investigate the combined effects of rotation and nonlinear interatomic interaction on the exact two - dimensional ( 2d ) topological structure and dynamics of a rotating sbfm , focusing on the superfluid @xmath2li-@xmath3li mixture case as the prototype . here
we present a phenomenological dissipation model @xcite combining with the unitary schrdinger equation for an interacting sbfm @xcite to describe the dynamics of a rotating sbfm .
a ground - state phase diagram for a nonrotating sbfm with specific parameters is given .
we find that the ground - state structure of a nonrotating sbfm is mainly determined by the ratio of the boson - fermion ( bf ) scattering length to the boson - boson ( bb ) scattering length .
furthermore , it is shown that the rotating sbfm with a sufficiently large rotation frequency can display four steady structures : a mixed phase and three typical layer separated phases .
in particular , we show that the generation of visible vortices in the fermionic superfluid displays an evident hysteresis effect during the time evolution of a rotating sbfm . due to the cooper pair entity of fermionic superfluid and the presence of bf and fermion - fermion ( ff ) interactions
, it is demonstrated that the topological structure and the dynamics of a rotating sbfm are evidently different from the usual cases of rotating two - component bose - einstein condensates ( becs ) .
the paper is organized as follows . in sec .
ii , we describe a phenomenological model for a rotating sbfm . in sec .
iii , we study the equilibrium structure of the rotating sbfm . a phase diagram for a nonrotating sbfm
is given , and the phase structures of the rotating sbfms with various parameter values are analyzed . in sec .
iv , we discuss the dynamics of vortex formation in a rotating sbfm .
the conclusion is outlined in the last section .
we consider a superfluid bose - fermi mixture composed of superfluid bosons and two - component superfluid fermions with equal populations of spin up and down , corresponding to the most favorable condition for cooper pairing .
the superfluid bose - fermi gas is confined in a harmonic trap rotating around the @xmath4 axis with angular velocity @xmath5 . in the present work we use a combined phenomenological model based on a phenomenological dissipation model @xcite and a galilei - invariant nonlinear unitarity model adhikari1 to investigate the dynamics of the vortex formation and the structure of the equilibrium state of a rotating sbfm . in this phenomenological model , the order parameter @xmath6 of a bosonic superfluid ( i.e. , a bose - einstein condensate ) and the order parameter @xmath7 of a fermionic superfluid @xcite obey the coupled equations in the rotating frame,@xmath8 \psi _ { b } \notag \\ & & + g_{bp}\left\vert \psi _ { p}\right\vert ^{2}\psi _ { b } , \label{sbe } \\ ( i-\gamma ) \hbar \frac{\partial \psi _ { p}}{\partial t } & = & \left [ -\frac{% \hbar ^{2}\nabla ^{2}}{2m_{p}}+u_{p}+\mu _ { p}(n_{p},a_{f})-\omega l_{z}% \right ] \psi _ { p } \notag \\ & & + g_{bp}\left\vert \psi _ { b}\right\vert ^{2}\psi _ { p } , \label{sfe}\end{aligned}\]]where for simplicity we assume that the degree of dissipation @xmath9 of the fermionic superfluid is the same as that of the bec .
@xmath10 is the mass of a bosonic atom , @xmath11 is the mass of a cooper pair with @xmath12 being the mass of a single fermion , and @xmath13 denotes the @xmath4 component of the angular - momentum operator .
@xmath14 is the external trapping potential for the bosons , @xmath15 is the trapping potential for the fermionic superfluid , and @xmath16(@xmath17 ) and @xmath18(@xmath19 ) are the radial trap frequency and the axial trap frequency for the bosons ( fermions ) , respectively .
@xmath20 , where @xmath21 is the bf @xmath22-wave scattering length , and @xmath23 is the bf reduced mass .
the bulk chemical potential of the bec is given by@xmath24where @xmath25@xmath26 is the local density of bosons , and @xmath27 is the bb @xmath22-wave scattering length . here
we choose @xmath28 , @xmath29 with @xmath30 @xmath31 , @xmath32 , and @xmath33 with @xmath34 adhikari1 .
the bulk chemical potential of the fermi superfluid is expressed as @xcite @xmath35where @xmath36@xmath37 is the local density of cooper pairs with @xmath38 being the total local density of fermions , @xmath39 denotes the attractive ff scattering length , @xmath40 , and @xmath41 with @xmath42 .
the choices of the parameters in eqs .
( bosonfunction ) and ( [ fermionfunction ] ) are consistent with the unitarity and the lee - yang - huang limits @xcite as well as the relevant results of the monte carlo calculations blume1,carlson , astrakharchik , blume2,chang .
this phenomenological model is a variation of that in tsubota , kasamatsu1 and a generalization of that of a rotating single - component bec @xcite , and it has good predictive power wen2 .
the largest merit of the phenomenological dissipation model is that by virtue of this model one can obtain not only the steady states of a rotating system but also reveal the dynamics of vortex formation .
in addition , the present model is valid from weak coupling to unitarity for both bosons and fermions @xcite . as a matter of fact , the unitary schrdinger model for nonrotating superfluid fermions
was presented in refs .
adhikari1,adhikari3,salasnich , where the bulk chemical potential was given by eqs . ( [ chempotentialfermion ] ) and ( [ fermionfunction ] )
( see also eqs .
( 10 ) , ( 11 ) and ( 28 ) in ref .
the lowest - order term @xmath43 in eq .
( [ fermionfunction ] ) results in the bulk chemical potential of a fermi superfluid in the absence of ff interaction ( @xmath44 ) .
furthermore , the next - order term @xmath45 can lead to a known analytical result in the small - gas - parameter regime ( medium value ff scattering length @xmath39 ) as obtained in ref .
when @xmath46 tends to negative infinity ( the so - called unitary limit ) , the function @xmath47 obviously approaches an asymptotic value @xmath48 .
thus the function ( [ chempotentialfermion ] ) with eq .
( fermionfunction ) provides a smooth interpolation between the bulk chemical potential of a fermi superfluid in the weak - coupling limit and that in the unitary limit for both the uniform case and the trapped case adhikari1,adhikari3,lee1 .
it was shown that the results with the choice of the fitting parameters @xmath40 , @xmath49 and @xmath42 @xcite agreed well with the corresponding monte carlo data @xcite .
therefore , we will use the choice in the present work .
next , we consider the two - dimensional problem by assuming the translation invariance along the @xmath4 axis ( i.e. , @xmath50 ) , reducing the order parameters as @xmath51 with the typical size @xmath52 along the @xmath4 axis . after a straight - forward calculation , we obtain 2d coupled equations,@xmath53 \psi _ { b } \notag \\ & & + \left [ \frac{m_{b}\omega _ { b}^{2}}{2}\left ( x^{2}+y^{2}\right ) + \frac{% g_{bp}}{r_{z}}\left\vert \psi _ { p}\right\vert ^{2}\right ] \psi _ { b } , \label{2dsbe } \\ ( i-\gamma ) \hbar \frac{\partial \psi _ { p}}{\partial t } & = & \left [ -\frac{% \hbar ^{2}}{2m_{p}}(\nabla _ { x}^{2}+\nabla _ { y}^{2})+j(\psi _ { b})-\omega l_{z}\right ] \psi _ { p } \notag \\ & & + \left [ m_{f}\omega _ { f}^{2}(x^{2}+y^{2})+\frac{g_{bp}}{r_{z}}\left\vert \psi _ { b}\right\vert ^{2}\right ] \psi _ { p } , \label{2dsfe}\end{aligned}\]]where@xmath54and @xmath55the initial order parameters are normalized as @xmath56 @xmath57 with @xmath58 being the initial number of bosons and @xmath59 being the initial number of cooper pairs ( @xmath60 and @xmath61 is the total number of fermions ) . by introducing the notations @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67 , and replacing the wave functions as @xmath68
, we obtained the rescaled dimensionless 2d equations for the mixture@xmath69 \psi _ { b } \notag \\ & & + \frac{4\pi \hbar } { m_{b}\omega _ { bf}}\frac{\left ( a_{1}\left\vert \psi _ { b}\right\vert ^{2}+\alpha b_{1}\left\vert \psi _ { b}\right\vert ^{4/3}\right ) } { 1+\lambda d_{1}\left\vert \psi _ { b}\right\vert + \beta e_{1}\left\vert \psi _ { b}\right\vert ^{5/3}}\psi _ { b } \notag \\ & & + \frac{m_{b}\omega _ { b}^{2}}{4m_{bf}\omega _ { bf}^{2}}\left ( x^{2}+y^{2}\right ) \psi _ { b } , \label{dimless2dsbe } \\ ( i-\gamma ) \frac{\partial \psi _ { p}}{\partial t } & = & \left [ -\frac{m_{bf}}{% m_{p}}(\nabla _ { x}^{2}+\nabla _ { y}^{2})+c_{2}\left\vert \psi _ { b}\right\vert ^{2}-\omega l_{z}\right ] \psi _ { p } \notag \\ & & + a_{2}\left\vert \psi _ { p}\right\vert ^{4/3}\left ( 1+\frac{\delta b_{2}\left\vert \psi _ { p}\right\vert ^{2/3}}{1-kb_{2}\left\vert \psi _ { p}\right\vert ^{2/3}}\right ) \psi _ { p } \notag \\ & & + \frac{m_{f}\omega _ { f}^{2}}{2m_{bf}\omega _ { bf}^{2}}\left ( x^{2}+y^{2}\right ) \psi _ { p } , \label{dimless2dsfe}\end{aligned}\]]where the number subscript @xmath70 is omitted for simplicity .
the corresponding coefficients are @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath77 , and @xmath78 . in the following ,
we numerically solve the 2d coupled equations ( dimless2dsbe ) and ( [ dimless2dsfe ] ) which requires an enormous computational effort .
the initial ground - state order parameters @xmath79 and @xmath80 of the system can be obtained by the imaginary - time propagation method @xcite based on the peaceman - rachford method @xcite .
here we consider a superfluid @xmath2li-@xmath3li mixture in harmonic traps satisfying @xmath81 , with @xmath82being the mass of @xmath83li atom and @xmath84 being the mass of the @xmath3li atom .
the parameters are chosen as @xmath85 hz , @xmath86 @xmath87 m , and @xmath88 , corresponding to a temperature of about @xmath89 @xcite .
recently , a mixture of @xmath2li superfluid and @xmath3li superfluid has been realized by salomon s group @xcite .
thus the above assumption is valid and feasible , and the relevant results can be tested under the current experimental conditions .
for convenience , we introduce two relative interaction strengths , @xmath90 and @xmath91 .
the ff attractive scattering length @xmath39 can be varied from zero to negative infinity , and the absolute value of the bf scattering length @xmath92 is assumed to be not too large ( otherwise , it requires special attention @xcite ) .
li-@xmath3li ( bose - fermi ) mixture , where @xmath93 and @xmath91 .
the parameters are @xmath94 , @xmath95 , and @xmath96 nm . regions i iv represent the system collapse regime , mixed phase , layer - separated phase , and inlaid separated phase , respectively . ,
width=264 ] figure 1 shows the ground - state phase diagram of a static superfluid @xmath2li@xmath3li mixture with fixed bb repulsive interaction .
the corresponding parameters are @xmath97 nm , @xmath94 , and @xmath95 . there exist four possible phases depending on the values of @xmath98 and @xmath99 , and the typical density profiles corresponding to phases ii - iv are displayed in fig .
2 . in fig . 1 , region i denotes the system collapse regime . for a sufficiently strong attractive bf interaction ( here ,
, i.e. , @xmath101 nm ) , the system undergoes a simultaneous collapse of the density profile of the bec and that of the fermionic superfluid .
physically , the critical value of bf scattering length is governed by the balance between the kinetic energy of bosons and cooper pairs and the mutual attractive bf interaction .
when the bf attraction becomes sufficiently strong , it can no longer be stabilized by the kinetic energy .
therefore the mixture lowers its energy via increasing the densities of bosons and cooper pairs , and finally the bosonic superfluid or the fermionic one or both the bosonic and fermionic superfluids collapse simultaneously due to instability .
region ii represents a miscible phase in which the density profile of the bec has a much larger spatial extension than that of the fermionic superfluid due to the bb repulsion and the ff attraction [ see figs .
2(a1 ) and 2(a2 ) ] .
region iii is a layer - separated phase , where the bec is completely expelled outside the fermionic superfluid because of the strong bf repulsive interaction [ figs .
2(b1 ) and 2(b2 ) ] , which is usually referred as demixing .
the similar mixing - demixing transition phenomenon has also been found in the studies of degenerate boson - fermion mixtures @xcite or two - component becs @xcite . finally , region iv marks an inlaid separated phase in which the fermionic superfluid is divided into two segments and inlaid into the outskirts of the bec [ figs .
2(c1 ) and 2(c2 ) ] .
labels the bosonic superfluid and @xmath102 denotes the fermionic one of the system .
( a ) @xmath103 ( b ) @xmath104 and ( c ) @xmath105 .
the other parameters are the same as those in fig .
1 . the darker color area indicates the lower density . @xmath106 and @xmath107 are in units of @xmath108 .
, width=294 ] the ground - state structures of sbfms are different from those of degenerate boson - fermion mixtures @xcite . in the latter case
, there is no @xmath22-wave interaction between identical fermions in the spin - polarized state due to the pauli exclusion principle .
depending on the bb and bf interactions , the density profile of a degenerate boson - fermion mixture may display a core - shell - shaped separated phase , where the fermi gas forms a shell around or a core inside the bec , or even both @xcite .
however , our simulation shows that in a wide range of parameter values , the sbfm supports neither a separated phase of fermions constituting a shell around the bec nor a staggered separated phase of fermions becoming both a shell around and a core inside the bec ( see figs . 1 and 2 ) ; even the total number of fermions is much larger than that of bosons .
in addition , for a fixed value of @xmath98 ( i.e. , a fixed bf interaction ) , the variation of the parameter @xmath99 ( ff interaction ) dose not influence the phase structure of the sbfm ( fig .
1 ) . furthermore ,
when the repulsive bf interaction of a sbfm is sufficiently strong , the system exhibits an inlaid separated phase , where the fermionic superfluid embeds into the periphery of the bosons by means of two fragment superfluids [ figs .
2(c1 ) and 2(c2 ) ] .
it is known that the fundamental entity of a fermionic superfluid is the cooper pair ramachandhran , adhikari5,gubbels and there exists attractive ff interaction between the fermion with spin up and that with spin down .
thus the center - of - mass momentum of each cooper pair in the fermionic superfluid remains zero , irrespective of the concrete value of @xmath39 , which may account for the above differences .
on the other hand , the equilibrium properties of the sbfms are also evidently different from those of two - component becs wen4,ho2,pu , trippenbach , lwen .
for the case of two - component becs , by varying the particle numbers of the two becs or the ratios of intra- and intercomponent interaction strengths , one can obtain many symmetric separated phases @xcite and even asymmetric separated phases @xcite .
physically , there exist three kinds of actual @xmath22-wave interactions ( including bb interaction , bf interaction , and ff interaction ) in a sbfm . in particular , when the bosonic and fermionic scattering lengths @xmath27 and @xmath109 tend to infinity ( i.e. , in the strong - coupling unitarity limit ) , both the bosonic and fermionic interactions exhibit unitarity saturation effects due to the constraints of quantum mechanics @xcite . in this context , the quasiparticle of a cooper pair in the fermionic superfluid can just be considered as a composite boson rather than a single bosonic atom , which largely results in the differences of the ground - state structures . in fig .
3 , we display the steady density profiles @xmath110 ( row @xmath111 ) and @xmath112 ( row @xmath113 ) , and the corresponding phase profiles of @xmath114 ( row @xmath115 ) and @xmath116 ( row @xmath117 ) at @xmath118 for a harmonic trap rotating with @xmath119 . here the value of the phase varies continuously from @xmath70 to @xmath120 , and the end point of the boundary between a @xmath120 phase line and a @xmath70 phase line denotes a phase defect ( i.e. , a vortex with anticlockwise rotation ) .
the density profiles of three initial states corresponding to the three columns ( left , middle , and right ) in fig .
3 are given in figs .
2(a)2(c ) , respectively . after rotating the superfluid bose - fermi mixture with @xmath121 , where @xmath111 and @xmath115 denote the bosonic superfluid , while @xmath113 and @xmath117 represent the fermionic superfluid .
the initial states corresponding to three cases of @xmath122 and @xmath123 ( from left to right ) are illustrated in figs .
2(a1,a2 ) , figs .
2(b1,b2 ) , and figs .
2(c1,c2 ) , respectively .
the other parameters are the same as those in figs . 1 and 2 .
the value of phase varies continuously from @xmath70 to @xmath124 .
the darker color area indicates lower density or phase . here ,
@xmath106 and @xmath107 are in units of @xmath108 .
, width=302 ] as shown in figs .
3(a1 ) and 3(a3 ) , when the rotating sbfm with an initial state of @xmath125 and @xmath126 reaches an equilibrium state , one can see that there are several visible vortices @xcite constituting a triangular lattice in the periphery of the bosonic superfluid . in the meantime
, the trap center is occupied by a density peak of bosons due to the attractive bf interaction .
however , no visible vortex can be generated in the density of the fermionic superfluid even for large rotation frequency of @xmath121 [ see fig .
3(a2 ) ] . from fig .
3(a4 ) , there exist some phase defects that are located on the outskirts of the superfluid fermions
. since these phase defects are invisible in the _ in situ _ density profile of fermionic superfluid and contribute to neither the angular momentum nor the energy of the sbfm , there are referred to as ghost vortices wen1,tsubota , kasamatsu1,wen2,wen4 .
for the case of @xmath127 and @xmath128 , there is a large circular density hole of bosons in the trap center which looks like a giant vortex ( a multiquantized vortex ) and is surrounded by an expected visible vortex lattice [ fig .
3(b1 ) ] . in the phase profile displayed in fig .
3(b3 ) , we find that there are three phase defects in the trap center which are close to each other .
the three singly quantized phase defects show that the circular density hole is not a giant vortex . actually , they are known as singly quantized hidden vortices wen1,wen2,wen4,mithun because they carry significant angular momentum , though they are invisible in the _ in situ _ density profile of the bec . only after including the hidden vortices
can the feynman rule be satisfied @xcite . as seen in figs .
3(b2 ) and 3(b4 ) , the fermionic superfluid with ghost vortices distributing on the outskirts of the cloud is completely pulled inside the trap center ( i.e. , the region of the large density hole of the bec ) due to the competition between the bf repulsion and the rotation repulsion , which indicates a fully separated phase of the rotating sbfm
. with the further increase of the bf repulsive interaction , a nonrotating sbfm will develop into an inlaid separated configuration , where the superfluid fermions lie on the outer edges of the bosons via the form of two fragments [ figs .
2(c1 ) and 2(c2 ) ] .
counterintuitively , we find that the steady structure of a rotating sbfm with @xmath129 and @xmath126 is similar to that with @xmath127 and @xmath130 as shown in figs .
3(c1)3(c4 ) .
here the fermionic superfluid is separately repelled to the trap center region rather than the outskirts of bec , which is quite different from the nonrotating case . in the presence of dissipation , the steady visible vortex lattice in the bec
is mainly formed by the competition between the rotating driving and the bb repulsive interaction .
the bf repulsion and the centrifugal force acting on the bosons tend to push the fermionic superfluid toward the outside or the inside .
considering as well the coherence of the fermionic superfluid with ff attractive interaction , it evidently prefers to occupy the trap center , especially in the presence of a central density hole for the bosons , because the corresponding potential energy for the superfluid fermions is relatively small such that the system energy of the sbfm reaches the minimum . in figs .
4(a1 ) and 4(a2 ) , we show the density profiles of the bec and the fermionic superfluid in a static sbfm with @xmath131 , @xmath132 , @xmath133 nm , @xmath129 , and @xmath126 .
compared with the layer - separated phase in figs .
2(a1 ) and 2(a2 ) , here the torus density profile of the bec becomes thinner while the circular density distribution of the fermionic superfluid gets larger .
the steady structure at @xmath118 of the sbfm rotating with @xmath134 is displayed in figs .
4(b1 ) and 4(b2 ) , and the corresponding phase profiles are given in figs .
4(c1 ) and 4(c2 ) . from figs .
4(b1 ) and 4(c1 ) , we can see that there are no visible vortices , but there exist some hidden vortices in the bec , where the trap center is occupied by multiple close singly quantized hidden vortices . here
the central blurred region in the phase profile mainly results from the inevitable numerical errors or fluctuations in the numerical computations .
in contrast to fig .
3 , a triangular visible vortex lattice made of six visible vortices forms in the rotating fermionic superfluid , as shown in figs .
4(b2 ) and 4(c2 ) .
although there is attractive ff interaction between two fermions in a cooper pair , the fermionic superfluid can display a weak effective repulsive interaction between the cooper pairs under the appropriate parameters according to eqs .
( [ sfe ] ) , ( [ chempotentialfermion ] ) , ( fermionfunction ) , and ( [ dimless2dsfe ] ) , which is consistent with the relevant theoretical prediction @xcite .
this point may explain why visible vortices can be generated in the rotating fermionic superfluid .
li-@xmath3li mixture with @xmath131 and @xmath132 .
( b1 ) , ( b2 ) the steady density profiles at @xmath118 for the superfluid bose - fermi system rotating with @xmath134 .
( c1 ) , ( c2 ) the corresponding phase profiles .
the relevant parameters are @xmath135 nm , @xmath129 , and @xmath136 . here , @xmath111 denotes the bosonic superfluid , @xmath113 represents the fermionic superfluid , and @xmath106 and @xmath107 are in units of @xmath108 .
the darker color area indicates lower density or phase .
, width=294 ]
in order to reveal the vortex formation process in the rotating superfluid mixture of bosons and cooper pairs , we consider a bigger superfluid bose - fermi system which requires more computation effort . in figs .
5(a1 ) and 5(a2 ) , we present the ground - state structure of a static sbfm , where the parameters are @xmath137 , @xmath132 , @xmath97 nm , @xmath129 , and @xmath138 . the steady structure at @xmath118 of the sbfm rotating with @xmath139 is shown in figs .
5(b1 ) and 5(b2 ) , and the corresponding phase profiles are displayed in figs .
5(c1 ) and 5(c2 ) .
the initial state of the sbfm at @xmath140 has a shell - shaped separated structure with the fermionic superfluid being surrounded by the bosonic superfluid , which is similar to that in figs .
2(b1 ) and 2(b2 ) . in the presence of dissipation , a steady visible vortex lattice forms eventually in the outer of the superfluid bec [ fig .
5(b1 ) ] , where the energy of the rotating sbfm reaches the minimum in the rotating frame . this point is similar to the case of fig .
the large density hole of the bosons corresponds to 22 singly quantized hidden vortices indicated by the phase profile in fig .
the central blurred region in the phase profile is mainly caused by the inevitable numerical errors or fluctuations in the numerical computations .
in contrast to the case of @xmath141and @xmath95 ( figs . 2 and 3 ) , seven evident visible vortices are generated in the fermionic superfluid [ see figs .
5(b2 ) and 5(c2 ) ] .
li-@xmath3li mixture with @xmath137 and @xmath132 .
( b1 ) , ( b2 ) the steady density profiles at @xmath118 for the superfluid bose - fermi system rotating with @xmath142 .
( c1 ) , ( c2 ) the corresponding phase profiles .
the relevant parameters are @xmath97 nm , @xmath129 , and @xmath143 . here , @xmath111 denotes the bosonic superfluid , @xmath113 represents the fermionic superfluid , and @xmath106 and @xmath107 are in units of @xmath108 .
the darker color area indicates lower density or phase .
, width=317 ] the dynamical evolution of the rotating sbfm is illustrated in fig .
6 , where the top two rows denote the time evolutions of the density profiles of the bec ( row 1 ) and fermionic superfluid ( row 2 ) , while the bottom two rows represent those of the phase profiles of the bec ( row 3 ) and fermionic superfluid ( row 4 ) .
the evolution times are @xmath144 ( left ) , @xmath145 ( middle ) , and @xmath146 ( right ) , respectively .
initially , the superfluid densities @xmath147 and @xmath148 in a stationary isotropic harmonic trap are shown in figs .
5(a1 ) and 5(a2 ) . with the development of time ,
the boundary surface of the bosonic superfluid undergo complex turbulent oscillation and many ghost vortices appear at the outskirts of the bec , which can be seen in figs .
6(a1 ) and 6(a3 ) .
in contrast , the fermionic superfluid remains basically unchanged , as shown in figs .
6(a2 ) and 6(a4 ) .
essentially , these ghost vortices are generated by collective excitations through the nonlinear atomic interactions and the landau instability associated with the negative excitation frequency @xcite because the rotating harmonic trap is isotropic and has rotation symmetry .
this characteristic is evident especially for the component of fermionic superfluid [ see figs .
6(a2)6(c2 ) and figs .
6(a4)6(c4 ) ] .
thus here the formation mechanism of topological defects is different from the case of a rotating anisotropic harmonic potential , where ghost vortices mainly result from the dynamical instability through the rapid modulation of trapping anisotropy @xcite .
the angular momentum is transferred into the sbfm via the excitations of surface modes or the generation of visible vortices and hidden vortices . with the further time evolution ,
some ghost vortices penetrate into the bec and become visible vortices or hidden vortices due to the landau instability [ see figs .
6(b1)6(c1 ) and figs .
6(b3)6(c3 ) ] , where the visible vortices arrange themselves irregularly .
when the rotating sbfm reaches an equilibrium state , the visible vortices form a triangular lattice such that the energy of the system approaches the minimum in the rotating frame . compared with the rotating bec , the vortex ( including ghost vortex and visible vortex ) formation in the rotating fermionic superfluid exhibits an evident hysteresis effect , as shown in figs . 6 and 5 .
( row 1 ) and @xmath149 ( row 2 ) and the corresponding phase profiles ( rows 3 and 4 ) after the superfluid bose - fermi system rotates with @xmath142 , where the other parameters are the same as those in fig .
5 . the value of the phase varies continuously from @xmath70 to @xmath124 .
the time is ( a1-a4 ) @xmath150 , ( b1b4 ) @xmath145 , and ( c1c4 ) @xmath146 .
the dark color area indicates the lower density or phase . here
@xmath106 and @xmath107 are in units of @xmath108 , and @xmath151 is in units of @xmath152.,width=317 ] the structure and dynamics of a rotating sbfm are remarkably different from those of rotating two - component becs .
first , in the latter case , the vortices in the two becs are always generated simultaneously kasamatsu2,woo , mason , while here the vortex formation in the superfluid fermions is far later than that in the bec due to the presence of cooper pair with attractive ff interaction .
second , in the presence of dissipation , there are triangular visible vortex lattices and singly quantized hidden vortices formed in a rotating sbfm . in our simulation , we did not observe the generation of vortex sheet or square vortex lattice occurring in a rotating two - component bec @xcite .
these structures as well as the giant vortex are suppressed by the dissipation term in our simulation .
in summary , we have studied the structures for rotating and non - rotating superfluid bose - fermi mixtures and the dynamics of vortex formation in a rotating superfluid bose - fermi mixture .
we show that the ratio of bf interaction to bb interaction plays a key role in determining the ground - state structure of a nonrotating sbfm . for fixed bb and bf @xmath22-wave scattering lengths , the variation of ff ( spin - up and -down )
@xmath22-wave scattering length does not influence the phase diagram for a nonrotating sbfm .
furthermore , depending on the choice of parameters , we find that a rotating sbfm with a sufficiently large angular velocity supports four typical steady structures : a mixed phase and three layer - separated phases .
in addition , it is shown that different separated phases at the initial time may result in almost the same steady structures .
in particular , we find that the generation of visible vortices in the fermionic superfluid exhibits an evident hysteresis effect during the time evolution of a rotating sbfm .
our results indicate that the topological structure and the dynamics of a rotating sbfm are remarkably different from the cases of rotating two - component becs .
we expect that our findings can be observed and tested in future experiments . in the mean time
, the present investigation provides a way to further test the validity of the unitarity model @xcite for a sbfm .
we thank biao wu , yongping zhang , li mao and yong xu for helpful discussions .
l.w . acknowledges the research group of professor chuanwei zhang at the university of texas at dallas , where part of the computations were carried out .
this work was supported by the national natural science foundation of china ( grants no .
11475144 , no . 11047033,and
11304270 ) , and the ph.d .
foundation of yanshan university ( no . b846 ) . | we investigate the structure and dynamics of a rotating superfluid bose - fermi mixture ( sbfm ) made of superfluid bosons and two - component ( spin up and down ) superfluid fermions . a ground - state phase diagram for the nonrotating case of a sbfm with specific parameters
is given , where the ground - state configuration of a nonrotating sbfm is mainly determined by the boson - fermion interaction .
for the rotating case of a sbfm with a sufficiently large rotation frequency , we show that the system supports a mixed phase and three typical layer separated phases .
in particular , the visible vortex formation in the fermionic superfluid exhibits a remarkable hysteresis effect during the dynamical evolution of a rotating sbfm , which is evidently different from the case of rotating two - component bose - einstein condensates . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.